these cases we can use all firms that were members of the S&P 500 at the
end of the years 1992 to 2003.
Matching the Execucomp data with the CRSP and the Compustat uni-
verse gives us ownership information for 15,600 firm years (97.5% of the
theoretically maximum firm years - the remaining firm years are missing
due to matching problems)5 from a total of 2,405 different firms.
The relevant data item in the Execucomp database is Shrownpc, which
gives the percentage of the firm’s shares owned by an officer.6 Table 1
describes the distribution of the fraction of the company’s shares owned by
the largest shareholder among the officers covered by Execucomp.
+ + + Please insert TABLE 1 about here + + +
Approximately one out of five S&P 500 firms have an officer who owns
a positive amount of shares (Panel A). The fraction of S&P 500 firms that
have a CEO who owns more than 5% of the firm’s outstanding shares ranges
5 To make sure that our results are not influenced by matching problems, we also
employ the dataset described in Dlugosz, Fahlenbrach, Gompers, and Metrick (2006),
which provides ownership information for 7,873 firm years between 1996-2001 from the
Investor Responsibility Research Center (IRRC). Our main results using this alternative
data source are very similar (see Table 4, Panel D).
6 Our results are unaffected if we restrict our analysis to non-restricted shares. We
construct a measure of ownership of unrestricted shares similar to the Shrownpc measure.
The results we obtain are very similar in economic as well as statistical terms. However,
our measures of the fractions of unrestricted shares is plagued by a data problem, because
the reporting dates for unrestricted shares and the reporting dates of outstanding shares
do not always conincide. Thus, we decided not to report them. Furthermore, creating
portfolios solely based on option ownership does not provide any significant abnormal
returns.