What drives the dating game of executive options exercise? Evidence from Taiwan
Ming-Cheng Wu acknowledges financial support from NSC 97-2410-H-018-007. We thank the anonymous reviewer and Robert Faff (Editor) for helpful comments, Shannon Hilscher and Debra Soled for editorial assistance.
Abstract
Examining Taiwanese firms from 2002 to 2008, this paper investigates the motivations behind backdating the exercising of executive stock options. The probability of suspect exercises (backdating) is positively related to the firm’s stock return, the value of the option, tax savings, institutional ownership and the extent of CEO equity ownership and negatively related to firm-specific risk and the use of Big Four accounting firms. Tax incentives motivate executives to backdate the exercise date, implying that the greater the potential for larger tax savings, the greater the likelihood of backdating. Backdating usually occurs in firms that have heavy ownership by the CEO, have more claims to executive stock options and are not family-run, confirming the presence of the agency cost problem.
1. Introduction
This paper examines factors that determine the likelihood of backdating in the exercise of executive stock options, which is the practice of selecting a retroactive date on which the stock price was particularly low for executives to exercise their stock options. Narayanan and Seyhun (2008) indicate that backdating is more likely in smaller firms and where large volumes of options can be exercised, while Dhaliwal et al. (2009) show that the rate of personal income tax influences the behaviour of executives in exercising their stock options. The topic of backdating executive stock options is important because it sheds light on different issues related to the agency cost issue concerning executives, corporate governance and transparency of the firm.1
There are two stock option exercise strategies characterized by executives’ stock disposition on the exercise date: exercise-and-hold and exercise-and-sell. The exercise-and-hold strategy refers to the situation in which executives hold all their acquired stock after exercising the option, while the exercise-and-sell strategy refers to the situation in which executives sell all or part of the acquired stock immediately after exercising the option.
The backdating of stock options exercise refers to the practice of retroactively selecting a date on which the stock price was particularly low to be the exercise date. When executives backdate their stock options exercise, we refer to this strategy as the exercise-and-hold strategy because they cannot sell the shares on the past date when the market price was low.
The tax treatment of stock options in the United States can be grouped into two categories:2 The difference between the stock price and the strike price is subject to ordinary income tax (now as high as 35 per cent). When executives dispose of their shares, they are subject to the capital gains tax. Executives who hold their shares for at least 1 year with capital appreciation can be taxed at a much lower capital gains tax rate. Currently, the ordinary income tax rate can be 233 per cent of the capital gains tax rate (which is at 15 per cent). If executives employ the exercise-and-hold strategy, they have a tax incentive to pay the lower capital gains tax instead of the ordinary income tax.
Executives who exercise stock options in Taiwan face one of two tax treatments.3 When executives exercise their stock options, the spread between the current market price and the strike price is taxed at the ordinary income tax rate. However, when they sell shares after exercising their stock options in Taiwan, they do not pay any capital gains tax because of a higher market share price. In other words, when executives select a date with a low closing price to backdate their exercise date, they do not need to worry about capital gains tax.4
The difference between the tax treatment of stock options exercise in the United States and Taiwan arises because in Taiwan, there is no capital gains tax applied, whereas in the United States, capital gains tax is levied. Thus, Taiwanese executives can ignore capital gains tax in their decision making. As a result, the tax system in Taiwan provides executives a stronger incentive to backdate the exercise date than those executives in the United States.
US executives have to consider the trade-off between the ordinary income tax and the capital gains tax payment when backdating so that their tax bills (combining the ordinary income tax and capital gains tax) will be minimized while they still have the incentive to backdate the exercise date in order to reduce ordinary income tax.
The existing literature supports the notion that the exercise-and-hold strategy is optimal for executives with stock options. For example, McDonald (2003) indicates that if executives expect an increase in their rate of ordinary income tax, then they will find the exercise-and-hold strategy optimal. In addition, Cicero (2009) demonstrates that an exercise-and-hold strategy may be optimal if the exercise date can be chosen ex post to minimize the total tax burden and that this strategy is favoured over any alternative if the executives can backdate the exercise date. Under the current tax regimes in the United States and Taiwan, the exercise-and-hold strategy seems to be the best strategy for executives. This paper examines the effect of tax savings on backdating in Taiwan.
Dhaliwal et al. (2009) examine US CEO stock option exercise data and find that if executives exercise their options and hold shares for at least 1 year, they have an incentive to backdate the stock price on the exercise date to lower their ordinary income tax immediately. Doing so might increase their future income if share prices rise, which will make them liable for paying capital gains tax. When executives in Taiwan exercise their stock options, the spread between the current market price and the strike price is part of the consolidated income of an individual for the exercise year.5 Because Taiwan has no capital gains tax, executives have a stronger incentive to manipulate the exercise date with a low stock price. The tax incentive variable should be an important factor in option backdating in Taiwan.6 Thus, the stock option data for Taiwan offer a unique opportunity to assess the personal tax motivations behind the exercise of stock options.
When executives engage in backdating, they are generally more concerned with total capital gains from the stock price movements. In fact, many studies, such as Carow et al. (2009), Huddart and Lang (2003) and Narayanan and Seyhun (2008), present the same argument, using raw stock returns as the basis for their empirical backdating analysis. Thus, following the literature, we use raw stock returns in our analysis.
Our sample includes both family-run and non-family-run firms, enabling us to analyse the effect of agency costs on the exercise of executive stock options. That is, the results help us better understand how executives behave in the two types of firms. This analysis has important implications, particularly for Asian countries, which have a large number of listed firms that are family-run.
