Modeling stock market volatility using new HAR-type models

Modeling volatility with reasonable accuracy is essential in asset allocation, asset pricing, and risk management. In this paper we use the ensemble empirical mode decomposition method and Zhang et al. (2008, 2009)'s method to decompose realized volatility into different volatility components. Then, we propose two new heterogeneous autoregressive (HAR) models by combining with the volatility components and leverage effect. Finally, we use high-frequency data for the S&P 500 as the study sample and perform parameter estimations on eight HAR-type models (including two new models). The results indicate that our models that are used to model 1-day, 1-week and 1-month future volatilities have an advantage over other existing HAR-type models. This advantage is substantial in the case of 1-month future volatility. In addition, the leverage contains significant in-sample prediction information for future volatility. (C) 2018 Elsevier B.V. All rights reserved.