Exercises of executive stock options on the vesting date.
It is well known that it is never optimal to exercise an American call option on a non-dividend-paying stock prior to expiration. To do so surrenders the time value of money associated with the exercise price. It is also well known that if the stock is dividend paying, that early exercise may be optimal, primarily because the exercise price of the option is not adjusted to reflect the ex-dividend drop in the price of the stock. It is also well recognized that it may also be optimal to exercise an executive stock option (ESO) early for a number of reasons.
The nontradability of ESOs implies that the time value of the option can never be captured by the executive. Thus, an executive essentially has a choice between a lottery over future intrinsic values of the option and the current intrinsic value of the option. Clearly, in certain circumstances, a risk-averse executive might prefer the current intrinsic value to the lottery over future intrinsic values (Huddart, 1994; Kulatilaka and Marcus, 1994).
However, risk aversion may not be the only motivation for early exercise by executives. Ofek and Yermack (2000) indicate that for the four years from 1992 to 1995, most of the shares acquired through options are sold for cash immediately after exercise. They regard this behavior as perfectly consistent with executives applying modern portfolio theory to their own holdings. That is, they attempt to diversify away the unsystematic risk associated with concentrating a significant portion of their wealth in a single asset. Thus, executives may have a diversification objective when they exercise ESOs early and then dispose of the acquired shares. These incentives would be stronger the greater the concentration of the executive's wealth in his firm's shares or the greater the unsystematic risk of those underlying shares. Meulbroek (2001) provides a method for valuing the deadweight cost of this underdiversification.
Another factor potentially affecting early exercise is the liquidity need of the executive. Robicheaux, Fu, and Ligon (2008) find that the real value of executive cash compensation (i.e., salary and bonus) is essentially flat over the period 1992-2004 but that the percentage of stock option compensation is significantly higher over the period 1992-2004 versus the base year of 1992 (31.8% vs. 20.2%). Liu and Yermack (2007) find that CEOs frequently liquidate company shares and options to finance acquisition of new principal residences, and that the extent to which they do so is negatively correlated with future firm performance. As the proportion of executive compensation represented by incentive compensation rises, executives may have greater need to occasionally liquidate positions in order to accommodate their liquidity needs.
Finally, early exercise of ESOs could be related to insider information. A number of studies indicate such a link. Carpenter and Returners (2001) test the relation between private information and the timing of ESO exercises by examining a data sample from 1984 to 1995. They find that during the period 1984 to 1991, when the SEC mandated insiders to hold stocks for six months after ESO exercises, exercises preceded positive abnormal returns. After this mandate was eliminated in 1991, however, they observe ESO exercises are followed by negative abnormal stock returns. Huddart and Lang (2003) examine option exercises by both executive and junior employees at seven firms and find that when option exercise frequency is high, stock returns are 10% lower than when option exercise frequency is low. Brooks, Chance, and Cline (2009) document negative abnormal returns in the year following exercises between 1996 and 2003.
We analyze the motives for the exercise of executive stock options at the point when it first becomes possible to do so. In particular, we focus on exercise decisions within a two-day period after the vesting date. We focus on this window because it is the point when the maximum time value of money is lost by exercise and because vesting date exercises are likely to represent the solution to a constrained optimization problem. That is, executives may have desired to exercise at an earlier date but were prevented from doing so because their options were not vested. As a result, we might expect the motivations for vesting exercises to be less time sensitive than the motivations for early exercise in general. For example, the need for portfolio diversification may be less time sensitive than private information regarding expected stock performance.
We observe that 12.06% of executive option exercises occur within two days after their vesting dates. What motivates these vesting date exercises? Executives clearly lose time value of money in the options by exercising them early, but their attitudes toward risk, their diversification needs, their liquidity needs, and their private information may offset this. We examine this issue by comparing vesting exercises to other early exercises to determine the relative importance of various factors to a vesting date exercise versus an early exercise at a later date. While recent work (e.g., Carr and Linetsky, 2000; Finnerty, 2005; Cvitanic, Wiener, and Zapatero, 2008) has incorporated the reality of early exercise into valuation models of ESOs, to the best of our knowledge, this paper is the first in the literature to study the relative importance of the specific motivations of vesting date exercise versus other early exercises.
Our tests provide strong evidence that ESO exercises on the vesting date are highly motivated by executives' needs to manage their personal wealth portfolios. Less diversified executives with riskier stocks are more likely to exercise on the vesting date than at any other point prior to option expiration. We also find that, in general, while inside information may produce an incentive to early exercise, it does not produce a differential incentive to vesting date exercise. This is intuitive since private information might have a relatively short shelf life compared to diversification needs, and since vesting exercise is likely to be a constrained optimization, short-lived factors might be expected to be less important. Company size and industry also influence vesting date exercise versus other early exercises.
The rest of the paper is organized as follows. Section I discusses our hypothesis development and model. Section II describes our data. Section III presents our empirical findings regarding motives for vesting date exercises versus other early exercises. Section IV provides robustness tests and Section V concludes.
I. Testable Hypotheses and Model
A. Testable Hypotheses
The primary purpose of this study is to study the motivations of vesting date exercises. We define a vesting date exercise as one where the option is exercised within two calendar days after the ESO becomes vested. A two-day window is set for a vesting date exercise so that we do not discriminate against options vested on Saturdays or Sundays.
We test two hypotheses related to the relative likelihood of vesting date exercises versus other early exercises:
H1: Executives' needs to manage their personal wealth portfolios are more likely to trigger exercise on the vesting date than at any other single point prior to expiration.
H2: Negative private information that the executive possesses is less likely to trigger exercise on the vesting date than at any other single point prior to expiration.
The rationale for our hypotheses is that since vesting exercise is the result of a constrained optimization problem, factors that are relatively time sensitive, like private information, are relatively less likely to trigger a vesting exercise, which may occur later than the optimal exercise date. On the other hand, factors that are relatively less time sensitive, like the need to diversify a portfolio, may be relatively more likely to trigger an exercise on the vesting date than on any other single date prior to expiration.
Our model of vesting exercise is as follows:
y = [[beta].sub.0] + [[beta].sub.1] + W [[beta].sub.2] [PHI] + [[beta].sub.3] X + [epsilon], (1)
y is a dummy variable taking on the value one if the option exercise occurs within two days of the vesting date and zero if exercise occurs at any other date prior to one month prior to expiration,
[[beta].sub.0] is the regression intercept,
W is a vector of variables related to executive personal wealth management,
[PHI] is a vector of variables related to executive private information,
X is a vector of control variables,
[[beta].sub.1] are vectors of coefficients (i = 1, 2, 3), and
[epsilon] is an error term.
In testing Hypotheses 1 and 2, we run this model on all options exercises between the vesting date and one month prior to expiration with the control variables defined in relation to the exercise date. (1) The regression methodology used to estimate Equation (1) is the logit model. Our dependent variable is binary, and the logit regression yields unbiased results about vesting exercise propensity.
C. Test and Control Variables
We test for the relation between vesting date exercises and wealth management concerns by including variables that relate to three aspects of portfolio management: the degree of risk aversion, diversification concerns, and liquidity issues.
The degree of risk aversion is related to personal background, experience, and net worth and is frequently directly unobservable by the empirical researcher. Becker (2006) finds that wealthy CEOs are less risk averse and take more incentive-based compensation. CEOs might also be expected to be wealthier than other executives and decreasing absolute risk aversion (DARA) is frequently assumed to hold in order to generate comparative statics consistent with observed behavior (e.g., Ligon and Thistle, 2008). Thus, wealthier CEOs might be expected to be less risk averse than junior executives. The less risk averse an executive, the less need for an early exercise and, in particular, a vesting date exercise. CEOs might, however, also be more concerned about control rights, which might lead them to exercise early to obtain voting shares. To control for the potential differences in CEO and other executive behavior, we use a dummy variable equal to one if the executive is CEO, and zero otherwise.
Executives' net worth could be suboptimally diversified if a large portion of their wealth is tied to company performance through stock ownership and option holdings. Exercising options on the vesting date and immediately selling the shares received is one approach for executives to manage the excessive risk imposed by the financial commitment to their companies. Diversification needs depend on current exposure to the firm's equity risk, including both stock ownership and ESOs, and the level of that risk. The effect of stock and option ownership need not be monotonically related to vesting exercise, however.
Assuming they have equal total wealth, an executive holding a large amount of company stock would clearly have a greater portion of his portfolio wealth concentrated in a single asset and thus be less optimally diversified than an executive holding a smaller amount of company stock. Thus, executives with larger stock holdings would clearly appear to have a greater diversification motivation for vesting date exercise. On the other hand, a large stock ownership could also imply a wealthier and, hence, less risk-averse executive. Furthermore, executives with a large percentage stock ownership consider the control rights associated with the large holding and portfolio diversification might be relatively less important for them. In short, we cannot predict the effect of executive stock ownership on the propensity for vesting date exercise. On its face, we would expect that a large stock holding might represent a positive effect on vesting date exercise because of diversification needs. On the other hand, the decreasing risk aversion derived from higher wealth (i.e., a larger stock holding) and the possible role played by the utility of control rights to the executive could suggest a negative impact on vesting exercise. Which effect is stronger is an empirical question. We include both percentage of stock owned and value of stock owned as proxies for the amount of executive stock holdings.
