In this paper we propose an innovative measure for information flows between stock exchanges. We develop an intensity-based information share using Russell’s (1999) autoregressive conditional intensity model. Thereby we maintain the irregular nature of financial high frequency data and use durations and timing of price changes to determine the informationally dominant market. From our empirical application to US-listed Canadian stocks we conclude that the home market mostly reflects information first. On the basis of a cross-sectional analysis we find a positive correlation between the intensity-based information share and liquidity.

The trading of securities on multiple markets raises the question of each market’s share in the discovery of the informationally efficient price. We exploit salient distributional features of multivariate financial price processes to uniquely determine these contributions, thereby resolving the main drawback of the widely used Hasbrouck (1995) methodology, which merely provides upper and lower bounds of a market’s information share. We show how tail dependence of price changes, which may emerge as a result of differences in market design, can be exploited to estimate unique information shares. Two empiricalapplications illustrate the practical use of the new methodology.