Finally, we provide a comprehensive analysis of the factors behind backdating, including stock return momentum, the value of the options exercise and institutional ownership. In addition, we investigate the effect of the public firm-specific news hypothesis proposed by Roll (1988) on the likelihood of exercising stock options. That is, how is the likelihood of option exercises affected by more firm-specific information that has been capitalized or incorporated into the stock prices? This study adds to the literature on backdating the exercise of executive stock options and has implications for both public policy and corporate governance.
The results indicate that about 17.87 per cent of executive stock options are exercised on the day of the month with the lowest closing price, a rate significantly higher than the 4.8 per cent exercised on any other day of the month.7 Additional analysis shows that for the 381 suspect exercises in our samples, executives saved over NT$130,795 in tax per stock option exercise. That is, by exercising their stock options on the day of the lower closing price, executives realized a total tax savings of more than NT$49.83 million.8 The result implies that tax motivations are an important factor in backdating the exercise of stock options. In addition, the probability of suspect exercises (backdating) is positively related to the firm’s stock return, the value of the option, institutional ownership and the extent of CEO equity ownership and is negatively related to firm-specific risk and the use of Big Four accounting firms. The backdating of the exercise of stock options occurs most often in non-family-run firms, confirming the agency cost problem.
The rest of the article is organized as follows. In Section 2, we briefly discuss the methodology, data and related hypotheses. In Section 3, we present the results and implications. The final section offers our conclusions.
2. Methodology and data
2.1. Method and hypotheses
We use a logistic regression to examine the different hypotheses, as shown in the following model:
(1)Nopshare is the number of stock options exercised in logarithmic form. Balachandran et al. (2008) point out that the incentive for managers to manipulate investors’ perceptions and personally benefit from positive price effects is stronger when managers have equity ownerships. Therefore, we expect this variable to have a positive effect on the dependent variable because the larger the variable, the greater the gain to the option holders.
Momentum is the firm’s stock returns over the exercise date t through t + 30. Past studies argue that backdated stock options are more likely to occur during a month with a high volatility of stock returns, when there are greater opportunities for gains through backdating (Heron and Lie, 2007). An increase in returns from stocks seems to be a critical factor affecting the behaviour of managers, as suggested by Carpenter and Remmers (2001). We hypothesize that the larger this momentum, the higher the probability of backdating.
Value is the stock options exercise volume times the value of the stock options. Brockman et al. (2010) show that the greater the value of the CEO option, the CEO has an incentive to manipulate the timing and content of voluntary disclosures. So, the value variable is expected to have a positive effect on the probability of backdating. We estimate the stock options’ value based on the Black and Scholes (1973) model. However, Rubinstein (1995) argues that the Black–Scholes (BS) model tends to cause overvaluation theoretically, and Marquardt (2002) indicates that the BS method can overvalue the economic cost to shareholders of issuing stock options to executives.9 Executive stock options are generally not standard options (like different vesting periods, non-tradable feature and risk-aversion of executives are) so it is not easy to estimate their parameters. As a result, many studies, such as Bebchuk et al. (2010), Bergman and Jenter (2007), Core and Guay (2002), McAnally et al. (2008) and Mehran (1995), continue to use the BS model to estimate the value of executive stock options. The slight overvaluation result is not expected to affect the sign or direction of the relation, so we follow prior studies and use the BS model here.
Concurrent is a dummy variable that equals one if multiple executives who work at the same firm exercise stock options on the same day and zero otherwise. Hall and Liebman (1998) show that when executives hold large stock options and multiple executives at the same firm exercise their stock options on the same date, they do so at a favourable stock price. Cicero (2009) argues that if multiple executives at the same firm exercise options around the release of information or backdate exercises with favourable prices, it is more likely that they will occur on the same day. Thus, we include the concurrent variable in the regression model, which indicates option exercises with multiple executives on the same day in order to examine whether suspect exercises are associated with favourable stock prices for executives. As a result, this variable is expected to have a positive effect on the probability of backdating.
Big4 is a dummy variable that equals one if the auditor is from one of the Big Four accounting firms (Deloitte, KPMG, PwC, or Ernst & Young) and zero otherwise. According to Taiwan’s regulations, the auditing firm is responsible for ensuring that stock option information has been adequately disclosed in the financial statements and for ascertaining whether proper accounting procedures have been followed.10 Firms audited by the major accounting companies are less likely to backdate if they are experienced in detecting such behaviour and face a greater risk to their own reputation if it occurs on their watch so they might not encourage their clients to backdate the exercising of their stock options. Collins et al. (2009) find that firms that engage in backdating are more likely to be audited by non-Big Four accounting firms than those that are audited by the Big Four.
Tax savings represents the potential tax savings, i.e., 
, where N is the number of stock options exercised; tax is the personal income tax rate; 
is the average closing price during the exercise month; 
is the closing price of the exercise date; and 
is the potential reduction in taxable income per share achieved by exercising the options at 
.
Dhaliwal et al. (2009) show that executives consider the effect of personal income tax liability on their decision to exercise their stock options. The Taiwan Income Tax Act states that the spread between the closing price on the exercise date of stock options and the strike price of stock options should be added to gross income. As mentioned earlier, Taiwan did not levy capital gains tax on the securities’ transactions during our sample study, unlike the United States, which does assess these transactions for capital gains. Because they are not subject to capital gain taxes, executives in Taiwan have a stronger incentive to identify a date with a low closing price. Thus, we expect this variable to have a positive effect on the probability of backdating.