The amount of ESOs held also would be expected to be related to the executive's diversification needs, but again the empirical relationships may not be straightforward. An executive's unexercised ESOs consist of two types: vested options and unvested options. Executives can exercise the former and rebalance their portfolios with freedom. Unvested options, however, cannot be converted into other assets. They represent an illiquid undiversifiable stake in an asset in which the executive may already be overinvested. Hence, we predict that between these two unexercised option variables, the amount of unvested options is a more significant measure for executives' diversification needs, while the amount of vested options is a relatively less significant measure. Vested options, being liquid, share some of the same ambiguities as stock holdings. The diversification need to reduce exposure to the company's equity risk suggests a positive relation. On the other hand, those exercisable option holdings increase wealth and decrease risk aversion, suggesting a negative relation. Also, a larger amount of vested options means that if the executive exercises options on a first-in first-out or pro rata basis, then the probability of a vesting date exercise would fall. Finally, we also analyze the effect of a new option grant around the exercise date on the propensity of a vesting date exercise. We include a dummy variable that takes on the value of one if the executive is awarded a new option grant within a one-month period around (i.e., -1, +1) the exercise date. While we might expect option reloads or other new grants to spur early exercise, whether they have a differential affect on vesting date exercise is an empirical question.
Finally, the need to diversify would be expected to be positively related to the risk of the company stock. The cost of the suboptimal diversification implied by the concentration of the executive's wealth in the firm will be higher the greater the risk of that undiversified holding. Higher stock volatility would make a normal option more valuable. However, Kulatilaka and Marcus (1994) suggest that the values of ESOs fall when stock volatility rises. Since the executive cannot capture the time value of the nonmarketable option, the increase in volatility does not increase the value of the option to the executive. Thus, we predict that stock volatility, as a measure of risk, is positively related to vesting date exercise.
Ofek and Yermack (2000) find that executives tend to sell the shares received from an ESO exercise for cash immediately after the exercise. Hence, the exercise of the stock options could be related to an executive's need for liquidity. We use the sum of previous-year salary and bonus as a measure of liquidity needs. Salary and bonus is measured as of the exercise date and hence the sign and significance of this coefficient indicates whether liquidity is more important than (negative significance), less important than (positive significance), or equally affects (insignificant) vesting versus other early exercises.
Since the 1992 change in the tax law limiting the deductibility of non-performance-related executive compensation, the variation in the cash component of compensation (i.e., salary and bonus) has been quite small (Perry and Zenner, 2001; Robicheaux, Fu, and Ligon, 2008). Consequently, we include an additional measure related to the timing of the option exercise in an attempt to capture liquidity-related effects. Most US corporations have fiscal years that correspond to the calendar year and most pay bonuses related to year-end performance measures, based on either accounting measures or stock performance. As a consequence, many executives in the United States have a surge of cash receipts around the turn of the year, so they should have less liquidity needs in January compared to other months of the year. (2) If one exercises at a time when one's cash holdings are flush, then liquidity needs are less of a motivation for that exercise. Thus, if liquidity is relatively less important to vesting exercises, the coefficient on this variable will be positive; if it is relatively more important, the coefficient will be negative. (3)
Since liquidity needs might be considered more time sensitive, we might expect a positive coefficient on the liquidity-related variables. However, if the liquidity is desired for big-ticket discretionary consumption, then the liquidity needs might be considered more time invariant and we might expect a negative coefficient on these variables. Ultimately, whether liquidity needs are transitory or more long-lived is an empirical question.
With respect to the role of private information, the finance literature on executive stock options documents the association between ESO exercise and future stock returns. Hence, the early exercise of ESOs is likely to be consistent with the strategy of avoiding future losses anticipated through private information. To test whether negative insider information leads to a differential effect on vesting date exercises, we study the correlation between stock returns after the exercise date. One of the variables we use is the intercept of a postexercise date one-year Fama-French-Carhart four-factor model, which is a commonly used measure of abnormal returns. (4) We also test whether executives can spot changes in long-run stock returns by including a variable that is the difference between the postexercise one-year four-factor intercept and preexercise one-year four-factor intercept. If this variable has a negative relationship to the propensity for vesting exercise, it suggests that private information is relatively more important to vesting exercise than to later early exercise. Long-run stock performance measures are, of course, sensitive to the methodologies used to generate them. To provide robustness to our results, we also include the postexercise one-year buy-and-hold abnormal return adjusted by the market index as one of our test variables for long-run performance. We also examine whether vesting date exercise reflects insider information in terms of short-run stock price movement. We include two variables to capture the possible insider information effect from short-run performance. One is the postexercise two-week abnormal stock returns, where the risk adjustment is the market index. The other is the postexercise two-week abnormal returns minus the preexercise two-week abnormal returns, where the risk adjustment is the market index.
Jeng, Metrick, and Zeckhauser (2003) and Lakonishok and Lee (2001) indicate that insider open market transactions are a good proxy for future stock return. We calculate the net firm-level insider open market transactions in one month preceding the exercise date. That is, we calculate the number of transactions where shares are purchased by firm insiders less the number of transactions where shares are sold by firm insiders over the total number of transactions by firm insiders. The insider transaction data used to calculate the net purchase ratio are the open market transactions that occur within one month before exercise dates and are not related to any ESO transaction. For observations without any insider purchase or sales over this one-month period, we assign zero to be the net purchase ratio. If private information is a less important factor for vesting exercise than other early exercise, we should observe a positive relationship between the insider open market net purchase ratio and vesting exercise.
Control variables include the number of days from exercise to expiration, the previous year's dividend amount, the risk-free rate, company size, a high-tech company dummy, a utility firm dummy, a financial firm dummy, and year dummies. The number of days from exercise to expiration is a measure of the time value of money lost on exercise. The value of options is negatively related to the amount of dividends. The time value of money lost through early option exercise is also positively related to the risk-free rate. The risk-free rate in our tests is the one-month US Treasury bill rate on the exercise date and is obtained from Kenneth French's Data Library at Dartmouth College. Company size is the company's equity market capitalization, where the market capitalization is in millions. The high-tech company dummy, utility firm dummy, and financial firm dummy are to control for any industry effects. (5) The proportion of vesting exercise differs over the years, and we use year dummies to control for this variation.
Detailed definitions of the variables used to test Hypotheses 1 and 2 and the expected signs of these variables, as well as the control variables included in Equation (1), appear in Table I.
Our data sample contains information about option exercise and insider open market transactions, executive compensation, and stock price. We obtain option exercise and insider open market transaction information from the Thompson Financial Network Insider Filing Data (TFI) database. TFI contains information on insider option exercises and open market transactions compiled from Forms 3, 4, 5, and 144 filed with the SEC from 1996 forward. Executive compensation information is provided by Standard & Poor's (S&P) ExecuComp data tape. ExecuComp covers the S&P 1500, but it also includes companies that were once part of the S&P 1500, but were subsequently removed from the index, yet are still trading. The S&P 1500 includes the S&P 500, S&P MidCap 400, and S&P SmallCap 600. A firm must have at least a market cap of $300 million to be included in the S&P SmallCap 600, so ExecuComp includes relatively larger firms. ExecuComp normally reports compensation of five executives for a given firm-year. Stock market data are from the CRSP database.
Our data collection process is as follows. First, we collect all insider option exercises from the TFI database from 1996 to 2005. We only keep observations that are reported as employee stock options, incentive stock options, or nonqualified stock options. We then exclude observations with missing transaction dates, vesting dates, or expiration dates, and this leads to 112,838 insider option exercise observations. The TFI database reports multiple observations if an insider exercises options with different vesting dates or different expiration dates on the same day, and these multiple observations are assigned the same document control number. Thus, we collapse these 112,838 observations into 54,459 observations by document control number. Because the vesting date is a critical variable in our paper, we exclude 15,624 observations that exercise options with multiple vesting dates from our primary analysis and that leaves us with 38,835 ESO exercises. (6) We do consider multiple vesting date exercises in our robustness tests.
We merge data on those 38,835 option exercises with the executive compensation data from ExecuComp by exercise years, companies, and executive names. The compensation data we obtain from ExecuComp are from the previous year-end relative to the exercise date, regardless of the vesting exercise decision. Because ExecuComp focuses on S&P 1500 firms and, normally, the top five executives in a company, our sample size decreases to 7,527 option exercises after this matching.
Our primary purpose in this paper is to examine executives' motivations for exercising stock options on the vesting date versus other early exercises. As it is irrational to exercise out-ofthe-money options, we exclude 605 observations where the options are out of the money on the vesting date. This leaves 6,922 exercises that are in the money on the vesting dates. The TFI database records option exercises from 1996 forward, and this implies that for ESO exercises with vesting dates before 1996, they would be recorded by TFI only if they are not vesting exercises. Including these observations could lead to the wrong conclusion that insiders do not engage in vesting exercise before 1996. Hence, we exclude these observations from our sample to avoid potential bias to our results. Our sample is then 6,109 ESO exercises by 2,223 executives from 846 companies.
Table II presents the number and percentage of vesting date exercises by the year ESOs become vested. For ESOs that become vested in 1996, only 29 exercises on the vesting dates occur in that year, while the number is significantly higher in other years. For instance, 124 vesting date exercises are observed out of 780 total exercises for ESOs vested in 1997. The percentage of vesting date exercises ranges from 5.03% in 1996 to 32.70% in 2005. However, the percentage of vesting date exercises in 2005 will be upwardly biased because of the TFI database structure. We obtain insider option exercise data until 2005 from TFI, so for the ESOs vested in 2005, they would be included in our TFI data set only if they are also exercised in 2005. This decreases the number of nonvesting exercises for ESOs vested in 2005 and inflates the percentage of vesting date exercises. For similar reasons, we should view 12.06% as an upper bound statistic on vesting date exercises from 1996 to 2005. If we exclude ESOs vested from 2003 to 2005, 496 (10.11%) out of 4,905 total ESO exercises are vesting date exercises.
