A0712
Title: Adjusted-range self-normalized autocorrelation tests
Authors: Jiajing Sun - University of Chinese Academy of Sciences (China)
Yongmiao Hong - Cornell University (United States)
Oliver Linton - University of Cambridge (United Kingdom)
Xiaolu Zhao - Dongbei University of Finance and Economics (China)
Xiaolu Zhao - Dongbei University of Finance and Economics (China) [presenting]
Abstract: Two autocorrelation tests are proposed using the adjusted-range based self-normalization, extending previous work. Despite focusing on testing the autocorrelation, our method can be generalized to testing the significance of the class of approximately linear statistics, such as the marginal mean, marginal variance and quantiles. Through comprehensive simulation studies on both standard time series and count series data, we confirm that our adjusted-range based tests are particularly suitable for detecting the presence of serial correlation, significantly outperforming existing self-normalized tests. Empirical results on the periodic presence of unit root in realized volatility series of the world's major stock indices as well as on testing the serial autocorrelation in COVID-19 count series further confirm the validity of our approach.