Board is a (1,0) dummy variable that equals one if the executives who exercise stock options are members of the board of directors and zero otherwise. Regarding the role of the board of directors, Ryan and Wiggins (2002) indicate that when the CEO also chairs the board of directors, the decision making and the monitoring of those decisions cause a potential conflict of interest, implying a less-effective board. We expect this variable to be positive because of the agency cost problem.
Size is the natural logarithm of a firm’s market capitalization. It serves as a control variable. Bhushan (1989) and Lang and Lundholm (1993) indicate that smaller firms’ managers are more likely to engage in self-interested behaviour because their firms are subject to less scrutiny and transparency than larger firms, while Narayanan and Seyhun (2008) demonstrate a small-firm effect on backdating. Dhaliwal et al. (2009) use firm size as a proxy for the strength of internal supervision, finding that backdated exercises are more likely when the firm has relatively weak internal oversight. If such logic prevails, the size variable would be expected to be negative.
Institution is the sum of ownership by the government, domestic financial institutions and foreign financial institutions compared with total ownership. Pound (1988) relates the effects of institutional ownership to firm performance through three hypotheses: (i) the efficient-monitoring hypothesis, (ii) the conflict of interest hypothesis and (iii) the strategic-alignment hypothesis. The efficient-monitoring hypothesis predicts a positive effect of institutional ownership on the value of the firm because of proper monitoring on improper managerial behaviour, thus implying a negative effect of institutional ownership on the likelihood of backdating. By contrast, the conflict of interest hypothesis and the strategic-alignment hypothesis both predict a negative relation between institutional ownership and the value of the firm arising from collusion by managers against the best interest of atomistic shareholders, implying a positive effect on the likelihood of backdating. In short, the effect of institutional ownership, Institution, on the likelihood of backdating can be positive or negative.
Non-family is a (1,0) dummy variable that equals one if the executives exercise stock options in a non-family-run firm and zero otherwise. We identify a firm as family-run or not as measured by ownership: (i) the shares are directly owned by family members; (ii) the cross-shareholdings of listed companies are in the same conglomerate; and (iii) the shareholdings of the nominal agent are controlled by the family. If the family’s holding of the firm’s outstanding shares exceeds 20 per cent, then it is a family-run firm.
Publicly traded family-run firms are quite common around the world (Burkart et al., 2003). Compared to non-family-run firms, family-run firms appear to face less-severe agency problems, which typically result from the separation of ownership and management (Type I agency problem).11Ryan and Wiggins (2002) indicate that CEOs who are members of the founding family can closely monitor the firm’s decisions. McConnell and Servaes (1990) and Morck et al. (1988) observe that managers’ and shareholders’ interests become more closely aligned as managerial ownership increases, resulting in improved firm performance. However, as managers’ equity stakes increase, their interests begin to diverge from those of the shareholders, leading to greater agency problems and declining firm performance.
We examine the agency cost problem using data for family-run and non-family-run firms in backdating the exercise of stock options. Such backdating is more likely to occur in non-family-run firms than in family-run firms. We predict that this variable will be positive and significant.
CEOown is the percentage of the executive’s equity ownership during the month of the stock options exercise. If executives have a higher stock ownership in the firm, it can imply managerial entrenchment, resulting in a positive association between CEOown and the probability of suspect option exercises. Collins et al. (2009) find that higher CEO ownership in a firm increases the CEO’s ability to influence his/her own pay through backdating stock options. Because of the self-interest argument pertaining to executives, we expect this variable to be positively related to backdating as executives may have more claims to executive stock options with a higher CEOown.
Firm-specific risk is measured by idiosyncratic risk (Roll, 1988). We follow Bali et al. (2005) in computing the idiosyncratic risk for each stock, which is estimated within 30 days of the exercise date using daily return data. Firm-specific risk (variation) and disclosure are an integral part of corporate governance. Morck et al. (2000) document that higher firm-specific risk (variation) is associated with stronger public investor property rights because these firms are better known to the public. In addition, Li et al. (2004) find that higher firm-specific risk implies greater capital market openness in emerging markets, including Taiwan. It appears that greater firm-specific return variation indeed provides more informed stock pricing for firms (Durnev et al., 2003). Ferreira and Laux (2007) indicate that (i) firms with high levels of transparency display high levels of idiosyncratic risk and (ii) poor firm-level corporate governance is associated with low levels of idiosyncratic volatility. Thus, high idiosyncratic volatility reflects higher corporate transparency and good corporate governance.
Thus, it seems reasonable to assume that firms with higher firm-specific variation have greater transparency and supply more public information about the firm. Higher transparency and better disclosure reduce the information asymmetry between a firm’s management and the financial stakeholders (equity and bond holders), mitigating the agency problem in corporate governance (Patel et al., 2002). In the light of the above literature, we hypothesize that the firm-specific risk variable will have a negative effect on the likelihood of suspect exercise.
2.2. Data
We examine a sample of 1321 executive exercise stock options from 2002 to 2008 for firms from the Taiwan Stock Exchange Corporation (TSEC) and the Gre Tai Securities Market (an over-the-counter market). We collect data on the exercising of stock options from the Market Observation Post System (MOPS) and the Taiwan Economic Journal Database (TEJDB).
We exclude the exercising of stock options for which we have insufficient stock price data. Our final sample consists of 1214 individual exercises, with option exercises across 209 companies.
The data on option exercises are selected for this study because executives have the power to make decisions that use a company’s inside information. Thus, the data set enables us to analyse the different motivations behind the decision to exercise stock options. The executive positions in this study include chief executive officer, president, vice-president, chief financial officer, chief operating officer, general manager and deputy general manager.