Early exercise becomes more prominent after we break down the ESO exercises based on the number of years or months from their vesting dates. We report the categorization of exercises by year and month in Table III. As shown in Panel A, more than half of the exercises in our full sample occur within two years after the vesting date, and the proportion of exercises in the first year after the vesting date is 36.1%. The proportion of exercises also decreases gradually by year with the only exception being year 9, as the number of exercises in year 9 is higher than those of years 7 and 8. As shown in Panel D, the mean (median) number of days from vesting to expiration is 2,484 (2,559), which means that if executives want to maximize the time value of their ESOs, they should wait approximately seven years after the vesting date to exercise. (7) However, more than one-third of the ESOs are exercised within the first year after vesting even though executives supposedly lose the most time value of money in so doing.
The results in Panel A could be subject to the same data issue we previously discussed. For instance, for the ESOs vested in 2005, they will be included in our full sample only if they are also exercised before the end of year 2005, which is within the one-year period after vesting. This issue could artificially inflate the proportions of exercise in the earlier years after vesting. To examine this potential issue, we partition a subsample that only includes observations with expiration dates in or before the year 2005, and we show the subsample results in Panel B. The results of Panel B, however, are generally consistent with Panel A. Though the proportions of exercises in years 2 and 3 decrease compared to the Panel A results, the proportion of exercises in the first year after the vesting date actually increases from 36.1% to 37.6%. These results give us greater confidence in the representativeness of our sample.
We next analyze the categorization of those first year exercises by the number of months after vesting in Panel C. The aforementioned data structure issue has little effect on the first year exercises, so we use 2,205 first-year exercises from the full sample. Among these 2,205 first-year exercises, 737 (33.4%) are exercised within two days after the vesting date, while the rest of the first month accounts for 273 (12.4%) exercises. Furthermore, the number of exercises in each of the first three months is higher than any of the remaining nine months. Taking these results together, Table III documents that early exercises are a common practice.
We then match these 6,109 exercises with ExecuComp, Compustat, and CRSP data based on the exercise date. For example, if an option exercise happens on May 20, 2005, then we locate compensation data and accounting data of the year 2004, and we also obtain daily stock price information relative to May 20, 2005. We are not able to locate data from ExecuComp for 282 observations because these 282 option exercises happen after executives leave their companies or step down from the top five highest paid positions. This procedure leaves us with 5,827 option exercises. We then exclude option exercises, which occur within 30 days of the option expiration date because they are clearly expiration related, and our final sample is 5,437 early option exercises. For some option exercises with multiple expiration dates, we take the earliest expiration date to determine whether it is an expiration-related exercise.
Descriptive statistics of the primary sample are shown in Table IV. To make salary and bonus, value of shares owned, value of vested options, value of unvested options, and previous-year dividends amount comparable over different years, we adjust them to year 2000 dollars. The executives in our sample on average earn around $1.707 million in salary and bonus a year, and the median salary and bonus is about $727,000. The value of option compensation is, on average, much larger. The mean (median) dollar value of vested options is $15.077 million ($2.274 million), and these numbers for unvested options are $7.744 million (1.258 million). (8) The mean amount of shares owned constitutes 0.43% of the company shares outstanding and the median is 0.054%. In dollar amounts, executives own $28.515 million ($2.164 million) of company stock in terms of means (medians). We find that the postexercise long-term performance is not as good as the preexercise long-term performance as the mean (median) change in the four-factor model intercept is -0.0006 (-0.0005). Moreover, the firms in the sample are fairly large with mean (median) total market cap of about $15.085 billion ($2.808 billion). Value of shares owned and equity market cap are apparently strongly right-skewed, as the mean of the former variable is more than three times as large as the third quartile number, while the mean of the latter is about 1.5 times as large as the third quartile number. In our multivariate tests in Section V, we take the natural logarithm of these two variables. For observations with zero dollar value of shares owned, we designate the dollar value to be $0.001.
III. Determinants of Vesting Exercise
A. Pearson Correlation Tables
Table V shows the Pearson correlation coefficients between vesting date exercise and all of the test variables. In this table, correlation coefficients are shown in the first row, and the p-values are reported below each correlation in the second row. In Row 3, we record the number of pairs used to calculate the correlations.
Most of the correlations between the executive wealth management variables and vesting date exercise are significant and consistent with the predictions of our hypotheses. Being a CEO is significantly negatively correlated with the propensity of vesting date exercise. This suggests that wealthier executives, who are less risk averse, are less likely to exercise their ESOs on the vesting date. Salary and bonus is negatively and January vesting is positively correlated with vesting date exercise. Taken together, these present conflicting signals on the importance of liquidity in the choice between a vesting exercise and a later exercise. However, salary and bonus may have overall wealth, and hence risk-aversion consequences, while the January dummy may have tax implications, which would tend to shade the signs in the direction we observe.
Our results also show that diversification and rebalancing needs are related to vesting date exercise. Consistent with our predictions, the value of unvested options, the recent new grant dummy, and the volatility of stock are significantly positively correlated with vesting exercise, suggesting that portfolio considerations are more important to vesting exercises than other early exercises. The value of shares owned and the value of vested options are negatively related to vesting exercise. Both of these variables may be proxies for wealth, and hence risk aversion, with wealthier executives being more willing to delay exercise beyond the vesting date. The latter variable may also reflect a first-in first-out exercise strategy on the part of executives. Percentage of shares owned is insignificant.
The correlations on the information variables present a much more muddled picture. The most significant correlations are the postexercise four-factor model intercept and the insider open market net purchase ratio, which are both positive. This is consistent with information being less important to vesting date exercises since they suggest vesting exercises are more likely to occur when there is good news. The change in the four-factor model intercept and the change in the two-week abnormal return are both significant at the 10% level, but have opposite signs. One-year buy-and-hold abnormal returns and postexercise two-week stock returns are insignificant. Taken as a whole, however, the observed patterns are consistent with the hypothesis that executive portfolio diversification is a primary motivation for exercise on the vesting date.
B. Multivariate Test Results
We next estimate the multivariate model specified in Equation (1) using the logit methodology. The percentage of shares owned and the value of shares owned are highly correlated as their correlation coefficient is 23.32%, which is significant at the 0.01% level. Hence, to reduce multicollinearity, we use only one of these two variables in Models 1, 2, 4, and 5. However, because the percentage of shares owned and the value of shares owned might have different implications, as the former is more control-right related and the latter is more wealth related, we also show the test results when both of these variables are included (Models 3, 6, and 7). We also specify alternate models using the four-factor model intercept and postexercise two-week abnormal returns (Models 1, 2, and 3) and the change in four-factor model intercept and change in two-week abnormal returns (Models 4, 5, and 6) around the exercise date. In Model 7, we have one-year buy-and-hold abnormal returns as a postexercise long-run return proxy. (9) The test results of these seven models are presented in Table VI.
Liquidity and risk aversion do not appear to be strong differential motivators of vesting exercise versus other early exercises. The CEO dummy, admittedly a noisy proxy for risk aversion, is insignificant as is salary and bonus. However, the January dummy is positive and significant. Since we expect executives to have greater liquidity in January, a positive sign indicates liquidity is relatively less important to a vesting date exercise (i.e., the executive is more likely to do a vesting exercises when cash flow is high). Since liquidity needs may be relatively time sensitive, we might expect liquidity to matter less to a vesting date exercise than to other early exercises. The positive sign is consistent with liquidity mattering less to a vesting date exercise, but could also reflect the tax considerations discussed in Footnote 3. Thus, there is only weak evidence that liquidity is less important to vesting exercises than other early exercises.
The diversification needs of the executive receive considerable support as motives for vesting exercises versus other early exercises. Although percentage of stock owned and the value of stock owned were insignificant across all specifications, the value of unvested options, the volatility of the firm's stock, and the recent new grant dummy were all positively significant at the 1% level. This means that large quantities of unvested options, a volatile stock, and a new grant around the exercise date are more likely to spur a vesting exercise than other early exercises. Thus, diversification needs disproportionately affect vesting exercises. The quantity of vested options is negatively significant at the 1% level, meaning a large quantity of vested options makes a vesting exercise less likely than other early exercises. This may reflect a first-in first-out option exercise strategy on the part of executives or may reflect wealth-related effects on risk aversion (i.e., DARA).
Private information, on the other hand, does not appear to be a significant motivator of vesting exercise versus other early exercises. The postexercise four-factor model intercept, the postexercise two-week abnormal return, the one-year buy-and-hold abnormal return, and the change in the four-factor model intercept are never significant in any of the specifications. This suggests that private information produces no differential effect between vesting exercises and other early exercises. The change in the two-week abnormal return is positive and significant at the 10% level in Models 4, 5, and 6, which is consistent with negative private information being less important to a vesting exercise than to other early exercises. Also, the coefficient on the insider open market purchase ratio is positive and significant across all specifications at the 1% level. Since a lower insider open market purchase ratio is consistent with negative private information, we would expect this variable and exercises to be negatively correlated, in general. Since this variable is positively correlated with the probability of a vesting exercise (i.e., a higher open market purchase ratio, which suggests positive information, is associated with a greater likelihood of a vesting exercise than other early exercises), it suggests negative private information is less significant to vesting exercises than to other early exercises. Since private information is time sensitive and vesting exercises are the solution to time-constrained optimizations, these results are intuitive.
A number of the control variables are also significant. The monthly risk-free rate and the number of days from exercise to expiration are negative and significant at the 1% level in all specifications, suggesting that executives do consider the time value of money in weighing a vesting exercise versus a later early exercise, just as one might expect. Also, executives from larger firms and utility firms are more likely to exercise on the vesting date, while executives from high-tech companies are less likely to exercise on the vesting date. These results are consistent with executives from companies with greater growth opportunities being more willing to retain options past the vesting date.
