A0595
Title: Dynamic conditional correlation models with time-varying parameters incorporating realized covariance matrices
Authors: Hideto Shigemoto - Kwansei Gakuin University (Japan) [presenting]
Takayuki Morimoto - Kwansei Gakuin University (Japan)
Abstract: Novel realized dynamic conditional correlation (Realized DCC) models with measurement errors are proposed to forecast the covariance of asset returns. The aim is to estimate the conditional covariance of asset returns and to ensure the bound of the forecasted correlation matrix by incorporating the measurement error in estimating the conditional variance and using the BEKK and HAR models to estimate the conditional correlation matrix. The models that we propose can incorporate measurement errors into not only variance estimations but also correlation estimations. These models can keep the persistence of volatility and correlation at high levels when the asymptotic variances of realized volatility and realized correlation are small. By incorporating measurement errors, models can decrease the persistence when the asymptotic variances are large. We show, in our empirical results, a significant improvement in predictive performance over the benchmark BEKK-type model and the usual Realized DCC model.