Table 1 provides a descriptive summary of the key variables. One interesting point is the tax savings variable. The average tax savings is NT$56,489, while the maximum value of tax savings is NT$2,120,075. These tax savings are substantial, indicating that tax incentives do motivate executives to manipulate the exercise date of their stock options.
| Variable | Mean | Std. Dev. | Median | Minimum | Maximum |
|---|---|---|---|---|---|
| Nopshare | 11.1085 | 1.1314 | 11.0509 | 5.2983 | 14.9141 |
| Momentum | 0.3392 | 1.8530 | 0.2113 | −10.0021 | 11.0947 |
| Value | 32.6213 | 57.8871 | 13.8300 | 0.0447 | 749.5786 |
| Concurrent | 0.3995 | 0.4900 | 0 | 0 | 1 |
| Big4 | 0.9522 | 0.2134 | 1 | 0 | 1 |
| Tax savings | 0.5649 | 1.5893 | 0.0263 | 0.0000 | 21.2008 |
| Board | 0.4283 | 0.4950 | 0 | 0 | 1 |
| Size | 8.8300 | 1.2079 | 8.6648 | 6.0890 | 14.2744 |
| Institution | 0.3452 | 0.2020 | 0.3040 | 0.0004 | 0.8523 |
| Non-family | 0.6647 | 0.4723 | 1 | 0 | 1 |
| CEOown | 0.1755 | 0.3243 | 0.0043 | 0 | 0.3735 |
| Firm-specific risk | 5.7173 | 3.9080 | 4.7509 | 0.3010 | 33.7061 |
- Number of observations is 1214. Nopshare is the number of stock options exercised in log. Momentum is the firm’s compounded return at exercise date t through t + 30. Value is the stock option exercise volume times the stock option price, as calculated by the Black–Scholes model in NT$100,000. Concurrent = 1 if multiple executives who work at the same firm exercise stock options on the same day; otherwise = 0. Big4 = 1 if firm’s auditor is Deloitte, KPMG, PwC, or Ernst & Young; otherwise = 0. The tax savings variable, Tax savings, is constructed as the potential tax savings = N × tax × (P* − Pexercise) in NT$100,000. N is the number of stock options exercised. tax is the personal income tax rate. P* is the average closing price during the exercise month. Pexercise is the closing price of the exercise date. Board = 1 if the executives who exercise stock options are members of the board of directors, otherwise = 0. Size is a firm’s market capitalization in log. Institution is the sum of ownership by the government, domestic financial institutions and foreign financial institutions relative to total ownership. Non-family = 1 if executives are from non-family-run firms; otherwise = 0. CEOown is the percentage of the manager’s equity ownership during the stock option exercise month. Firm-specific risk is measured by idiosyncratic volatility computed from the capital asset pricing model (CAPM).
3. Empirical results
3.1. The incidence of suspect stock options exercise
We classify suspect stock options as those that were exercised if the stock price on the exercise date is one of the three lowest closing prices during the month of the exercise date. Figure 1 shows that 17.87 per cent of executives exercise stock options on the day with the lowest closing price of the month, even though the expected percentage would be 4.8 per cent (see footnote 7). The frequency of exercising stock options on the second-lowest closing price of the month is 6.59 per cent and on the third-lowest closing price of the month is 6.92 per cent. Clearly, the results indicate that when Taiwanese executives exercise their stock options, they tend to backdate their exercise date.
Percentage of executive stock options exercised on the day of the month and its closing price. This figure plots the percentage of executive stock options exercised on each day of the month and its closing price. The lowest closing price has a rank of 1, the second-lowest price 2 and so on.
Table 2 provides a statistical analysis of the raw stock returns surrounding when executives exercise their stock options. The table indicates positive increases after the stock options exercise date. The subsequent stock returns are significantly higher than zero. These results are consistent with our hypothesis that executives backdate the exercising of their stock options.
| Window | All samples N = 1214 | <100,000 shares N = 841 | >100,000 shares N = 373 | Difference (a−b) |
|---|---|---|---|---|
| Cumulative raw returns | Cumulative raw returns (a) | Cumulative raw returns (b) | ||
| (−20,0) | 0.0803 (0.142) | −0.2495 (−0.312) | 0.6718 (0.886) | −0.9213 (−1.032) |
| (−10,0) | 0.3288 (0.809) | 0.3666 (0.791) | 0.2605 (0.517) | 0.1061 (0.169) |
| (−5,0) | −0.1360 (−0.457) | −0.1149 (−0.318) | −0.1731 (−0.430) | 0.0582 (0.122) |
| (−3,0) | 0.0634 (0.453) | 0.2081 (1.088) | −0.2023 (−0.993) | 0.4104 (1.307) |
| (1,3) | 0.5774*** (3.645) | 0.4435*** (3.809) | 0.8392** (2.022) | −0.3957 (1.215) |
| (1,5) | 0.4531 (1.311) | 0.3967 (1.592) | 0.5832 (0.927) | −0.1865 (−0.444) |
| (1,10) | 0.9153** (2.541) | 0.7299** (2.530) | 1.2659* (1.868) | −0.5360 (−0.891) |
| (1,20) | 2.1499*** (3.772) | 1.8459*** (3.467) | 2.7132*** (3.063) | −0.8673 (−1.000) |
- t-Values are in parentheses. ***Significant at the 1 per cent level; **significant at the 5 per cent level; and *significant at the 10 per cent level.
