Book Description
The SEC's EDGAR log files provide a direct, powerful measure of attention from relatively sophisticated investors. We apply this measure to a sample of earnings announcements from 2003 to 2016. We find that the stock market is less surprised, and the post-earnings-announcement drift is weaker for earnings announcements receiving more pre-announcement investor attention, measured in downloads by humans from EDGAR. We further show that it is profitable to utilize the different drift patterns. An attention-based portfolio without the SEC reporting lag that longs stocks with the lowest investor attention and most positive earnings surprises and shorts stocks with the lowest attention and most negative earnings surprises generates a statistically significant monthly alpha of 1.24% after adjusting for standard asset pricing factors.