Year: 2026
Type: article
Abstract: ABSTRACT Recent literature documents that the variance risk premium (VRP) loses its predictive power for market returns during the pandemic. We show that this failure is largely attributable to the standard measurement of realized variance, which equally weights overnight squared returns and intraday realized variance. Simulations demonstrate that the predictive performance of the conventional VRP deteriorates as overnight volatility and jump activity increase. To address this issue, we propose a new linear forecasting model for expected variance that separately captures the dynamics of daytime and overnight variance. Empirically, our model‐based VRP restores significant in‐sample predictability. In out‐of‐sample exercises, it achieves the best performance among competing predictors and generates substantial economic gains for a mean‐variance investor over short and medium horizons.
Source: Journal of Forecasting
DOI: 10.1002/for.70237
Authors: Zhenxiong Li, Xingzhi Yao, Marwan Izzeldin
Language: English
Subfield: Finance
Domain: Social Sciences
Open Access status: bronze
OpenAlex ID: W7225450846