It is expected that such backdating by executives is more pronounced as the number of stock options exercised rises because it increases the executives’ potential tax savings. As a result, we divide the whole sample into two subsamples according to the number of stock options exercised: <100,000 and >100,000 shares per exercise. The result is consistent with our expectations. The large volume of stock options exercised is positively correlated with cumulative raw stock returns. Table 2 shows that raw stock returns increase with the exercise volume after the stock options are exercised. However, there is no statistical significance in returns for an exercise volume of <100,000 and >100,000 shares, implying that there is no size effect.
Figure 2 shows greater cumulative raw stock returns for the non-family-run firms than for family-run firms. The 20-day post-exercise cumulative raw stock returns are 0.9234 per cent for a family-run firm and 2.7685 per cent for a non-family-run firm; these differences are statistically significant (see also Table 3). In our all option exercise samples, there are 209 companies (i.e. 72 family-run firms and 137 non-family-run firms). In our family-run and non-family-run firm samples, executives include the chief executive officer, president, vice-president, chief financial officer, chief operating officer, general manager and deputy general manager. The difference in market capitalization between family-run firms and non-family-run firms is not statistically significant.12 Family-run firms’ executives increase their own wealth by increasing the value of the firm because their wealth is tied up with the firm. Backdating can damage the firm’s reputation if it is detected.
Stock return behaviours around the exercise date for family-run and non-family-run firms. Samples are classified into two subsamples: family-run firms (N = 407) or non-family-run firms (N = 807).
| Window | All samples N = 1214 | Non-family N = 807 | Family N = 407 | Difference (a−b) |
|---|---|---|---|---|
| Cumulative raw returns | Cumulative raw returns (a) | Cumulative raw returns (b) | ||
| (−20,0) | 0.0803 (0.142) | 0.6618 (1.258) | −1.0729 (−1.089) | 1.7347** (2.010) |
| (−10,0) | 0.3288 (0.809) | 0.8288*** (4.233) | −0.6627 (−0.941) | 1.4915** (2.353) |
| (−5,0) | −0.1360 (−0.457) | 0.3193*** (9.017) | −1.0388** (−2.229) | 1.3581*** (2.823) |
| (−3,0) | 0.0634 (0.453) | 0.3260*** (7.320) | −0.4573 (−1.070) | 0.7833* (1.957) |
| (1,3) | 0.5774*** (3.645) | 0.8225*** (10.724) | 0.0912 (0.731) | 0.7313** (2.222) |
| (1,5) | 0.4531 (1.311) | 0.7392*** (18.656) | −0.1142 (−0.519) | 0.8534** (2.011) |
| (1,10) | 0.9153** (2.541) | 1.1307*** (18.545) | 0.1312 (0.304) | 0.9995* (1.941) |
| (1,20) | 2.1499*** (3.772) | 2.7685*** (12.615) | 0.9234 (1.182) | 1.8451** (2.106) |
- t-Values are in parentheses. ***Significant at the 1 per cent level; **significant at the 5 per cent level; and *significant at the 10 per cent level.
Table 4 presents summary statistics on suspect exercising and non-suspect exercising of stock options. The mean for tax-suspect exercises is NT$130,795, which is significantly higher than that for non-suspect exercises. This result also supports the notion that executives backdate the exercise of their stock options for tax reasons.
| Suspect exercises | Non-suspect exercises | P-value (difference) | |
|---|---|---|---|
| Number of observations | 381 | 833 | — |
| Potential tax savings (mean) | NT$130,795 | NT$22,503 | <0.001 |
| Potential tax savings (median) | NT$59,796 | NT$0 | <0.001 |
- Potential tax savings is N × tax × (P* − Pexercise). N is the number of stock options exercised. tax is the personal income tax rate. P* is the average closing price during the exercise month. Pexercise is the closing price of the exercise date. If the closing price on the exercise date is higher than the average closing price during the exercise month, the potential tax savings equals zero.
Table 5 reports summary statistics on executives’ exercise of stock options in family-run firms and non-family-run firms. The descriptive statistics differ greatly between family-run and non-family-run firms.13 The mean of family-held equity is 0.3066 in family-run firms and 0.1154 in non-family-run firms. The mean of backdating (suspect exercise) is 0.2752 in family-run firms and 0.3333 in non-family-run firms. The results indicate that the likelihood of suspect exercises is higher in non-family-run firms than in family-run firms.