IV. Robustness Tests
A. Missing Observation Issue
Our primary sample size prior to matching with ExecuComp and CRSP is 6,109. However, we are only able to use around 5,040 observations in our primary multivariate tests. This significant decrease in the number of observations is primarily caused by the missing observations for the value of vested options and the value of unvested options variables. Those two variables have 772 missing observations and serve as the major data limiter for our multivariate regressions. These missing observations are a potential source of bias.
To address this issue, we omit these two variables from the model and rerun our multivariate regression. We report the regression results in Column 1 of Table VII. For the sake of brevity, in Table VII, we include both the percentage of shares owned and the dollar value of shares owned as proxies for share ownership in each of these regressions and include only the postexercise four-actor model intercept (the postexercise two-week abnormal returns) as a measure of long-run (short-run) performance. (10)
The results in Column 1 of Table VII support our major findings. The significant relations we document in Section III remain largely unchanged. The primary differences are that salary and bonus becomes negatively significant at the 5% level and value of shares owned becomes negatively significant at the 1% level. Since a higher salary and bonus implies fewer liquidity problems, and hence less need to exercise options for liquidity reasons, a negative coefficient here implies liquidity may be more important to vesting exercises than later early exercises. As is true in the univariate correlations, the salary and bonus variable and the January dummy present conflicting evidence on the role of liquidity, but the observed signs are consistent with a wealth/risk aversion interpretation for salary and bonus and a tax-based interpretation for the January dummy. The negative coefficient on the value of shares owned may imply that in the absence of the vested options variable, this variable proxies for some of the same factors as vested options and may imply greater wealth and hence less risk aversion, with lower risk aversion making vesting exercise less likely.
B. Company Size and Age Issue
As we discuss in Section III, our sample set is a merged product of the TFI database, the ExecuComp database, and the CRSP database. Though the inclusion of ExecuComp provides us valuable executive compensation and stock ownership data, it also decreases our sample size by nearly 75% because the coverage of ExecuComp is limited to five executives in each of the S&P 1500 firms. Moreover, for a firm to be included in the S&P 1500 index it has to satisfy size and age requirements. A $300 million market cap is the threshold to be an S&P SmallCap 600 firm and the firm has to be at least six months removed from its IPO. Hence, our findings reported in Section III could be merely the behavior of executives from larger older companies.
To test whether our findings are also representative of executives in smaller and younger firms, we expand our sample by only merging the TFI option exercise data and CRSP stock price information. In so doing, we are not able to locate executive compensation information for some of the wealth management variables, but the remaining variables still provide some controls to test the wealth management hypothesis, as well as the private information hypothesis. The CEO variable is still loosely related to wealth and, hence, to the degree of risk aversion; liquidity issues are reflected in January vesting; and whether an executive has another new option grant around the vesting date is related to diversification needs. The risk of company stock is still indicated by stock volatility. We then obtain a sample with 28,660 executive exercises by 12,953 executives from 3,174 companies, and 2,466 of these exercises are vesting exercises. (11) For this larger sample, the mean market cap is $13,700.11 million, somewhat smaller than the $15,085 million mean market cap of the primary sample. (12) However, the distributions of market cap in the primary versus this larger sample are rather different. The median market cap of the larger sample (primary sample) is $1,462.73 million ($2,808 million), and the 25th percentile market cap of the larger sample (primary sample) is $412.45 million ($1,126 million). These comparisons suggest that our larger sample includes a fair number of small firms and therefore expands the coverage of our study. For about 19.72% of the observations in our new sample, the firm market cap is less than $300 million. This makes them ineligible for inclusion in the S&P 600 SmallCap index and the ExecuComp database. The regression results using this larger sample are illustrated in Column 2 of Table VII.
The results in Column 2 of Table VII are consistent with our primary findings in many respects, but there are also significant differences. The recent new grant dummy and volatility of stock remain significantly positive consistent with diversification needs being more important to vesting exercises than other early exercises. The high-tech dummy and number of days from exercise to expiration remain significantly negative, although the significance of the former drops to the 10% level. The utility dummy remains significantly positive. The risk-free rate is no longer significant, however, and company size actually changes sign and becomes significantly negative. The latter result implies that executives in smaller firms are more likely to exercise on the vesting date than their larger firm counterparts. The CEO dummy and the January dummy also change sign and become negatively significant, at the 5% and 1% levels, respectively. The result on the CEO dummy likely flows from the fact that without information on stockholdings and vested option holdings, the CEO dummy may serve more strongly as a proxy for wealth and hence risk aversion, with wealthier less risk-averse CEOs less likely to exercise on the vesting date. The result on the January dummy suggests that when small-firm executives and non-top-five executives at larger firms are included in the sample, liquidity is more important to vesting date exercises than other early exercises. Taken together with the result on firm size, this suggests that the propensity of small-firm executives to exercise on the vesting date may be due to greater liquidity constraints.
The private information variables provide conflicting results here. The insider purchase ratio changes sign and is negatively significant at the 1% level, while the postexercise four-factor intercept and the postexercise two-week abnormal return become positively significant at the 5% and 1% levels, respectively. The insider purchase ratio result suggests that private information is more important to vesting exercises than later early exercises when smaller firms are included, since executives are exercising when they are selling other shares. However, the abnormal returns results are consistent with private information being less important to vesting exercises since executives are exercising on the vesting date when subsequent stock returns are better. If we were to assume that executives in small firms are more likely to exercise and hold rather than exercise and sell, these results would suggest that information is longer-lived and perhaps more valuable in smaller firms. (13) Such a conclusion would be intuitive since the absence of analyst coverage and lesser institutional interest in smaller firms may imply greater asymmetric information between firm insiders and other investors. It would also be consistent with the evidence of Carpenter and Remmers (2001) that exercises at small firms may be more information related. However, we cannot distinguish cases where executives exercise and hold versus exercise and sell and any such interpretation of the results is speculative.
C. Multiple-Vesting-Date Exercise Issue
As we previously discussed, we exclude option exercise observations with multiple vesting dates from our primary test sample to preserve a clear definition of a vesting date exercise. In reality, executives often exercise ESOs with different vesting dates on a single day. In this section, we include these multiple-vesting-date option exercises in the test sample and investigate whether the inclusion of those exercises with multiple vesting dates would change our results.
We obtain 54,459 option exercise observations containing both single-vesting-date exercises and multiple-vesting-date exercises from the TFI database. The sample size becomes 15,447 after the merge with the ExecuComp database. After excluding option exercises with vesting dates before 1996 and exercises within 30 days of expiration dates, our sample is 12,975 ESO exercises by 4,059 executives from 1,121 companies.
For those exercises with multiple vesting dates, we define whether they are vesting exercises by three methods, including two that represent alternate extremes (the loosely defined method and the strictly defined method). In the loosely defined method, we take a multiple-vesting-date option exercise as a vesting exercise as long as some portion of the exercised options is newly vested. In the strictly defined method, a multiple-vesting-date options exercise is regarded as a vesting exercise only if all of the exercised options are newly vested. In addition, we also create a variable Percentage of vesting exercise, which is the percentage of the exercised options that are newly vested for that particular multiple-vesting-date exercise. These methods &defining vesting exercise for multiple-vesting-date option exercises do not affect our original definition of vesting exercise for single-vesting-date option exercises. For option exercises with a single vesting date, their Percentage of vesting exercise is either one or zero. Vesting exercises consist of 7.86% of our sample if the loosely defined method is adopted. This number decreases to 4.96% if we use the strictly defined method. The average Percentage of vesting exercise is 6.22%. Summary statistics of other test variables and control variables for this expanded sample with multiple-vesting-date exercises are similar to those for the primary sample as shown in Table IV. (14)
We rerun our test model with the expanded sample and the results appear in Table VIII. The dependent variable in Column 1 is the loosely defined vesting exercise and the dependent variable in Column 2 is the strictly defined vesting exercise. Due to the binary nature of these dependent variables, we apply the logit methodology for Columns 1 and 2. The dependent variable is Percentage of vesting exercise in both Column 3 and 4. The Tobit methodology is used for Column 4 since the dependent variable "percentage of vesting exercise" is censored at zero and one. We present OLS results in Column 3 for comparison.
Out test results in Table VIII are generally consistent with the regression results based on the primary sample. The value of unvested options, volatility of stock, and new option grant dummy remain significantly positive at the 1% level in all specifications suggesting that diversification remains a differentially significant motivation of vesting date exercises. The value of vested options, the number of days from exercise to expiration, and the high-tech dummy remain significantly negative at the 1% level. The January vesting and utility firm dummies remain significantly positive at the 1% level. The insider open market purchase ratio remains significantly positive, although its significance level drops from 1% to 5% in Model 1. The differences are that company size becomes insignificant in some specifications, the CEO dummy becomes positively significant in some specifications, the percentage of shares owned becomes negatively significant in some specifications, and the postexercise four-factor intercept becomes positively significant at the 10% level in one specification. Since the primary difference between this sample and the primary sample is that we added exercises with multiple vesting dates, the significance and sign of the CEO dummy across specifications is difficult to interpret. In the loosely defined (Column 1) and percentage (Columns 3 and 4) cases it suggests that the positive coefficient results because CEOs are more likely to do multiple-vesting-date exercises and adding these cases as vesting exercises generates the positive coefficient. However, in the strictly defined case (Column 2), a positive coefficient would result because non-CEOs are more likely to do multiple-vesting-date exercises, leaving CEOs more likely to do pure vesting date exercises. We note, however, the OLS specification in Column 3 is insignificant and the strictly defined case (Column 2) and the percentage case estimated using Tobit (Column 4) are significant at only the 10% level. The percentage of shares owned has some of the same implications as vested options and its role in these regressions is similar. On the whole though, there is nothing that disturbs our original conclusions that vesting exercises are disproportionately driven by portfolio diversification considerations and private information is relatively less important in triggering vesting exercises.
