A0656
Title: Semiparametric estimation method for quantile coherence with an application to financial time series clustering
Authors: Cristian Felipe Jimenez Varon - King Abdullah University of Science and Technology (Saudi Arabia) [presenting]
Ying Sun - KAUST (Saudi Arabia)
Ta-Hsin Li - IBM Watson Research Center (Austria)
Abstract: In multivariate time series analysis, the coherence measures the linear dependency between two-time series at different frequencies. However, real data applications often exhibit nonlinear dependence in the frequency domain. Conventional coherence analysis fails to capture such dependency. Quantile coherence, conversely, characterizes nonlinear dependency by defining the coherence at a set of quantile levels. Although quantile coherence is a more powerful tool, its estimation remains challenging due to the high level of noise. A new semi-parametric estimation technique is proposed for quantile coherence. The method uses the parametric form of the spectrum of the vector autoregressive (VAR) model as an approximation to the quantile spectral matrix, along with a nonparametric smoother. For each quantile level, the VAR parameters from the quantile periodograms are obtained, and then, using the Durbin-Levinson algorithm, the initial estimate of quantile coherence is calculated. Finally, it is smoothed across quantiles with a nonparametric smoother. Numerical results show outperformance over nonparametric methods. It is shown that quantile coherence-based time series clustering has advantages over ordinary coherence. For applications, the identified clusters of financial stocks by quantile coherence with a market benchmark are shown to have an intriguing and more accurate structure of diversified investment portfolios that may be used by investors to make better decisions.
