Volatility forecast of country ETF: The sequential information arrival hypothesis

Abstract

This paper uses daily data to examine whether the sequential information arrival hypothesis is supported in single country Exchange Traded Fund (ETF) market, and to model the forecast of ETF's volatility. The work is based on incorporating lagged trading volume into the ‘heterogeneous auto-regressive’ (HAR) model of regression realized range-based volatility (RRV) on realized range-based bi-power variance (RBV) (HAR–RRV–RBV-cum-Vol model, hereafter), in an attempt to improve the overall forecast of realized variance. We find that the forecasting performance of the HAR–RRV–RBV-cum-Vol model is better than other models for both in-sample and out-of-sample forecasts. The results support the sequential information arrival hypothesis in single country ETF market, by which lagged volume is available to predict current volatility.