In summary, we find that our primary findings are generally not sensitive to missing observation issues, large/old firm bias, or the exclusion of exercises with multiple vesting dates. Our robustness tests in general strengthen our confidence in the primary results of the study. The primary difference is that private information may be relatively more important to vesting exercises in smaller firms, although only if we are willing to assume that an exercise-and-hold strategy is frequently used by small-firm executives.
We show that early exercises of ESOs are a common practice in the business world. This study examines the motivations for vesting date exercise of ESOs versus other early exercises. We hypothesize that vesting date exercises are more strongly related to personal wealth management concerns of executives than other early exercises and less strongly related to negative private information regarding the firm's future prospects than other early exercises. We conduct both correlation analysis and multivariate tests to study these hypotheses. Our results indicate that executives exercise their ESOs on the vesting date in order to manage their diversification needs, including underlying stock risk. Furthermore, vesting exercises are also less likely to be triggered by negative private information regarding long-run stock performance. All these findings appear generally robust to sample construction, methodology, and variable definitions.
One of the practical implications of our results is that the success of a firm's attempt to align the interests of shareholders and managers through its compensation structure will be affected by the personal wealth management considerations of those executives. Executives with greater diversification concerns who face greater risk in holding their companies' stocks are more prone to undo the incentive effects of option compensation at the earliest possible moment by exercising their ESOs on the vesting date.
Aboody, D., J. Hughes, J. Liu, and W. Su, 2008, "Are Executive Stock Option Exercises Driven by Private Information?" Review of Accounting Studies 13, 551-570.
Becket, B., 2006, "Wealth and Executive Compensation," Journal of Finance 61, 379-397.
Brooks, R., D. Chance, and B. Cline, 2009, "Private Information and the Exercise of Executive Stock Options," Clemson University Working Paper.
Carhart, M., 1997, "On Persistence in Mutual Fund Performance," Journal of Finance 52, 57-82.
Carpenter, J. and B. Returners, 2001, "Executive Stock Option Exercises and Inside Information," Journal of Business 74, 513-534.
Carr, P. and V. Linetsky, 2000, "The Valuation of Executive Stock Options in an Intensity-Based Framework," European Finance Review 4, 211-230.
Cvitanic, J., Z. Wiener, and F. Zapatero, 2008, "Analytic Pricing of Employee Stock Options," Review of Financial Studies 21, 683-724.
Fama, E. and K. French, 1993, "Common Risk Factors in the Returns on Stocks and Bonds," Journal of Financial Economics 33, 3-56.
Finnerty, J.D., 2005, "Extending the Black-Scholes-Merton Model to Value Employee Stock Options," Journal of Applied Finance 15, 25-54.
Huddart, S., 1994, "Employee Stock Options," Journal of Accounting and Economics 18, 207-231.
Huddart, S. and M. Lang, 2003, "Information Distribution within Firms: Evidence from Stock Option Exercises," Journal of Accounting and Economics 34, 3-31.
Jeng, L., A. Metrick, and R. Zeckhauser, 2003, "Estimating the Returns to Insider Trading: A Performance Evaluation Perspective," Review of Economics and Statistics 85, 453-471.
Kulatilaka, N. and A.J. Marcus, 1994, "Valuing Employee Stock Options," Financial Analysts Journal 50, 46-56.
Lakonishok, J. and I. Lee, 2001, "Are Insiders' Trades Informative?" Review of Financial Studies 14, 79-111.
Ligon, J.A., 1997, "A Simultaneous Test of Competing Theories Regarding the January Effect," Journal of Financial Research 20, 13-32.
Ligon, J.A. and P.D. Thistle, 2008, "Moral Hazard and Background Risk in Competitive Insurance Markets," Economica 75,700-709.
Liu, C.H. and D. Yermack, 2007, "Where Are the Shareholders' Mansions? CEOs Home Purchases, Stock Sales, and Subsequent Company Performance," New York University Working Paper.
Loughran, T. and J. Ritter, 2004, "Why Has IPO Underpricing Changed Over Time?" Financial Management 33, 5-37.
Meulbroek, L., 2001, "The Efficiency of Equity-Linked Compensation: Understanding the Full Cost of Awarding Executive Stock Options," Financial Management 30, 5-44.
Ofek, E. and D. Yermack, 2000, "Taking Stock: Equity-Based Compensation and the Evolution of Managerial Ownership," Journal of Finance 55, 1367-1384.
Ogden, J., 1990, "Turn-of-Month Evaluations of Liquid Profits and Stock Returns: A Common Explanation for the Monthly and January Effects," Journal of Finance 45, 1259-1272.
Perry, T. and M. Zenner, 2001, "Pay for Performance? Government Regulation and the Structure of Compensation Contracts," Journal of Financial Economics 62, 453-488.
Robicheaux, S.H., X. Fu, and J.A. Ligon, 2008, "Lease Financing and Corporate Governance," Financial Review 43, 403-437.
(1) We also run the model on all option exercises (i.e., including expiration date exercises) and on all option exercises prior to three months before expiration. The results, available upon request, are qualitatively similar to those presented here. Thus, our results are not sensitive to the definition of an expiration-related exercise.
(2) Note, for example, the importance of liquidity to the January effect in articles by Ogden (1990) and Ligon (1997).
(3) Taxes introduce a potential complexity to the interpretation of the January dummy. For executives who pay estimated tax (and most do) the optimal date for option exercise is September 16 since the estimated tax payment associated with the exercise can be delayed until January 15 of the following year. However, the second best month for exercise is January, since estimated tax payments on January exercises may be delayed until April 15. Taxes thus introduce a potential positive bias into the coefficient. Thus, while a positive coefficient would suggest liquidity is less important for vesting date exercises, it may also reflect the tax effect.
(4) Specifically, we employ calendar-time factor models to generate the abnormal long-term return over a 365-calendar-day period (the actual number of trading days varies from 250 to 252) after exercise as one of our proxies for long-term return. We obtain the performance of the underlying stock relative to the Fama and French (1993)-Carhart (1997) four-factor model. The four factors are:
MKTRF: excess return on the market, measured as the value-weighted market return minus risk-free rate,
HML: average return on the two highest value (high book-to-market) portfolios minus the average return on the two highest growth (low book-to-market) portfolios,
SMB: average return on the three smallest capitalization portfolios minus the average return on the three biggest capitalization portfolios, and
UMD: return on winner minus loser portfolio, where winners and losers are measured by returns from month -12 to -2.
For the 365 calendar days after the exercise date, observed daily excess returns of one underlying stock are regressed on these four factors:
[R.sub.t] - [R.sub.Ft] = [alpha] + [[beta].sub.1] [IMKTRFs.bu.t] + [[beta].sub.2][HML.sub.t] + [[beta].sub.3][SMB.sub.t] + [[beta].sub.4][UMD.sub.t] + [epsilon],
where [R.sub.t] is the daily stock return, RF, is the daily risk-free rate, and [R.sub.t] - [RF.sub.t] is the excess return. Alpha represents the difference between predicted daily returns by the four factors and actual daily returns, so alpha is viewed as the long-run abnormal return.
(5) See Appendix D of Loughran and Ritter (2004) for the definition of a high-tech industry.
(6) We exclude option exercises with multiple vesting dates from our primary analysis because we want to have a clear definition of vesting date exercise. Thus, we restrict our primary analysis to pure cases of vesting date exercise. We do consider multiple vesting date exercises in our robustness tests and the results are qualitatively similar to the primary analysis.
(7) 2,484/365 = 6.81 years, 2,559 / 365 = 7.01 years.
(8) The values of vested (unvested) options are provided by the ExecuComp database. In the ExecuComp database, the values of vested (unvested) options represent the value the executive would have realized at year end if he had exercised all of his vested (unvested) options that had an exercise price below the market price. In other words, these figures represent the intrinsic value of in-the-money options.
(9) The correlation coefficient between one-year buy-and-hold abnormal returns and the four-factor model intercept is 54.76%.
(10) Our robustness test results are quantitatively unchanged if we use buy-and-hold abnormal returns instead. These test results are available upon request.
(11) Again, all these 28,660 executive exercises are in the money on the vesting date.
(12) The mean market cap does not drop as much as one might expect because in addition to picking up executives at smaller firms we also add executives beyond the top five executives at larger firms.
(13) Aboody et al. (2008) present evidence that executives exercise and retain shares in over 25% of exercises, more frequently than the evidence of Offek and Yermack (2000) would suggest, and that exercise and hold predicts positive returns.
(14) Results are available upon request.
Xudong Fu and James A. Ligon *
We thank Anup Agrawal, Brandon Cline, Doug Cook. Junsoo Lee, Gregory Nagel, Mary Stone, Pengcheng Zhu. and an anonymous referee for comments on prior drafts. We also thank Ron Masulis, Harold Mulherin, Vikram Nanda, and David Yermack for helpful conversations. Remaining errors are our own.
* Xudong Fu is an Assistant Professor of Finance at Southern Illinois University, Edwardsville in Edwardsville, IL. James A. Ligon is a Professor of Finance at the University of Alabama in Tuscaloosa, AL.
