Many academic studies use fixed preannouncement event days (e.g., -20,-42, or -63) to measure takeover premiums. In this paper, we show that the use of traditional fixed windows generates premiums that are underestimated by as much as 8 percentage points. This downward bias is especially severe for transactions with long processes (e.g., target-initiated deals). We take account of this bias by hand collecting deal initiation dates and show that using these dates results in measured premiums that give contradictory conclusions to those found in existing literature. We also offer guidance for measuring premiums if hand collecting data is impractical.

Using proprietary institutional trade data, we construct a price impact measure that represents the costs faced by institutional investors. We show that many widely used liquidity measures do not adequately capture institutional trading costs. We then find that institutional trading costs are not dramatically impacted by decimalization, casting doubt on the widely used identification strategy that employs decimalization as an exogenous shock to liquidity, particularly institutional liquidity. Indeed, we find that conclusions from prior research are significantly altered when we measure liquidity using institutional trading data.