- Email: [email protected]
Stock options: From backdating to spring loading
G Model
ARTICLE IN PRESS
QUAECO-866; No. of Pages 7
The Quarterly Review of Economics and Finance xxx (2015) xxx–xxx
Contents lists available at ScienceDirect
The Quarterly Review of Economics and Finance journal homepage: www.elsevier.com/locate/qref
Stock options: From backdating to spring loading Giuliano Bianchi ∗ Ecole hôtelière de Lausanne, HES-SO // University of Applied Sciences Western Switzerland, Route de Cojonnex 18, Lausanne 1000, Switzerland
a r t i c l e
i n f o
Article history: Received 10 June 2014 Received in revised form 11 June 2015 Accepted 23 July 2015 Available online xxx Keywords: CEO pay Stock options Price targets Backdating Spring loading
a b s t r a c t In this article I explore the impact of the introduction of the Sarbanes–Oxley Act (SOX) in 2002 and the Securities and Exchange Commission’s implementation of the Act in 2006 on the options granting process. I show that after the introduction of the SOX and its implementation the practice of backdating options was substituted with the practice of “spring loading” options around analysts’ price targets announcements. © 2015 The Board of Trustees of the University of Illinois. Published by Elsevier B.V. All rights reserved.
1. Introduction Managers are contracted to act in the best interests of shareholders and to maximize shareholders’ wealth. Ideally, managers’ pay is set according to their effort and productivity, a solution known as the first best solution. However, managers’ interests are not fully aligned with those of shareholders. Plus, the board of directors cannot completely monitor managers’ work, and it does not possess all of the information that is available to managers. Thus, managers might engage in rent-seeking activities at the shareholders’ expense. These types of conflicts lead to ‘agency problems’. Under such circumstances, the first best solution is not possible and agency theory suggests adopting the “second best solution”1 . The second best solution consists of sharing part of the company’s ownership with managers in order to realign the interests of the agent with those of the principal. Stock options are among the many financial tools used to link CEOs’ compensation to companies’ performance. Stock options became very popular in the United States, especially in 2000–2001 when they were the largest single component of CEO compensation (as shown in Fig. 12 ).
∗ Tel.: +0041 217 851 456. E-mail address: [email protected] 1 For a comprehensive review of this concept see Eisenhardt (1989). 2 Data on the composition of CEOs compensation by fiscal year are collected from Execucomp. I retain only CEOs data, using the variable CEOANN that indicates whether the officer served as CEO for the entire or most of the fiscal year or not. The estimated value of the stock options granted to CEOs is approximated by Execucomp
Stock options give the option holder the right to buy company stock at a pre-established strike price (also referred to as the exercise price). In the U.S., stock options are usually granted “at the money”, i.e. the strike price is the share price at the grant date. As the Wall Street Journal (Forelle, Bandler, & Anders, 2006) pointed out, granting options “in the money” (i.e. when the strike price is below the share price) is not illegal per se but granting options in the money is a cost in terms of accounting rules, which reduces the profit and thus must be booked in the company’s accounts. Besides, in most cases the shareholder-approved stock option plan explicitly requires that options be set at the money. Violating this condition could result in allegations of securities fraud (Forelle et al., 2006). When options are granted at the money, executives can opportunistically manipulate the timing of the transaction in order to inflate the value of the options. For instance, a CEO might withhold good news or release bad news before she/he is awarded options—a practice known as spring loading. The subsequent fall in the share price will guarantee a lower strike price for the person receiving the options. Subsequently, a CEO might release positive information right before exercising her/his options. Although these forms of timing options/manipulating news announcements would be considered illegal, and analogous to insider trading if the person was to purchase or sell shares, it is viewed as legal in the granting or selling of options, though many regard it as amoral. An alternative method used to inflate the value of stock options is backdating. Backdating occurs when managers choose a past date as the grant
until 2006, when the FAS123R changed the reporting rules requiring companies to report the fair value of the stock options in the proxy statement.
http://dx.doi.org/10.1016/j.qref.2015.07.004 1062-9769/© 2015 The Board of Trustees of the University of Illinois. Published by Elsevier B.V. All rights reserved.