Table I. Test Variable Descriptions and Predictions This table provides a listing and definitions of the test variables used in the analysis including those related to executive wealth management factors including the degree of executive risk aversion, the executive's liquidity needs, and the executive's diversification needs and private information factors related to long-term performance, short-term performance, and insider open market transitions. The predicted signs of the test variables are indicated. The table also includes the definitions of the control variables included in the analysis. Variable Definition CEO Dummy variable equal to one if the executive is CEO Salary and bonus Salary + bonus (in thousands) January vesting Dummy variable equal to one if vesting date occurs in January Percentage of shares Percentage of shares owned owned Value of shares owned log (dollar value of shares owned in thousands) Value of vested Dollar value of vested options options (in thousands) Value of unvested Dollar value of unvested options options (in thousands) Recent new grant Dummy variable equal to one dummy if the executive obtains another option grant in a (-30, +30)-day period relative to exercise date Volatility of stock The standard deviation of daily stock return for the half year preceding the exercise date Postexercise Fama-French-Carhart four-factor model four-factor model intercept intercept one-year postexercise. See Footnote 2 for a detailed explanation Change in four-factor Postexercise four-factor model model intercept intercept--pre-exercise four-factor model intercept One-year One-year buy-and-hold stock buy-and-hold returns minus market index abnormal returns Postexercise Arithmetic mean of two-week stock postexercise two-week daily returns stock returns adjusted by market index Change in two-week Postexercise two-week stock abnormal returns returns--preexercise two-week stock returns Insider open market (Number of insider purchase transaction net --number of insider purchase ratio sales)/(number of insider purchase + number of insider sales) for the open market transactions one month before the exercise date Control Variables Number of days from Expiration date--exercise exercise to date expiration Previous-year Total dividends per share in dividends amount the year before exercise date Monthly risk-free rate One-month Treasury bill rate on the exercise date Company size ($M) Company equity market cap in 2000 dollars High-tech company A dummy variable equal to one if it is high-tech firm, zero otherwise Utility firm A dummy variable equal to one if it is utility firm, zero otherwise Financial firm A dummy variable equal to one if it is financial firm, zero otherwise Variable Predictions Executive Wealth Management Factors Degree Liquidity Diversification of Risk Needs Needs Aversion CEO - Salary and bonus - ? January vesting ? Percentage of shares - + owned Value of shares owned - + Value of vested - + options Value of unvested + options Recent new grant + dummy Volatility of stock + Postexercise four-factor model intercept Change in four-factor model intercept One-year buy-and-hold abnormal returns Postexercise two-week stock returns Change in two-week abnormal returns Insider open market transaction net purchase ratio Control Variables Number of days from exercise to expiration Previous-year dividends amount Monthly risk-free rate Company size ($M) High-tech company Utility firm Financial firm Variable Predictions Private Information Factors Long-Term Short-Term Insider Performance Performance Open Market Transaction CEO Salary and bonus January vesting Percentage of shares owned Value of shares owned Value of vested options Value of unvested options Recent new grant dummy Volatility of stock Postexercise + four-factor model intercept Change in four-factor + model intercept One-year + buy-and-hold abnormal returns Postexercise + two-week stock returns Change in two-week + abnormal returns Insider open market transaction net purchase ratio Control Variables Number of days from exercise to expiration Previous-year dividends amount Monthly risk-free rate Company size ($M) High-tech company Utility firm Financial firm Table II. Vesting Exercise by Year This table presents the number of vesting and nonvesting option exercises by year. The data come from the Thompson Financial Network Insider Filing Data database, which contains information compiled from Forms 3, 4, 5, and 144 filed with the SEC, and cover the period from 1996 to 2005. Vesting No. of No. of No. of Percentage of Year Total Vesting Nonvesting Vesting Exercises Exercises Exercises Exercises 1996 577 29 548 5.03% 1997 780 124 656 15.90% 1998 515 35 480 6.80% 1999 878 99 779 11.28% 2000 689 92 597 13.35% 2001 916 72 844 7.86% 2002 550 45 505 8.18% 2003 513 91 422 17.74% 2004 480 81 399 16.88% 2005 211 69 142 32.70% 1996-2005 6,109 737 5,372 12.06% Table III. Categorization of Exercise by Year/Month This table presents the categorization of exercises by year and month. The data come from the Thompson Financial Network Insider Filing Data database, which contains information compiled from Forms 3, 4, 5, and 144 filed with the SEC, and cover the period from 1996 to 2005. The results of Panel A and Panel C are based on the full sample, and the results of Panel B are based on the subsample that only includes observations with expiration dates prior to the end of year 2005. Panel A. Categorization of Exercise by Number of Years after Vesting (Full Sample) Year 1 Year 2 Year 3 Year 4 Year 5 Year 6 Year 7 N 2,205 889 816 634 535 319 238 % 36.1% 14.6% 13.4% 10.4% 8.8% 5.2% 3.9% Year 8 Year 9 Year 10 Sum N 220 245 8 6,109 % 3.6% 4.0% 0.1 Panel B. Categorization of Exercise by Number of Years after Vesting (Partitioned Sample) Year 1 Year 2 Year 3 Year 4 Year 5 Year 6 Year 7 N 822 206 224 233 238 131 91 % 37.6% 9.4% 10.3% 10.7% 10.9% 6.0% 4.2% Year 8 Year 9 Year 10 Sum N 70 167 3 2,185 % 3.2% 7.6% 0.1 Panel C. Categorization of First-Year Exercise by Number of Months after Vesting (Full Sample) Day 0 to The Rest Month Month Month Month Month Day 2 of the 2 3 4 5 6 Month 1 N 737 273 194 139 113 100 96 % 33.4% 12.4% 8.8% 6.3% 5.1% 4.5% 4.4% Month Month Month Month Month Month Sum 7 8 9 10 11 12 N 116 94 86 88 82 87 2,205 % 5.3% 4.3% 3.9% 4.0% 3.7% 3.9% Panel D. Statistics of (Expiration Dory--Vesting Day) Mean Q1 Median Q3 (Expiration Day--Vesting Day) 2,484 1,827 2,559 3,287 Table IV. Descriptive Statistics Table IV presents descriptive statistics for the merged sample. Option exercise and insider trading data come from the Thompson Financial Network Insider Filing Data database, which contains information compiled from Forms 3, 4, 5, and 144 filed with the SEC, and cover the period from 1996 to 2005. Executive and compensation information is from ExecuComp. Stock market information is from the CRSP database. Salary and bonus, value of shares owned, value of vested options, value of unvested options, and previous year dividends amount are inflation adjusted to year 2000 dollars. N stands for the number of nonmissing observations. See Table I for detailed variable explanations. Mean SD Dependent Variable Vesting exercise 0.1148 0.3188 Test Variable CEO 0.2546 0.4356 Salary and bonus ($T) 1,707.65 3,186.50 January vesting 0.0668 0.2496 Percentage of shares 0.4274 1.7002 owned (%) Value of shares owned 28,515.21 2,46,593.66 ($T) Value of vested options 15,077.43 41,589.38 ($T) Value of unvested 7,744.45 28,250.68 options ($T) Recent new grant 0.2200 0.4143 Volatility of stock 0.0248 0.0128 Postexercise four-factor 0.0003 0.0019 model intercept Change in four-factor -0.0006 0.0022 model intercept One-year buy-and-hold -0.0679 0.4995 abnormal returns Postexercise two-week 0.0005 0.0080 stock returns Change in two-week -0.0024 0.0116 abnormal returns Insider open market -0.6500 0.5216 transaction net purchase ratio Control Variables Number of days from 915.81 893.00 exercise to expiration Previous-year 0.4114 1.2092 dividends amount Monthly risk-free rate 0.227 0.145 Company size ($M) 15,085.78 41,132.95 High-tech company 0.1402 0.3472 Utility firm 0.0259 0.1590 Financial firm 0.2218 0.4155 Q1 Median Dependent Variable Vesting exercise 0 0 Test Variable CEO 0 0 Salary and bonus ($T) 424.69 726.71 January vesting 0 0 Percentage of shares 0.0091 0.0542 owned (%) Value of shares owned 317.32 2,164.82 ($T) Value of vested options 474.53 2,274.57 ($T) Value of unvested 301.53 1,258.40 options ($T) Recent new grant 0 0 Volatility of stock 0.0162 0.0218 Postexercise four-factor -0.0005 0.0001 model intercept Change in four-factor -0.0016 -0.0005 model intercept One-year buy-and-hold -0.3390 -0.1134 abnormal returns Postexercise two-week -0.0034 0.0004 stock returns Change in two-week -0.0079 -0.0021 abnormal returns Insider open market -1 -1 transaction net purchase ratio Control Variables Number of days from 138 678 exercise to expiration Previous-year 0 0 dividends amount Monthly risk-free rate 0.090 0.160 Company size ($M) 1,126.44 2,808.19 High-tech company 0 0 Utility firm 0 0 Financial firm 0 0 Q3 N Dependent Variable Vesting exercise 0 5,437 Test Variable CEO 1 5,437 Salary and bonus ($T) 1,401.48 5,437 January vesting 