A0708
Title: The tail efficiency hypothesis: Extreme-value perspective on market efficiency
Authors: Raphael Huser - King Abdullah University of Science and Technology (Saudi Arabia)
Jordan Richards - King Abdullah University of Science and Technology (Saudi Arabia)
David Bolin - Chalmers (Sweden)
Junshu Jiang - King Abdullah University of Science and Technology (Saudi Arabia) [presenting]
Abstract: In economics, the market efficiency hypothesis posits that asset prices reflect all available information. Several empirical investigations show that market efficiency under extreme situations is relatively low. Although many models for extremal dependence have been developed over the last few decades, these mainly focus on characterizing the behaviour of a random vector when it exhibits only positive extremal dependence, i.e., all components are jointly extreme in the same direction. This limitation makes them unsuitable for studying dependence in financial markets where it is not uncommon for data to exhibit both positive and negative extremal dependence. To jointly model positive and negative extremal dependence, regular variation models are constructed on the entirety of Rd space and develop a bivariate measure for the asymmetry between the strength of extremal dependence in adjacent orthants. The so-called directional tail dependence (DTD) measure allows us to construct the tail efficiency hypothesis, an analogue of the market efficiency hypothesis, for behavior of the market in its tails. Asymptotic results for estimators of the DTD are described, and testing is discussed via permutation-based methods. Empirical results for China's derivatives market support that the market is generally tail-efficient, with a few exceptions which indicate potential arbitrage opportunities.
