In the last few years,
alternative methodological approaches have been proposed to measure relative contributions
when the trading sessions of the different markets overlap. Harris et al. (1995) infer
each market’s weight in price discovery from the error correction terms of a vector error
correction (VEC) model. Hasbrouck (1995) proposes a common trend representation to
model all the markets’ quotes. The fraction of the variance of the common stochastic trend
that is explained by each market’s innovations defines its information share. Finally,
Harris et al. (2002) use the common factor estimation method proposed by Gonzalo and
Granger (1995). In this last methodology, the long memory component of stock prices is
characterized as a weighted average of the contemporaneous trade prices. The weights
signify the incidence of trades that permanently move prices on each market. A special
issue of the Journal of Financial Markets (2002, vol. 5, (3)) provides some discussion on
these econometric methodologies.