A1260
Title: Robust testing for correlation in a non-i.i.d. setting
Authors: Liudas Giraitis - Queen Mary University of London (United Kingdom)
Yufei Li - Kings College London (United Kingdom) [presenting]
Abstract: The methodology of testing for correlation and cross-correlation is extended to a wider class of data and the finite sample performance of the robust testing procedures are studied. Models with non-smooth deterministic and stochastic (unit root type) scale factors, which are not covered in previous research, are studied theoretically and in the Monte Carlo experiments to compute and compare the size of both standard and robust tests. In the Monte Carlo study, we use models that include deterministic and stochastic scale factors mentioned above, demonstrating the performance of the robust statistics for testing for cross-correlation. In an empirical exercise, we test for autocorrelation and cross-correlation for some financial data to show the applications of the robust testing method. By comparing the results based on the standard statistics and the robust ones, the advantages of the robust testing procedures in empirical research are further uncovered and emphasized.
