 |
For an asset traded in multiple venues, an outstanding problem is how those places individually
contribute to the price discovery mechanism (the incorporation of information into
prices). I show that existing measures of price discovery lead to misleading conclusions when
using High-frequency data, due to uninformative microstructure noises. I then propose robustto-
noise measures, good at detecting “which market incorporates quickly new information”.
Using the Dow Jones stocks traded on NYSE and NASDAQ on the period March 1st to May
30th 2011, I show that the data are in line with my theoretical conclusions. In addition, when
the Information Share measure gives wide bounds making it unusable, my proposed robust IS
has very close bounds. I later obtain that price discovery mostly happens on NYSE and is positively
correlated with its liquidity and its market share in small and big size transactions. For
NASDAQ-listed stocks, large quantities trades do not convey information and NASDAQ contribution
to price discovery increases slightly the days with macroeconomic announcements.
