Earnings Announcement Premia and the Limits to Arbitrage

We re-examine the existence of earnings announcement-day premia and find that they persist beyond the sample period used in prior studies (ending in 1988). The magnitude of the premia is lower for all sub-periods when we use the expected rather than actual announcement dates used in prior studies. Moreover, the premia are not present following earnings pre-announcements. Finally, we find that limits to arbitrage are a likely explanation for the continuing presence of the premia despite the fact that firms on earnings announcement dates have a significantly positive Jensen's alpha and a higher Sharpe Ratio than on non-announcement dates.