| Variables | Mean | Std. Dev. | Median | Minimum | Maximum |
|---|---|---|---|---|---|
| Family-held equity | |||||
| Family sample | 0.3066 | 0.1020 | 0.2757 | 0.2007 | 0.6710 |
| Non-family sample | 0.1154 | 0.0623 | 0.1270 | 0.0000 | 0.1995 |
| Backdating | |||||
| Family sample | 0.2752 | 0.4472 | 0.0000 | 0.0000 | 1.0000 |
| Non-family sample | 0.3333 | 0.4717 | 0.0000 | 0.0000 | 1.0000 |
| Nopshare | |||||
| Family sample | 4.9074 | 0.5066 | 4.8968 | 2.8751 | 6.4771 |
| Non-family sample | 4.7825 | 0.4783 | 4.7690 | 2.3010 | 6.4771 |
| Momentum | |||||
| Family sample | 0.2176 | 1.7200 | −0.0151 | −4.4663 | 4.6485 |
| Non-family sample | 0.4004 | 1.9143 | 0.3937 | −10.0021 | 11.9047 |
| Value | |||||
| Family sample | 34.6456 | 65.0319 | 14.8345 | 0.1581 | 749.5786 |
| Non-family sample | 31.6000 | 53.9404 | 13.5428 | 0.0447 | 535.9256 |
| Concurrent | |||||
| Family sample | 0.3342 | 0.4723 | 0.0000 | 0.0000 | 1.0000 |
| Non-family sample | 0.4325 | 0.4957 | 0.0000 | 0.0000 | 1.0000 |
| Big4 | |||||
| Family sample | 0.9509 | 0.2164 | 0.0000 | 0.0000 | 1.0000 |
| Non-family sample | 0.9529 | 0.2120 | 0.0000 | 0.0000 | 1.0000 |
| Tax savings | |||||
| Family sample | 0.5558 | 1.5694 | 0.0000 | 0.0000 | 14.6250 |
| Non-family sample | 0.5694 | 1.6002 | 0.0420 | 0.0000 | 21.2008 |
| Board | |||||
| Family sample | 0.3882 | 0.4879 | 0.0000 | 0.0000 | 1.0000 |
| Non-family sample | 0.4486 | 0.4977 | 0.0000 | 0.0000 | 1.0000 |
| Size | |||||
| Family sample | 8.9709 | 1.0812 | 8.9276 | 6.4583 | 12.2378 |
| Non-family sample | 8.7590 | 1.2616 | 8.5475 | 6.0890 | 14.2744 |
| Institution | |||||
| Family sample | 0.4229 | 0.4374 | 0.2126 | 0.0027 | 0.8523 |
| Non-family sample | 0.3061 | 0.1845 | 0.2697 | 0.0400 | 0.8308 |
| CEOown | |||||
| Family sample | 0.0173 | 0.0433 | 0.0027 | 0.0000 | 0.3735 |
| Non-family sample | 0.0177 | 0.0253 | 0.0058 | 0.0000 | 0.1711 |
| Firm-specific risk | |||||
| Family sample | 5.3345 | 3.7588 | 4.1811 | 0.4431 | 21.2189 |
| Non-family sample | 5.9104 | 3.9708 | 5.0051 | 0.3010 | 33.7061 |
- Family-held equity is the percentage of the sum of the shares directly owned by family members, the cross-shareholdings of listed companies are in the same conglomerate and the shareholdings of the nominal agent controlled by the family.
The stock return momentum is much higher in non-family-run firms (0.4004) than in family-run firms (0.2176). That is, the non-family-run firms’ stock return momentum is about twice as high as in family-run firms. These results also support the conclusion that suspect exercises are more likely in non-family-run firms. Executives who work at non-family-run firms appear to display more self-interest because of the agency cost problem.
3.2. Logistic regression results
We use a logistic regression model to examine the relation between various factors behind backdating and the likelihood of suspect exercises. Table 6 reports the results of the logistic regression. Two versions are presented. In Model 1, the likelihood of suspect exercises is positively and significantly related to (i) Nopshare, which is the number of stock options exercised (coeff. = 0.429, P < 5 per cent), (ii) the stock return Momentum (coeff. = 0.597, P < 1 per cent) and (iii) Value, which is the value of stock options (coeff. = 0.019, P < 1 per cent) and negatively related to the auditor type (coeff. = −1.118, P < 1 per cent). The estimated marginal effect of Nopshare is 0.0713. This result implies that an increase in Nopshare of one standard deviation will increase the probability of backdating by 7.13 per cent. The estimated marginal effect of Momentum is 0.1631, implying that an increase of one standard deviation in Momentum will increase the probability of backdating by 16.31 per cent. Similarly, the results indicate that an increase of one standard deviation in Value will increase the probability of backdating by 17.37 per cent. The firms audited by the Big Four accounting firms (Big4) appear to have a negative and significant coefficient, which implies a lower likelihood of suspect exercises. That is, an increase of one standard deviation in Big4 is associated with a 3.52 per cent reduction in the probability of backdating. Higher-quality auditors are expected to follow rules and regulations and have incentives to protect their reputation and brand, thus lowering the likelihood of suspect exercises, a result consistent with our prediction.