Please cite this article in press as: Bianchi, G. Stock options: From backdating to spring loading. The Quarterly Review of Economics and Finance (2015), http://dx.doi.org/10.1016/j.qref.2015.07.004
G Model QUAECO-866; No. of Pages 7
ARTICLE IN PRESS G. Bianchi / The Quarterly Review of Economics and Finance xxx (2015) xxx–xxx
0
Percent of Total CEO Pay by Year 20 40 100 60 80
2
96 97 98 99 00 01 02 03 04 05 06 07 08 09 10 11 Salary Options LTIP Others
Bonus Stock Deferred Non Equity Incentive
Fig. 1. Percentage of total CEO pay by year.
date, i.e. when the share price was particularly low in order to have a favorable strike price. Specifically, officers that backdate options look back through the share price history for a date when the share price was particularly low. Once found, they announce the options as granted at the money that day. As a result, they are able to obtain an option that appears granted at the money when in fact it is awarded in the money. Opportunistic option timing is found to be associated with weaker corporate governance. Indeed, practices such as backdating and spring loading raise governance concerns. When governance is weak, managers have more influence over the board of directors in contracting the terms of their pay (Bebchuk, Grinstein, & Peyer, 2010; Collins, Gong, & Li, 2009). Eventually, the opportunistic option timing casts doubt on the efficacy of incentives to address the principal agent models. Bebchuk and Fried (2004) divide scholars into two contrasting views. In one corner there are those who embrace the so-called “arm’s-length bargaining model” and in the other corner there are those who are in favor of the “managerial power view” (Bebchuk & Fried, 2004). According to the arm’s-length bargaining model, the second best solution is efficient and the board of directors sets optimally managers’ compensation solving the disalignement of interest between shareholders and managers. Scholars that adhere to the managerial power theory sustain instead that the board of directors is not independent from managers and the compensation package is incapable of solving the agency problem. As option timing is more likely to happen under the managerial power model, practices such as backdating and spring loading might shed light into which one of the two models is a better indicator of the compensation of top executives/senior managers. The relative importance of the two opposing views returned to the fore in 2002 when the U.S. Senate introduced the Sarbanes–Oxley Act (SOX) in an attempt to restore the public’s trust in the market, which the Enron debacle – among other corporate and accounting scandals – had destroyed along with millions of dollars of shareholder wealth. Among other countermeasures, the SOX required firms to report to the SEC options granted to company insiders within two business days. If enforced, the reform would have eliminated backdating but not spring loading. Before August 2002, insiders had to file Form 4 and submit it to the SEC “within ten days after the close of the calendar month in which the transaction had occurred” (Brochet, 2010, p. 420). In 2006, the SEC implemented the rules to increase the transparency of granting options. They required top officers to disclose if they were “timing options grants to make them more lucrative to executives and other
employees” (Bickley & Shorter, 2008, p. 15). Firms had to report to the SEC the share price on the grant day, the grant date, if the strike price was lower than the share price as well as justification for having chosen a particular date as a grant date (Bickley and Shorter, 2008). This article explores the impact of the introduction of the Sarbanes–Oxley Act (SOX) in 2002 and its further implementation in 2006 on the options granting process. Consistent with previous literature (Collins et al., 2009; Heron & Lie, 2007; Hossain, Mitra, Rezaee, & Sarath, 2011; Tang, 2014; Lie, 2005), I found that the SOX affected backdating. I show that the introduction of SOX and its further implementation in 2006 effectively reduced and eventually eliminated the practice of backdating options (granting options at time “t” but dating them at some earlier time when the share price was lower). However, in contrast with previous literature, I show that the SOX is also associated with greater “spring loading” of option granting, to time them with analysts’ price target announcements. In particular, I show that springloading options substituted the practice of backdating options in 2002 and that price targets have increasingly played a crucial role in the options granting process. I conclude that the backdating phenomenon ended but it was substituted with the practice of spring loading. The new evidence fits better with the managerial power model than the arm’s length bargaining model. Thus, this study might provide further insight into the current debate regarding managerial compensation. To the best of my knowledge, analysts’ price targets constitute a new indicator to measure spring loading. The remainder of this paper is organized as follows. In the next section, I provide a literature review. In Section 3, I discuss the data and methodology. Section 4 discusses the results. Section 5 is a conclusion.