0 5,437 Percentage of shares 0.2200 5,435 owned (%) Value of shares owned 9,765.79 5,433 ($T) Value of vested options 9,837.97 5,048 ($T) Value of unvested 5,082.02 5,048 options ($T) Recent new grant 0 5,437 Volatility of stock 0.0291 5,432 Postexercise four-factor 0.0009 5,434 model intercept Change in four-factor 0.0004 5,432 model intercept One-year buy-and-hold 0.0959 5,434 abnormal returns Postexercise two-week 0.0043 5,429 stock returns Change in two-week 0.0031 5,428 abnormal returns Insider open market 0 5,437 transaction net purchase ratio Control Variables Number of days from 1,432 5,437 exercise to expiration Previous-year 0.5815 5,437 dividends amount Monthly risk-free rate 0.360 5,437 Company size ($M) 10,541.02 5,434 High-tech company 0 5,437 Utility firm 0 5,437 Financial firm 0 5,437 Table V. Pearson Correlation Coefficients between Vesting Exercise and Test Variables Correlation coefficients are shown in the first row, and the p-values are reported below each correlation in the second row. In the third row, we record the number of pairs used to calculate correlations. Option exercise and insider trading data come from the Thompson Financial Network Insider Filing Data database, which contains information compiled from Forms 3, 4, 5, and 144 filed with the SEC, and cover the period from 1996 to 2005. Executive and compensation information is from ExecuComp. Stock market information is from the CRSP database. Salary and bonus, value of shares owned, value of vested options, value of unvested options, and previous-year dividends amount are inflation adjusted to year 2000 dollars. See Table I for detailed variable explanations. CEO Salary and January Percentage of Bonus Vesting Shares Owned Vesting -0.0488 -0.0609 0.0586 0.0050 exercise (0.0003) (0.0000) (0.0000) (0.7133) N = 5437 N = 5437 N = 5437 N = 5435 Value of Value of Value of Recent Volatility Shares Vested Unvested New of Stock Owned Options Options Grant Dummy Vesting -0.0918 -0.0833 0.0848 0.1236 0.1412 exercise (0.0000) (0.0000) (0.0000) (0.0000) (0.0000) N = 5433 N = 5048 N = 5048 N = 5437 N = 5,432 Post-Exercise Change One-Year Post-exercise Four-Factor in Four- Buy-and-Hold Two-Week Stock Model Factor Abnormal Returns Intercept Model Returns Intercept Vesting 0.0473 -0.0230 0.0167 0.0131 exercise (0.0005) (0.0907) (0.2185) (0.3349) N = 5,434 N = 5,432 N = 5,434 N = 5,429 Change in Insider Open Two-Week Market Abnormal Transaction Returns Net Purchase Ratio Vesting 0.0244 0.1332 exercise (0.0725) (0.0000) N = 5,428 N = 5,437 Table VI. Motives for Vesting Exercise This table shows logit model results of the propensity that executives exercise in-the-money executive stock options within two days of the vesting date. The dependent variable, vesting exercise, is a binary variable that equals one if the option is exercised within two days of the vesting date and zero if it is exercised at any other point prior to one month prior to expiration. Option exercise and insider trading data come from the Thompson Financial Network Insider Filing Data database, which contains information compiled from Forms 3, 4, 5, and 144 filed with the SEC, and cover the period from 1996 to 2005. Executive and compensation information is from ExecuComp. Stock market information is from the CRSP database. Salary and bonus, value of shares owned, value of vested options, value of unvested options, and previous-year dividends amount are inflation adjusted to year 2000 dollars. See Table I for detailed variable explanations. Model 1 Coef. Z-Stat CEO 0.0605 0.45 Salary and bonus 0.0000 -1.00 January vesting 0.6716 *** 3.85 Percentage of shares 0.0026 0.09 owned Value of shares owned Value of vested -0.00003 *** -5.29 options Value of unvested 0.00001 *** 5.83 options Recent new grant 0.5850 *** 4.89 dummy Volatility of stock 31.8802 *** 6.82 Postexercise 11.2655 0.50 four-factor model intercept Postexercise 4.3539 0.74 two-week abnormal returns One-year buy-hold abnormal returns Change in four-factor model intercept Change in two-week abnormal returns Insider open 0.3136 3.33 market net purchase ratio Monthly risk-free -288.84 *** -2.77 rate No. of days from -0.0005 *** -9.02 exercise to expiration Previous-year -0.0574 -0.77 dividends amount Company size 0.0822 ** 2.04 High-tech -0.5794 *** -3.41 company Utility firm 0.7602 ** 2.50 Financial firm 0.1879 1.24 Year dummies Yes Number of 5,040 observations used Pseudo [R.sup.2] 0.2169 Model 2 Coef. Z-Stat CEO 0.0802 0.59 Salary and bonus 0.0000 -0.96 January vesting 0.6732 *** 3.86 Percentage of shares owned Value of shares -0.0093 -0.66 owned Value of vested -0.00003 *** -5.28 options Value of unvested 0.00001 *** 5.84 options Recent new grant 0.5869 *** 4.91 dummy Volatility of stock 31.8666 *** 6.81 Postexercise 11.9528 0.53 four-factor model intercept Postexercise 4.3847 0.75 two-week abnormal returns One-year buy-hold abnormal returns Change in four-factor model intercept Change in two-week abnormal returns Insider open 0.3150 *** 3.35 market net purchase ratio Monthly risk-free -288.55 *** -2.77 rate No. of days from -0.0005 *** -9.11 exercise to expiration Previous-year -0.0581 -0.78 dividends amount Company size 0.0878 ** 2.15 High-tech -0.5867 *** -3.44 company Utility firm 0.7566 ** 2.49 Financial firm 0.1960 1.28 Year dummies Yes Number of 5,040 observations used Pseudo [R.sup.2] 0.2170 Model 3 Coef. Z-Stat CEO 0.0785 0.58 Salary and bonus 0.0000 -0.98 January vesting 0.6731 *** 3.86 Percentage of shares 0.0073 0.25 owned Value of shares -0.0101 -0.70 owned Value of vested -0.00003 *** -5.27 options Value of unvested 0.00001 *** 5.82 options Recent new grant 0.5876 *** 4.91 dummy Volatility of stock 31.8586 *** 6.81 Postexercise 11.9120 0.53 four-factor model intercept Postexercise 4.3240 0.74 two-week abnormal returns One-year buy-hold abnormal returns Change in four-factor model intercept Change in two-week abnormal returns Insider open 0.3148 *** 3.34 market net purchase ratio Monthly risk-free -288.89 *** -2.77 rate No. of days from -0.0005 *** -9.04 exercise to expiration Previous-year -0.11578 -0.77 dividends amount Company size 0.0899 ** 2.15 High-tech -0.5856 *** -3.44 company Utility firm 0.7571 ** 2.49 Financial firm 0.1970 1.29 Year dummies Yes Number of 5,040 observations used Pseudo [R.sup.2] 0.217 Model 4 Coef. Z-Stat CEO 0.0566 0.42 Salary and bonus 0.0000 -1.06 January vesting 0.6470 *** 3.70 Percentage of shares 0.0026 0.09 owned Value of shares owned Value of vested -0.00003 *** -5.29 options Value of unvested 0.00001 *** 5.81 options Recent new grant 0.5917 *** 4.94 dummy Volatility of stock 32.1023 *** 6.96 Postexercise four-factor model intercept Postexercise two-week abnormal returns One-year buy-hold abnormal returns Change in -17.73 -0.90 four-factor model intercept Change in two-week 6.4399 * 1.66 abnormal returns Insider open 0.3156 *** 3.31 market net purchase ratio Monthly risk-free -297.15 *** -2.85 rate No. of days from -0.0005 *** -8.99 exercise to expiration Previous-year -0.0606 -0.80 dividends amount Company size 0.0845 ** 2.10 High-tech -0.5947 *** -3.51 company Utility firm 0.7312 ** 2.41 Financial firm 0.1863 1.22 Year dummies Yes Number of 5,040 observations used Pseudo [R.sup.2] 0.2176 Model 5 Coef. Z-Stat CEO 0.0781 0.58 Salary and bonus 0.0000 -1.02 January vesting 0.6483 *** 3.70 Percentage of shares owned Value of shares -0.0102 -0.73 owned Value of vested -0.00003 *** -5.28 options Value of unvested 0.00001 *** 5.82 options Recent new grant 0.5940 *** 4.96 dummy Volatility of stock 32.1336 *** 6.96 Postexercise four-factor model intercept Postexercise two-week abnormal returns One-year buy-hold abnormal returns Change in -17.62 -0.90 four-factor model intercept Change in two-week 6.6300 * 1.70 abnormal returns Insider open 0.3169 *** 3.33 market net purchase ratio Monthly risk-free -296.86 *** -2.84 rate No. of days from -0.0005 *** -9.09 exercise to expiration Previous-year -0.0615 -0.81 dividends amount Company size 0.0909 ** 2.22 High-tech -0.6031 *** -3.95 company Utility firm 0.7270 ** 2.39 Financial firm 0.1950 1.28 Year dummies Yes Number of 5,040 observations used Pseudo [R.sup.2] 0.2177 Model 6 Coef. Z-Stat CEO 0.0764 0.56 Salary and bonus 0.0000 -1.03 January vesting 0.6483 *** 3.70 Percentage of shares 0.0078 0.26 owned Value of shares -0.0111 -0.77 owned Value of vested -0.00003 *** -5.28 options Value of unvested 0.00001 *** 5.80 options Recent new grant 0.5947 *** 4.96 dummy Volatility of stock 32.1236 *** 6.96 Postexercise four-factor model intercept Postexercise two-week abnormal returns One-year buy-hold abnormal returns Change in -17.67 -0.90 four-factor model intercept Change in two-week 6.6107 * 1.70 abnormal returns Insider open 0.3166 *** 3.32 market net purchase ratio Monthly risk-free -297.17 *** -2.85 rate No. of days from -0.01105 *** -9.02 exercise to expiration Previous-year -0.0611 -0.81 dividends amount Company size 0.0931 ** 2.23 High-tech -0.6018 *** -3.54 company Utility firm 0.7277 ** 2.39 Financial firm 0.1961 1.28 Year dummies Yes Number of 5,040 observations used Pseudo [R.sup.2] 0.2177 Model 7 Coef. Z-Stat CEO 0.0828 0.61 Salary and bonus 0.0000 -1.01 January vesting 0.6668 *** 3.82 Percentage