CUSUM control schemes for monitoring the covariance matrix of multivariate time series. Issue 4 (4th July 2017)

• Record Type: Journal Article
• Authors: Bodnar, Olha; Schmid, Wolfgang
• Is Part Of: Statistics. Volume 51:Issue 4(2017)
• Page Start: 722
• Page End: 744
• Publication Date: 2017-07-04
• Subjects: CUSUM control charts -- statistical process control -- multivariate time series -- financial application; Mathematical statistics -- Periodicals; 519.505
• DOI: 10.1080/02331888.2016.1268616
• Languages: English
• ISSNs: 0233-1888

Abstract

ABSTRACT: Modified cumulative sum (CUSUM) control charts and CUSUM schemes for residuals are suggested to detect changes in the covariance matrix of multivariate time series. Several properties of these schemes are derived when the in-control process is a stationary Gaussian process. A Monte Carlo study reveals that the proposed approaches show similar or even better performance than the schemes based on the multivariate exponentially weighted moving average (MEWMA) recursion. We illustrate how the control procedures can be applied to monitor the covariance structure of developed stock market indices.