Title: Correlations between irregularly spaced time series
Authors: Jonas Andersson - Norwegian School of Economics (Norway) [presenting]
Abstract: The problem of analyzing relations between time series observed irregularly and at different points in time is addressed. In particular, computing covariances and correlations and the extension to linear regression is treated. The first difference of processes which are integrated of order one, I(1)-processes, are of special interest because they are often used as models for stock returns. The covariances of stock returns, in turn, is interesting since they are used in portfolio optimization. Different interpolation methods and modeling the irregularly spaced observations explicitly are investigated as well as simulation based methods. For the case where the return process and the process governing the times of trade are dependent, an unbiased estimation method is proposed and investigated.