of shares 0.0080 0.27 owned Value of shares -0.0106 -0.73 owned Value of vested -0.00003 *** -5.23 options Value of unvested 0.00001 *** 5.80 options Recent new grant 0.5886 *** 4.92 dummy Volatility of stock 32.5349 *** 7.09 Postexercise four-factor model intercept Postexercise 3.5009 0.60 two-week abnormal returns One-year buy-hold 0.1095 1.20 abnormal returns Change in four-factor model intercept Change in two-week abnormal returns Insider open 0.3191 *** 3.39 market net purchase ratio Monthly risk-free -287.16 *** -2.76 rate No. of days from -0.0005 *** -9.05 exercise to expiration Previous-year -0.0526 -0.72 dividends amount Company size 0.0933 * 2.24 High-tech -0.5743 *** -3.37 company Utility firm 0.7628 ** 2.91 Financial firm 0.1990 1.30 Year dummies Yes Number of 5,040 observations used Pseudo [R.sup.2] 0.2174 *** Significant at the 0.01 level. ** Significant at the 0.05 level. * Significant at the 0.10 level. Table VII. Robustness Tests This table shows logit model results of the propensity that executives exercise in-the-money executive stock options within two days of the vesting date. In Column 1, we exclude the vested and unvested options variables. In Column 2, we use a new and larger sample without ExecuComp variables. The dependent variable is a binary variable that equals one if the option is exercised within two days of the vesting date and zero if it is exercised at any other point prior to one month before expiration. Option exercise and insider trading data come from the Thompson Financial Network Insider Filing Data database, which contains information compiled from Forms 3, 4, 5, and 144 filed with the SEC, and cover the period from 1996 to 2005. Executive and compensation information is from ExecuComp. Stock market information is from the CRSP database. Salary and bonus, value of shares owned, and previous-year dividends amount are inflation adjusted to year 2000 dollars. See Table I for detailed variable explanations. Column 1 Coef. Z-Stat CEO -0.0059 -0.05 Salary and bonus -0.0001 ** -2.07 January vesting 0.6439 *** 3.90 Percentage of shares owned 0.0218 0.82 Value of shares owned -0.0325 *** -3.74 Recent new grant dummy 0.7560 *** 7.00 Volatility of stock 29.4098 *** 6.81 Postexercise four-factor model intercept 21.6962 1.00 Postexercise two-week abnormal returns 3.6939 0.69 Insider open market transaction net 0.3600 *** 4.13 purchase ratio Monthly risk-free rate -235.42 ** -2.45 Number of days from exercise to expiration -0.0005 *** -9.21 Previous-year dividends amount -0.0751 -1.03 Company size 0.0902 *** 2.67 High-tech company -0.6451 *** 4.10 Utility firm 0.7706 *** 2.76 Financial firm 0.2251 1.61 Year dummies Yes Number of observations used 5,812 Pseudo [R.sup.2] 0.2306 Column 2 Coef. Z-Stat CEO -0.1636 ** -2.17 Salary and bonus January vesting -0.3344 *** -4.31 Percentage of shares owned Value of shares owned Recent new grant dummy 0.7872 *** 14.11 Volatility of stock 18.5726 *** 12.09 Postexercise four-factor model intercept 17.7696 ** 2.16 Postexercise two-week abnormal returns 7.3551 *** 3.36 Insider open market transaction net -0.8710 *** -18.68 purchase ratio Monthly risk-free rate 16.6754 0.33 Number of days from exercise to expiration -0.0006 *** -22.38 Previous-year dividends amount -0.0185 -0.83 Company size -0.0271 ** -2.12 High-tech company -0.1363 * -1.92 Utility firm 0.3388 ** 1.98 Financial firm 0.3500 *** 5.74 Year dummies Yes Number of observations used 27,095 Pseudo [R.sup.2] 0.1400 *** Significant at the 0.01 level. ** Significant at the 0.05 level. * Significant at the 0.10 level. Table VIII. Robustness Tests on Expanded Sample with Multiple Vesting Exercises This table shows regression results when vesting exercise decisions are regressed on the test variables and other control variables. In Column 1, an option exercise is regarded as a vesting exercise as long as some portion of the exercised options is newly vested. In Column 2, an option exercise is regarded as a vesting exercise only if all portions of the exercised options are newly vested. The dependent variables for both Columns 1 and 2 are binary variables equal to one if the observation is defined as a vesting exercise and zero for any other early exercise and are estimated using logit. In Columns 3 and 4, the dependent variables are the percentage of the exercised options that are newly vested. We use OLS for Column 3 and the Tobit methodology for Column 4. Option exercise and insider trading data come from the Thompson Financial Network Insider Filing Data database, which contains information compiled from Forms 3, 4, 5, and 144 filed with the SEC, and cover the period from 1996 to 2005. Executive and compensation information is from ExecuComp. Stock market information is from the CRSP database. Salary and bonus, value of shares owned, value of vested options, value of unvested options, and previous-year dividends amount are inflation adjusted to year 2000 dollars. See Table I for detailed variable explanations. Column 1 Dependent Variable: Vesting Exercise (Loosely Defined) Coef. Z-Stat CEO 0.2362 ** 2.25 Salary and bonus 0.000015 0.50 January vesting 0.5675 *** 4.28 Percentage of shares owned -0.0503 ** -2.23 Value of shares owned 0.0033 0.30 Value of vested options -0.000039 *** -7.35 Value of unvested options 0.000016 *** 7.87 Recent new grant dummy 0.6589 *** 7.18 Volatility of stock 30.2630 *** 8.99 Postexercise four-factor 16.1119 0.99 model intercept Postexercise two-week 0.3431 0.08 abnormal returns Insider open market 0.1770 ** 2.48 transaction net purchase ratio Monthly risk-free rate -52.2097 -0.68 Number of days from -0.0005 *** -11.36 exercise to expiration Previous-year dividends -0.0197 -0.56 amount Company size 0.0581 * 1.86 High-tech company -0.3605 *** -3.01 Utility firm 0.7147 *** 2.87 Financial firm 0.1701 1.52 Year dummies Yes Test methodology Logit Number of observations 9,387 used Pseudo [R.sup.2] (adjusted [R.sup.2] for 0.1959 Column 3) Column 2 Dependent Variable: Vesting Exercise (Strictly Defined) Coef. Z-Stat CEO 0.2265 * 1.82 Salary and bonus 0.000006 0.17 January vesting 0.4666 *** 2.96 Percentage of shares owned -0.0156 -0.69 Value of shares owned 0.0084 0.61 Value of vested options -0.000038 *** -6.34 Value of unvested options 0.000014 *** 7.20 Recent new grant dummy 0.5362 *** 4.87 Volatility of stock 29.6399 *** 6.86 Postexercise four-factor 11.9330 0.60 model intercept Postexercise two-week -2.0327 -0.38 abnormal returns Insider open market 0.2669 *** 3.12 transaction net purchase ratio Monthly risk-free rate -248.4580 *** -2.61 Number of days from -0.0005 *** -8.84 exercise to expiration Previous-year dividends -0.0516 -0.72 amount Company size 0.1135 *** 3.00 High-tech company -0.6971 *** -4.39 Utility firm 0.7556 *** 2.65 Financial firm -0.0181 -0.13 Year dummies Yes Test methodology Logit Number of observations 9,387 used Pseudo [R.sup.2] (adjusted [R.sup.2] for 0.1639 Column 3) Column 3 Dependent Variable: Percentage of Vesting Exercise Coef. Z-Stat CEO 0.0097 1.58 Salary and bonus 0.000003 ** 2.12 January vesting 0.0335 *** 3.53 Percentage of shares owned -0.0022 * -1.95 Value of shares owned 0.0010 1.48 Value of vested options -0.000001 *** -7.71 Value of unvested options 0.000001 *** 9.68 Recent new grant dummy 0.0426 *** 7.07 Volatility of stock 2.2902 *** 9.08 Postexercise four-factor 2.2852 * 1.82 model intercept Postexercise two-week 0.0177 0.06 abnormal returns Insider open market 0.0164 *** 3.48 transaction net purchase ratio Monthly risk-free rate -3.4334 -0.67 Number of days from 0.0000 *** -8.75 exercise to expiration Previous-year dividends -0.0013 -1.00 amount Company size 0.0029 1.57 High-tech company -0.0342 *** -4.90 Utility firm 0.0496 *** 3.01 Financial firm 0.0101 1.50 Year dummies Yes Test methodology OLS Number of observations 9,387 used Pseudo [R.sup.2] (adjusted [R.sup.2] for 0.1191 Column 3) Column 4 Dependent Variable: Percentage of Vesting Exercise Coef. Z-Stat CEO 0.3473 * 1.91 Salary and bonus 0.000001 0.01 January vesting 0.9491 *** 3.93 Percentage of shares owned -0.0842 ** -2.19 Value of shares owned 0.0132 0.72 Value of vested options -0.000055 *** -7.23 Value of unvested options 0.000024 *** 7.95 Recent new grant dummy 1.0976 *** 6.45 Volatility of stock 49.8589 *** 7.55 Postexercise four-factor 21.5634 0.71 model intercept Postexercise two-week 1.3277 0.17 abnormal returns Insider open market 0.3799 *** 2.99 transaction net purchase ratio Monthly risk-free rate -142.7498 -1.06 Number of days from -0.0008 *** -9.40 exercise to expiration Previous-year dividends -0.0555 -0.69 amount Company size 0.0914 * 1.68 High-tech company -0.6239 *** -3.06 Utility firm 1.1231 ** 2.55 Financial firm 0.1945 0.98 Year dummies Yes Test methodology Tobit Number of observations 9,387 used Pseudo [R.sup.2] (adjusted [R.sup.2] for 0.1551 Column 3) *** Significant at the 0.01 level. ** Significant at the 0.05 level. * Significant at the 0. 10 level.
|Printer friendly Cite/link Email Feedback|
|Author:||Fu, Xudong; Ligon, James A.|
|Date:||Sep 22, 2010|
|Previous Article:||Locked up by a lockup: valuing liquidity as a real option.|
|Next Article:||Can operating leverage be the cause of the value premium?|