The Gonzalo-Granger Component Share

When the same economic asset trades in two places at once — a stock on its primary exchange and in dark pools, or a bond and its futures contract — their prices are cointegrated: they can wander apart briefly but are tied to a single shared long-run "efficient price." A natural question is which venue actually discovers that shared price first, so that the other is just following along with a lag. The Gonzalo-Granger component share answers this by looking at the error-correction model that links the two price series: whichever venue's price barely reacts when the pair drifts apart, while the other venue's price does most of the adjusting back toward equilibrium, is assigned the larger share of price discovery — the one that moves toward the deviation is doing the following, and the one that stays put is the one setting the pace.

Concretely, the model estimates how much each venue's price correction coefficient responds to the prior period's price gap between the two; the venue with the smaller correction coefficient is contributing more to the permanent, common component, so its component share is higher. A component share near 1 for a lit exchange and near 0 for a dark venue would mean the exchange is doing essentially all of the price discovery, with the dark venue simply printing at whatever the lit market has already settled on. This is one of two commonly used decompositions for this question, alongside the related Hasbrouck information share, and both are standard tools when comparing how much dark or off-exchange trading contributes to setting prices versus simply consuming them.