| Variable | Predicted sign | Model 1 | Model 2 | ||
|---|---|---|---|---|---|
| Coefficient | Marginal effect | Coefficient | Marginal effect | ||
| Constant | ? | −3.544*** (0.002) | −2.217** (0.056) | ||
| Nopshare | + | 0.429** (0.031) | 0.0713 | 0.321 (0.112) | 0.0532 |
| Momentum | + | 0.597*** (0.001) | 0.1631 | 0.707*** (0.001) | 0.1890 |
| Value | + | 0.019*** (0.001) | 0.1737 | 0.020*** (0.001) | 0.1737 |
| Concurrent | + | 0.213 (0.174) | 0.0152 | 0.176 (0.269) | 0.0127 |
| Big4 | − | −1.118*** (0.001) | −0.0352 | −1.287*** (0.001) | −0.0399 |
| Tax savings | + | 0.814*** (0.001) | 0.1971 | 0.912*** (0.001) | 0.2114 |
| Board | + | −0.225 (0.174) | −0.0163 | −0.188 (0.260) | −0.0134 |
| Size | − | 0.072 (0.374) | 0.0133 | 0.473 (0.567) | 0.0085 |
| Institution | ± | 1.185** (0.018) | 0.0354 | 1.280** (0.012) | 0.0376 |
| Non-family | + | 0.500*** (0.005) | 0.0350 | 0.553*** (0.002) | 0.0378 |
| CEOown | + | 4.871** (0.046) | 0.2332 | 4.326* (0.080) | 0.2040 |
| Firm-specific risk | − | −0.113*** (0.001) | −0.0625 | ||
| #obs. | 1214 | 1214 | |||
| % Suspect exercises | 31.39% | 31.39% | |||
| −2 log likelihood | 1101.308 | 1079.232 | |||
| Cox & Snell R2 | 28.6% | 29.9% | |||
| Nagelkerke R2 | 40.2% | 42% | |||
- The dependent variable is a dummy variable (backdating), which is 1 if executives exercise stock options with lowest, second-lowest and third-lowest closing stock price during the month, and 0 otherwise (i.e. backdating = 0). Nopshare is the number of stock option exercises in log. Momentum is the firm’s compounded return at exercise date t through t + 30. Value is the stock option exercise volume times the stock option price by the Black–Scholes model in NT$100,000. Concurrent = 1 if multiple executives who work at the same firm exercise stock options on the same day; otherwise = 0. Big4 = 1 if firm’s auditor is Deloitte, KPMG, PwC, or Ernst & Young; otherwise = 0. Tax savings is constructed as potential tax savings = N × tax × (P* − Pexercise) in NT$100,000. N is the number of stock options exercised. tax is the personal income tax rate. P* is the average closing price during the exercise month. Pexercise is the closing price of the exercise date. Board = 1 if the executives exercising stock options are members of the board of directors; otherwise = 0. Size is a firm’s market capitalization in log. Institution is the sum of ownership by the government, domestic financial institutions, and foreign financial institutions relative to total ownership. Non-family = 1 if executives are from non-family-run firms; otherwise = 0. CEOown is the percentage of the manager’s equity ownership during the stock option exercise month. Firm-specific risk is measured by idiosyncratic volatility computed from the capital asset pricing model (CAPM). Marginal effect for one standard deviation of the variable (i) is beta (i) × [exp (backdating)/(exp (backdating)+1)2] × standard deviation (i). Tests of the multicollinearity among independent variables via variance inflation factor (VIF) show no evidence for multicollinearity between the model variables (tolerance range between 0.58 and 0.95, mean VIF 1.317). P-values are in parentheses. ***, ** and * denote significance at 0.01, 0.05 and 0.1 levels, respectively.
Tax savings (coeff. = 0.814, P < 1 per cent) are significantly positive in predicting the likelihood of suspect exercises. The estimated marginal effect of Tax savings is 0.1971, indicating that an increase in Tax savings of one standard deviation is associated with a 19.71 per cent increase in the probability of backdating. This finding is consistent with the hypothesis that tax incentives motivate executives to manipulate their option exercise date. The greater the potential tax savings, the greater the likelihood of suspect exercises.
The Institution variable is positive and has a significant effect (coeff. = 1.185, P < 5 per cent) on the likelihood of suspect exercises, implying that an increase in Institution of one standard deviation will increase the probability of backdating by 3.54 per cent. This result supports the conflict of interest hypothesis and the strategic-alignment hypothesis by Pound (1988).
It is interesting to note that such backdating practices usually occurred in non-family-run firms. The fact of being a non-family-run firm has a significant and positive effect (coeff. = 0.500, P < 1 per cent) on the likelihood of suspect exercises. An increase in Non-family of one standard deviation will trigger a 3.5 per cent increase in the probability of backdating. Executive stock options are part of the compensation package for executives, as part of an effort to reduce Type I agency problems. So, backdating violates the original purpose of granting executive stock options and creates a different serious agency problem.
The CEOown variable is positive and has a significant effect (coeff. = 4.871, P < 5 per cent) on the likelihood of suspect exercises. We find that an increase in CEOown of one standard deviation increases the probability of backdating by 23.32 per cent. It is important to note that good corporate governance principles need to be applied to deter self-interest behaviour by CEOs.
Model 2 shows results similar to those in Model 1, but it further demonstrates the importance of the public firm-specific risk (variation) that affects the suspect exercises. The Firm-specific risk variable is negative and significant (coeff. = −0.113, P < 1 per cent). The result indicates that an increase of one standard deviation in Firm-specific risk reduces the probability of backdating by 6.25 per cent, a finding that supports the argument that greater firm-specific risk (variation), a proxy for transparency, reduces the motivation for executives to backdate the exercising of their stock options. This is an interesting and important result because the current corporate literature does not seem to document corporate governance’s strong role in the backdating of exercising stock options.
In the light of the marginal effect that a one standard deviation change can have on the probability of backdating, we find that the Tax savings variable appears to have the most noticeable effect on backdating (i.e. a marginal effect of 21 per cent in Model 2). The other three important variables according to Model 2 are CEOown (about 20 per cent), Momentum (19 per cent) and Value (17 per cent).
To check for robustness, we examine different ways of defining the dependent variable (suspect exercises). We use firms in a group with the lowest closing price as the dependent variable.14 The results are shown in Table 7. Two interesting points are noted here. First, in Model 1 of Table 7, the Concurrent variable (coeff. = 0.376, P < 5 per cent) is positive and becomes significant when compared to the insignificant result in Table 6. The estimated marginal effect of the Concurrent variable is 0.0216, indicating that a one standard deviation increase in concurrent is associated with a 2.16 per cent increase in the probability of backdating. That is, when two or more executives simultaneously backdate to the same exercise day with the lowest closing price, the probability of backdating becomes stronger and significant. Second, the CEOown variable is no longer significant in Table 7, implying that the executive who has higher equity ownership intends to backdate the exercise date to a lower closing price, but not necessarily the lowest. The results in Table 7 in general are largely similar to those in Table 6. Thus, our conclusion about the driving forces behind the backdating decision is robust.