2. Literature review The economic literature has long investigated the opportunistic timing of executives’ stock options, including the two main techniques: backdating and spring-loading. Yermack (1997) conducted the first study on the opportunistic timing of option granting. Yermack (1997) showed that companies’ share prices tended to drop before options were granted and to increase afterward. Aboody and Kasznik (2000) and Chauvin and Shenoy (2001) argued that these abnormal patterns could be due to a manipulation of information by officers, a practice known as spring loading. Frydman and Jenter (2010) showed that spring loading alone cannot fully explain the abnormal return of option grants nor can “luck” (Lie, 2005). Specifically, Lie (2005) shows that the most likely reason executives have historically made abnormal returns from options is that the grant date was opportunistically chosen retroactively—a process known as backdating. Lie (2005) and Heron and Lie (2007) used the abnormally low prices of shares at the dates when firms grant stock options to show that backdating was frequent in the 1990s. Bebchuk et al. (2010) showed that opportunistic timing is coupled with weaker corporate governance. Similarly, Collins et al. (2009) found that backdating companies “have a higher proportion of inside and gray directors on the board, a higher proportion of outside directors appointed by the incumbent CEO, and a higher incidence of the CEO serving as chair of the board, relative to a control sample of nonbackdating firms” (Collins et al., 405, 2009). After the introduction of the SOX, however, backdating became more difficult (Heron and Lie, 2007; Hossain et al., 2011; Tang, 2014) but was still practiced. In this regard, Collins et al. (2009) revealed that when CEOs’ influence over the board is higher, there is also a higher tendency to violate the SOX’s two-day reporting rule.
Please cite this article in press as: Bianchi, G. Stock options: From backdating to spring loading. The Quarterly Review of Economics and Finance (2015), http://dx.doi.org/10.1016/j.qref.2015.07.004
G Model QUAECO-866; No. of Pages 7
ARTICLE IN PRESS G. Bianchi / The Quarterly Review of Economics and Finance xxx (2015) xxx–xxx
Although, spring loading and backdating are generally recognized as ethically and legally questionable, some scholars propose an alternative point of view arguing that opportunistic timing can be explained via the arm’s-length bargaining model. Accordingly, granting options at or out of the money has tax benefits. In particular, firms can take advantage of section 162 m of the tax code and deduct incentive pay above $1 million as a business expense, but they cannot deduct salaries above $1 million. Thus, the board of directors that opportunistically times stock options would be able to set CEO compensation at an optimal level, while taking advantage of tax benefits and thus saving shareholders from bearing unneeded expenses (Mahmudi & Gao, 2013). According to this view, timing grants is efficient and economically ideal (despite being ethically and legally dubious). In contrast with this interpretation, Bebchuk et al. (2010) provided evidence that gains from opportunistically-timed options are not a substitute for other forms of compensation (like salaries) and that “lucky” CEOs simply tend to have higher compensation. My study contributes to the ongoing debate about the opportunistic timing of executive stock options. First of all, I show that the SOX and its further implementation in 2006 ended this form of option granting. I then examine the extent to which firms substituted backdating with spring loading strategies, i.e. strategically timing their options granting around price target announcements, in order to ensure that executives profited from options. These results provide insight into the relative importance of the two main models (the arm’s-length bargaining model and the managerial power model) in explaining executive compensation and suggest that the new evidence fits better with the managerial power view. 3. Data and methodology To obtain detailed data on stock options granted to executives I used the Thomson–Reuters data set, and in particular the Insider Filing Data Feed (IFDF) Table II, which collects details on stock options and other derivatives paid to executives that firms report to the Securities and Exchange Commission. From this table, I obtain the official grant date, the exercise date, the expiry date and the strike price of stock options granted to corporate ‘insiders’, which allows me to infer if the options were backdated or if they were granted at the money, out of the money, or in the money. In particular, I obtain the daily stock price over a window of 41 days (the grant date plus 20 days before and after3 ) from CRSP. I define an option as backdated if the stock’s closing price on the grant date is equal to the minimum over the 41-day window. That is, backdated options are those set at the minimum of a share price valley. I focus on transactions involving CEOs and stock options granted at the money for fiscal years from 1996 to 2011. In Appendix A, I provide details regarding the methodology adopted to construct my dataset. I obtained price target data from the IBES Detail History Price Target file. Price targets are defined as “the projected price level[s] forecasted by an analyst within a specific time horizon” (Glushkov, 2009, p. 6). The IBES records the following variables: analysts’ price targets, the name of the analyst, the company he/she works for, company for which he/she issues the target price, the horizonperiod, the day the price target was announced and when it became active in the IBES data file, the company currency and whether or not the company is a U.S. firm. IBES has reported price targets since March 1999. I retain only price targets expressed in USD and I only take into account those that are forecasted for a 12-month horizon, which constitutes more than 90% of the entire dataset. First, I
3 I look at the stock’s closing price 20 days before and 20 days after, not trading days. Thus, for non-working days I have missing data.
