ESS Multiple Comparison Methods for Long Memory Processes: Application to US Stock Volatilities
Dr. Jaechoul Lee
The volatilities in stock prices often show long range dependence, showing significant autocorrelations for observations even in large time lags. A autoregressive fraction-ally integrated moving-average process adequately takes account of the long memory autocorrelation in stock price volatilities. A naive use of typical multiple comparison methods in volatilities can produce erroneous results since most multiple comparison methods are developed under independence or weak correlation assumptions. We propose modifying typical independence-based multiple comparison methods by incorporating the equivalent sample size method. Our simulation results show that our multiple comparison methods are appreciably accurate for comparing simulated long memory series. We also use our methods to identify high and low volatile companies in United States.
Bossart, Holly E. and Lee, Jaechoul, "ESS Multiple Comparison Methods for Long Memory Processes: Application to US Stock Volatilities" (2020). 2020 Undergraduate Research Showcase. 17.