Multiscale correlations of volatility patterns across the stock market

摘要

Volatility is an important variable in financial markets. In this paper we study to what extent patterns in volatility changes are shared across different sectors of the U.S. stock market as a function of the time horizon. Wavelets are used to handle the multiscale aspect of the problem. The log Garman-Klass estimator is used as a proxy to the unknown historical log volatility. Dissimilarities are calculated from correlation coefficients. Classical multidimensional scaling allows for the visualization of results. These results suggest that the multiscale aspect of the problem is a very crucial one as the proximity pattern changes as a function of the time scale. This supports the use of wavelets in the analysis of the characteristics of the stock market and shows that wavelets might be practically useful for understanding uncertainty in the market.