3
Table 1 Characteristics of options granted, 1996–2011. Year
Firms
Transactions
Backdated
%
Spring loaded
%
1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011
792 1093 1262 1395 1533 1797 1767 1837 1888 1848 1621 1673 1768 1636 1238 1436
1419 2283 2892 3873 5789 15,824 17,747 18,930 21,606 19,260 17,914 19,630 21,412 18,184 13,014 16,637
152 130 208 220 470 1582 1068 873 958 834 427 551 524 558 317 273
11 6 7 6 8 10 6 5 4 4 2 3 2 3 2 2
– – – 473 1366 3346 4854 5294 7056 6375 5999 6459 8068 7247 5792 6955
– – – 12 24 21 27 28 33 33 33 33 38 40 45 42
merge the Detail History price target file with CRSP. I am thus able to assign a PERMNO code for approximately 80% of observations. Then, I merge IFDF information with the IBES Detail History Price Target file. For every option granted, I determine whether IBES reports a price target announcement seven days before or after the grant, i.e. a 14-day window4 . If more than one announcement was made within this window, I use the announcement closest to the grant date. Specifically, I look to see if any price target was announced at the grant date, if not, I look to see if a price target was announced one day before or after the grant date. In some cases, price targets were announced both one day before and one day after the options were granted5 : I retain as a price target the announcement made before the option was granted. This method allows me to work under the null hypothesis of an ideal board (that grants options when the information is disclosed and available to the market). Besides, this approach favors disclosed – rather than undisclosed – information. If no price target was announced at the grant date or one day before/after I proceed as before. Namely, I look to see if any price target was announced two days before/after. . .and so on. I report the announced price target and the announcement date. I create a dummy (SL) taking the value of one if the option was granted seven days before or seven days after or at the announcement date and zero otherwise. SL is meant to identify the spring loading phenomenon. Spring loading occurs when options are granted at specific moments when pertinent information is disclosed. In this article, we argue that granting options around price target announcements is a possible way (i.e. by using analysts’ information) to inflate the value of options. In fact, as we will prove in Appendix B, price target announcements contain reliable information. Thus, after a positive (negative) announcement, the share price tends to go up (or down), and CEOs might benefit from such disclosures. Indeed, if a manager receives an option at the money before a positive announcement, she/he will most likely have an option in the money after the announcement and vice-versa (if a negative announcement is about to be announced, a CEO might wait for the announcement to be made before receiving the option). Table 1 summarizes statistics for the granting of options to CEOs from the IFDF Table II. The first three columns show the calendar year of the transaction, the number of firms in the data set, and the number of transactions. The remaining columns show the characteristics of the options granted. Column 4 reports the number of transactions
4 Again, I look at announcements that occurred days before and after and not working days. In Appendix C I provide evidence for a window of 0 days. 5 Which would imply that the chance that an option would be granted at the lowest value period within the option is 1/41 (2.44%).
