Estimating time-varying variances and covariances via nearest neighbour multivariate predictions: applications to the NYSE and the Madrid Stock Exchange Index

Abstract

In this article, we present a technique to obtain the time-varying covariance matrix for several time series for nearest neighbour predictors. To illustrate the use of this technique, we analyse the time-varying variances and correlations between the daily returns on two equity stock market indexes, the New York Stock Exchange and the Madrid Stock Exchange Index.

Acknowledgment

This research is supported by the Spanish Ministry of Science and Education through project SEJ2006-07701.2