Wavelet Transforms of Skewed Gaussian Long Memory Processes

• Activity Number: 235
• Type: Contributed
• Date/Time: Monday, August 4, 2014 : 2:00 PM to 3:50 PM
• Sponsor: Business and Economic Statistics Section
• Author(s): Kyungduk Ko*+
• Companies: Boise State University
• Keywords: Long Memory ; Skewed Gaussian ; Volatility ; Wavelet Transform

Abstract

Motivated by Lee and Ko (2007) but not limited to the study, we consider skewed Gaussian long memory processes for modeling asymmetric long memory data. Applications with long memory time series often need Gaussian assumption, but we encounter real data that exhibit long range dependence but do not follow a Gaussian distribution. Typical real-world examples can be found in volatilities in absolute (or squared) stock returns whose distributional shapes are severely skewed to the right. Here, we consider wavelet transforms of skewed Gaussian long memory processes in order to make statistical inference procedures simple as in the case of wavelet transformed Gaussian long memory series. We check and show if wavelet transforms of such processes still have uncorrelated feature. Moreover, we present an estimation procedure for the parameters of skewed long memory processes in wavelet domain.