**Expected Idiosyncratic Skewness Data**

Below you can find updated estimates of *expected* idiosyncratic skewness through the end of 2016, similar to those described in Boyer, Mitton, and Vorkink (2010). The first column identifies the year and month for date *t*, where *t* ranges from July 1969 through December 2016. The second column identifies the permno. The third column gives the forecast of expected *idiosyncratic* skewness observed at the end of month *t* for the distribution of daily returns over months *t*+1 through *t+*60. The fourth column gives the forecast of expected *total* skewness observed at the end of month *t* for the distribution of daily returns over months *t*+1 through *t+*60. The data file is approximately 80 megs.

Data
Methodology

To investigate the pricing of idiosyncratic skewness we sort securities into quintiles by expected idiosyncratic skewness at the end of month *t* and examine returns over month* t+1* (as in our paper). Below we report the average value-weighted return, as well as the Fama-French alpha for the value-weighted return of each quintile portfolio for *t+1* equal to January 1987 through December 2016. Standard errors are Newey-West adjusted. (We begin the sample in January 1987 due to data restrictions on turnover. See "Methodology" link for further details.)

We also compare the returns of a typical stock with high ex-ante skewness to that of a typical stock with low ex-ante skewness. To investigate this issue, we calculate the *median* return (in the cross section) each month for each skewness bin, and report both the time-series average and Fama-French alpha.