A0685
Title: Long memory and structural breaks in testing the implied-realized volatility relation
Authors: Yi-Chi Chen - National Cheng Kung University (Taiwan) [presenting]
Abstract: The recent literature has shown that the implied-realized volatility relation can be well-modelled as fractional integration and thus provide evidence for the implied volatility unbiasedness. On the other hand, it has been found that the observed long-range dependence could be generated by structural breaks, in particular, uncommon breaks. The unbiasedness hypothesis is revisited, and the problem is reformulated as testing for cross-correlation between two volatility series. This testing procedure takes advantage of the residual cross-correlation function by first pre-whitening two-time series. Such a correlation test is free from parametric volatility models that depend on specific distributional assumptions and are subject to size distortions. Simulation studies will be considered to demonstrate the finite sample properties of the proposed correlation test as opposed to its conventional counterparts. The volatility dynamics for the major exchange rates are investigated to empirically verify whether implied volatility is an unbiased forecast of realized volatility and whether the unbiasedness anomaly remains a statistical artefact in a pairwise cross-dependence testing framework.