Information Shares in Stationary Time Series and Global Volatility Discovery

Abstract

Standard price discovery measures, particularly information shares, rely on the concept of co-integration for non-stationary time series. For the definition of information shares, the existence of a permanent impact of innovations is crucial, as these shares measure the relative contribution of different markets to this permanent impact. For stationary time series such as interest rates, CDS prices, or volatilities, a permanent impact however does not occur and thus the concept fails and lacks an interpretation. In this paper we extend the concept of information shares to the case of stationary time series. We suggest a price discovery measure based on the well-known variance decomposition for stationary time series, aiming to be equivalent to Hasbrouck's information share. We show that this new price discovery measure converges to the standard information share in the border case of non-stationarity.