Please cite this article in press as: Bianchi, G. Stock options: From backdating to spring loading. The Quarterly Review of Economics and Finance (2015), http://dx.doi.org/10.1016/j.qref.2015.07.004
G Model QUAECO-866; No. of Pages 7
ARTICLE IN PRESS G. Bianchi / The Quarterly Review of Economics and Finance xxx (2015) xxx–xxx
4
that I infer were backdated. I explain in detail how I estimated the number of backdated options in Appendix A. The basic idea is that whenever firms report an excessively large number of options on days when the share price is especially low, they are likely to be backdating. If, for instance, firms have the possibility of granting options on any one of three particular days, it is reasonable to expect that one-third will be granted each day. If all the options are granted on the day with the lowest share price, that would be a sign of potential backdating. Column 4 shows a drop in inferred backdating after 2002. The most likely reason for this is the implementation of the Sarbanes–Oxley Act of 2002, which forced companies to notify, within two business days, any change in ownership of officers’ options. There is another drop in the percentage of backdated options in 2006 and 2007. The likely reason for this is that in 2006 the SEC strengthened the 2-day rule in an attempt to further reduce backdating (Bickley and Shorter, 2008). In my analysis using a window of 41 days6 , which are on average 28 trading days, the chance that an option would be granted at the lowest value period is 1/28 (3.6%). The average “% estimated backdating” from 2006 to 2010 is 2.33%, so I conclude that backdating has essentially ended.I found that approximately 76% of price targets reported by analysts in my IBES subsample foresee an increase in the share price (greater than or equal to zero). Once the IBES subsample is crossed with Thomson IFDF, I find that the number of positive price target announcements around stock awards is about 82%.To isolate the effect of the change in the law on the relationship between price target announcements and granting options, I generated a dummy variable called “SOX” that assumes the value of 0 if the transaction occurred before 2002 and 1 if it was granted during 2002 or later. This variable is designed to detect t the effect of change in the reporting rules pursuant to the new law that went into effect in 2002. Likewise, I also create a variable I have named “SEC” that takes the value of the unity if the options were granted after 2006, when the SEC strengthened the SOX criteria. 4. Results This section tests the effect of the change in reporting rules in preventing backdating in the options granting process, and if this had an unanticipated effect in reinforcing spring loading. The evidence tells a clear story. The introduction of the SOX in 2002 and its further implementation in 2006 reduced backdating. Before 2002, boards frequently backdated options by strategically picking a grant date in the past when the share price was particularly low. This retroactive strategy allowed CEOs to report a favorably low strike price, and in doing so, inflate the option’s value. After 2002, when the SEC required that insiders report the options they were granted within two days, backdating disappeared. But firms found another way to “inflate” the value of executives’ options by using analysts’ price targets, a forward-looking strategy. 4.1. Multivariate analysis To estimate the effect of the change in the law on optiongranting behavior, I run the OLS regression with fixed effects. In the panel database, I use the PERMNO code as the company identifier (id) and the transaction date as a time variable. I adopt the following specification: DEP VARt,i = ˛ + ˇ1 SOXt,i + ˇ2 SECt,i + ˇ3 Xt,i + ıyr∗t + i + εi,t (1)
6 I report a missing value if no transaction took place over the course of a specific day (e.g. when the stock exchange was closed).
where DEP VARi,t stands for dependent variables. The dependent variables are set as follows: • BD is a dummy taking the value of one if options were backdated or zero otherwise. An option is defined as backdated if the share price at the grant date is the lowest in a window of 40 days7 . As described in Appendix A, I define the transaction share price (and day) as the one that minimizes in absolute value the difference between the strike price and the share’s closing price at the reported transaction date or within two days before the transaction (because the SOX required transactions to be reported to the SEC within two business days)8 . • PR is a continuous variable measuring the probability that an option was backdated (or spring-loaded). To obtain it, I construct a symmetric window of 20 days before and after the transaction. For each of these 40 days, I record either the share price or a missing value if no transaction took place (i.e. when the stock exchange was closed). I count how many times the share price at the grant date is actually equal to or less than any share price reported in the 40-day window9 . For instance, according to the previous definition, backdated options are those set at the minimum of a share price valley. In this case, the number of times the stock’s closing price at the grant date is equal to or less than the stock’s closing price in the window of 40 days (20 days before and 20 days later) is 100%. On average, I expect that the closing price at the grant date will be 50% higher and 50% lower than the share prices of the transactions occurring in the window of 40 days. This variable is meant to detect all those transactions that are not necessarily made at the minimum of a share price valley but still were granted at a favorably lower share price. For instance, a board might decide to set the options at the second-lowest share price in order to appear less suspicious. It is more difficult to use spring loading to grant stock options at the bottom of the V share price valley. A share price that is falling might suddenly increase before falling again (whisk of the tail), so that the CEO/board that intends to spring load might grant the options too early or too late and not catch the exact minimum. Nonetheless, despite the fact that options are not at the lowest share price, the return is still favorable. • SL is a dummy taking the value of one if the option was granted seven days before or seven days after or at the announcement date, zero otherwise. The independent variables are: • SOX is a dummy that assumes the value of 0 if the transaction occurred before 2002 and it takes the value of 1 if it was granted during 2002 or later.
