Abstract
This study adds to the literature on estimating the probability of informed trading (PIN), which interests market microstructure empiricists, by proposing the q-adjustment to the process of estimating PIN. Due to challenges in accessing the data necessary for distinguishing between buyer- and seller-initiated transactions, prior studies commonly rely on trade classification algorithms. However, these algorithms result in misclassification and, as a result, the PIN may be underestimated. Specifically, this study shows that an underestimated PIN leads to a spuriously large slope coefficient but a low adjusted R-squared for the regression of opening bid-ask spread on the PIN and that both trade misclassification and PIN estimation error become more serious for recent-year data. The authors find that the q-adjustment is appropriate for all extended PIN models and is effective in both simulated and historical data. In addition, the q-adjustment outperforms the existing remedies across longer time spans.
Published Version
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call