| Variable | Predicted sign | Model 1 | Model 2 | ||
|---|---|---|---|---|---|
| Coefficient | Marginal effect | Coefficient | Marginal effect | ||
| Constant | ? | −5.136*** (0.001) | −3.599*** (0.008) | ||
| Nopshare | + | 0.567** (0.014) | 0.0758 | 0.420* (0.071) | 0.0554 |
| Momentum | + | 0.423*** (0.001) | 0.0927 | 0.559*** (0.001) | 0.1204 |
| Value | + | 0.022*** (0.001) | 0.1737 | 0.023*** (0.001) | 0.1737 |
| Concurrent | + | 0.376** (0.032) | 0.0216 | 0.355** (0.045) | 0.0201 |
| Big4 | − | −0.554 (0.152) | −0.0139 | −0.746* (0.058) | −0.0186 |
| Tax savings | + | 0.458*** (0.001) | 0.0858 | 0.502*** (0.001) | 0.0922 |
| Board | + | −0.193 (0.303) | −0.0114 | −0.133 (0.481) | −0.0079 |
| Size | − | 0.026 (0.773) | 0.0036 | 0.003 (0.976) | 0.0036 |
| Institution | ± | 1.794*** (0.001) | 0.0428 | 1.883*** (0.001) | 0.0442 |
| Non-family | + | 0.522*** (0.009) | 0.0293 | 0.552*** (0.006) | 0.0302 |
| CEOown | + | 1.889 (0.464) | 0.0723 | 1.402 (0.593) | 0.0529 |
| Firm-specific risk | − | −0.116*** (0.001) | −0.0547 | ||
| #obs. | 1214 | 1214 | |||
| % Suspect exercises | 31.39% | 31.39% | |||
| −2 log likelihood | 927.691 | 908.674 | |||
| Cox & Snell R2 | 16.0% | 17.3% | |||
| Nagelkerke R2 | 26.3% | 28.5% | |||
- The dependent variable is a dummy variable, which is equal to 1 if executives exercise stock options at the lowest closing stock price during the month (i.e. backdating = 1), and otherwise = 0. Nopshare is the number of stock option exercises in log. Momentum is the firm’s compounded return at exercise date t through t + 30. Value is the stock option exercise volume times the stock options price, as calculated by the Black–Scholes model in NT$100,000. Concurrent = 1 if multiple executives who work at the same firm exercise stock options on the same day; otherwise = 0. Big4 = 1 if the firm’s auditor is Deloitte, KPMG, PwC, or Ernst & Young; otherwise = 0. Tax savings is constructed as the potential tax savings = N × tax × (P* − Pexercise) in NT$100,000. N is the number of stock options exercised. tax is the personal income tax rate. P* is the average closing price during the exercise month. Pexercise is the closing price of the exercise date. Board = 1 if the executives who exercise stock options are members of the board of directors; otherwise = 0. Size is a firm’s market capitalization in log. Institution is the sum of ownership by the government, domestic financial institutions and foreign financial institutions relative to total ownership. Non-family = 1 if executives are from non-family-run firms; otherwise = 0. CEOown is the percentage of the manager’s equity ownership during the stock option exercise month. Firm-specific risk is measured by idiosyncratic volatility computed from the capital asset pricing model (CAPM). The marginal effect of one standard deviation of the variable (i) is beta (i) × [exp (backdating)/(exp (backdating)+1)2] × standard deviation (i). Tests of the multicollinearity among independent variables via variance inflation factor (VIF) show no evidence for multicollinearity between the model variables (tolerance range between 0.58 and 0.95, mean VIF 1.317). P-values are in parentheses. ***, ** and * denote significance at 0.01, 0.05 and 0.1 levels, respectively.
4. Conclusions
This paper conducts a comprehensive analysis of the motivations behind the exercising of executive stock options. We include in our analysis Taiwanese firms covering the 7-year span from 2002 to 2008. The data set contains observations of 1214 individual exercises. We focus on how tax incentives drive executives to manipulate the date of exercising their stock options. Executives reduced their tax liability by more than NT$49.83 million in 381 suspect exercises in all exercise samples, more than NT$130,795 per stock option. In addition, we find that 17.87 per cent of executives exercise their stock options on the day with the lowest closing price of the month. The frequency of exercises on the second-lowest closing price of the month is 6.59 per cent and the third lowest is 6.92 per cent.
In this study, we explore several motivations that may affect the exercising of stock options. Our results contribute to the existing literature in several ways. First, we document the presence of an upward-momentum factor, which is important in explaining the decision by executives to backdate the exercising of their stock options. The upward movement of firms’ stock prices provides executives with the best timing to manipulate the exercise date; hence, they benefit financially from backdating.
Second, we explore backdating practices in non-family-run firms. Issuing stock options may motivate executives to work hard, but this study shows that backdating the exercising of stock options confirms the agency cost problem. That is, family-run firms exert tight control over the firms, thus reducing the likelihood of backdating. The results imply that a better governance structure within the firm or through a regulatory body needs to be implemented effectively in order to reduce the agency problem among non-family-run firms listed on the stock exchange.
Finally, our results indicate that a greater firm-specific risk variable has a negative and significant effect on the likelihood of suspect exercises. Executives who work in firms with lower firm-specific risk (variation) are more likely to manipulate the date of exercising their stock options. This study demonstrates clearly the presence of self-interested motivations among executives in backdating the exercising of their stock options.
Our study contributes to the literature by providing a comprehensive framework for examining the exercising of executive stock options. Specifically, our results shed light on corporate governance, the agency cost problem among executives and the role of public firm-specific information relating to firm behaviour.
