A new correlation for bivariate time series with a higher order of integration

Abstract

A new correlation coefficient for a bivariate time series that takes into account the first differences of the data has recently been proposed. They have made use of extensive simulations to show that their new correlation is better than the usual Pearson’s correlation coefficient. In this work, we propose a new extended version of a correlation coefficient which is capable of capturing higher order integration between bivariate time series. We support this idea via numerous simulations under different settings.

Keywords:

• Correlation
• Bivariate
• Time series
• Cointegration
• Non-stationarity
• Random walk

SUBJECT CLASSIFICATIONS:

• 37M10
• 62M10
• 91B84

Disclosure statement

No potential conflict of interest was reported by the authors.