The FYL measure is an extreme measure of embedded capital gains in that it assumes that all of the shares of stock were purchased at their FYL price.
George and Hwang (2007) argue that EWGO and FYL measures should dominate the EWGL measure when testing the capital gains lock-in hypothesis.
(7) I use the terms [FYL.sub.f], [EWGO.sub.t], and [EWGL.sub.t] to denote the end of year t values of the FYL measure, the equally weighted gain only measure, and the equally weighted gain-loss measure, respectively.
For the FYL (EWGO) measure, the mean value of 2.02 (0.77) is much higher than the median value of 1.02 (0.43) suggesting the presence of some large FYL and EWGO values in the sample, which is also reflected in significantly large maximum values for each of these measures.
For the measures of embedded capital gains, I find that the difference in the average raw monthly returns of the top and bottom decile portfolios over the five-year period is a statistically significant -0.57% (t-statistic = -4.07) for the FYL measure and -0.51% (t-statistic = -3.43) for the EWGO measure.
For example, to examine the role of the FYL measure in the reversals of cumulative return winners and losers, I estimate the following regression for each month k from July of Year t+1 to June of Year t+6:
This aspect is reflected in the significantly higher magnitude of embedded gains measures, FYL and EWGO, in Subsample 1 as compared to Subsample 2.
Note that the measures of locked-in gains, FYL and EWGO are highly correlated with the cumulative return with correlations of 0.77 and 0.79, respectively.
In Panel A of Table VII, I include the FYL winner and loser indicator variables as controls in the baseline regression specification of Table III.
After controlling for the FYL winner (Panel A), as well the equally weighted gain only winner (Panel B), the intangible return winner portfolios have negative and statistically significant returns in each of the five years in the holding period, with and without the January returns.