A residual-based multivariate constant correlation test (Q1669884)

A constant correlation test is proposed which allows for non-constant marginal variances in multivariate time series analysis. A bootstrap approximation is used to obtain the corresponding critical values for the test. A dataset of eight European stocks (Euro Stoxx 50) was used to validate the proposed multivariate constant correlation test. The selected dataset includes the period of the recent global financial crisis. The constructed theory paves the road for the development of a general theoretical foundation of the validity of bootstrap approximation in this constant correlation test.