7 I report a missing value if no transaction took place over the course of a specific day (e.g. when the stock exchange was closed). 8 Since in one single day more than one transaction may occur and each transaction may report a different strike price, it might be possible that a transaction is assigned to a day (with its related share price) while another transaction is assigned to a different day. Therefore, it is possible that one or more transactions reported by a company are considered backdated, while others reported by the same company are not considered backdated even if these transactions occur on the same day. For those transactions, I retain the daily average of the true/false variable defining a transaction as backdated. The same issue also emerges on the variable PR and therefore I adopt the same methodology by retaining only the daily average. 9 The procedure I adopt is similar but different to the one described by Edelson and Whisenant (2009) Instead of considering whether the stock’s closing price at the grant date is lower or higher than the stock’s closing price of the symmetric window of trading days around the transaction, the authors ranked the stock’s closing price of a symmetric window around the grant day and test whether the stock price at the ¨ grant date “has equal probability of receiving any rank within the window(Edelson
and Whisenant, 2009, p. 3).
Please cite this article in press as: Bianchi, G. Stock options: From backdating to spring loading. The Quarterly Review of Economics and Finance (2015), http://dx.doi.org/10.1016/j.qref.2015.07.004
G Model
ARTICLE IN PRESS
QUAECO-866; No. of Pages 7
−0.213 (−0.85) −0.00000628 (−0.27) 0.101*** (13.43) Yes Yes −0.504*** (−10.83) 23595
(9) SL
0.186*** (12.40) 0.0926*** (5.76)
5
• SEC is a dummy that takes the value of the unity if the options were granted after 2006, when the SEC strengthened the SOX criteria, and zero otherwise.
(8) SL
0.156*** (10.57) 0.117*** (6.68) −0.00121 (−0.94) −0.160 (−0.61) −0.00000337 (−0.14) 0.101*** (12.83) Yes Yes −0.498*** (−10.27) 22118
(7) SL
0.120*** (8.71) 0.121*** (7.98)
Yes Yes 0.218*** (21.67) 25976 −2.523*** (−16.79) −0.00000404 (−0.52) 0.00159 (0.41) Yes Yes 0.609*** (25.54) 27513
(6) PR
−0.0563*** (−5.25) −0.0532*** (−5.73)
Yes Yes 0.614*** (68.93) 29959 −1.209*** (−9.36) 0.0000122* (1.83) −0.00566* (−1.69) Yes Yes 0.140*** (6.82) 27513
(5) PR (4) PR
N
Yes Yes 0.104*** (13.80) 29959 COMPANY DUMMY YEAR DUMMIES CONSTANT
AT
CSHO
EWRETX
EFFICIENCY
SEC
−0.0370*** (−2.65) −0.0208** (−2.31) −0.000877 (−1.35) −1.069*** (−8.04) 0.00000759 (0.63) −0.00494 (−1.24) Yes Yes 0.120*** (4.49) 22433 −0.0616*** (−6.81) −0.0138* (−1.77)
−0.0610*** (−6.23) −0.0108 (−1.25)
(2) BD (1) BD
(3) BD
−0.0452*** (−2.66) −0.0267** (−2.44) −0.000802 (−1.01) −2.564*** (−15.86) −0.00000156 (−0.11) 0.00550 (1.13) Yes Yes 0.597*** (18.35) 22433
−0.0601*** (−5.27) 0.00269 (0.27)
Xi,t is a vector of observed individual characteristics:
SOX
Table 2 Multivariate analysis of the effects of the law on timing strategy of granting options from 1999 to 2010. Estimates are reported. * , ** , *** indicate significance at 10%, 5% and 1% levels respectively. The t-statistics are reported in parentheses.
G. Bianchi / The Quarterly Review of Economics and Finance xxx (2015) xxx–xxx
• SALE is the natural logarithm of the gross sales or of the amount of billing for regular sales in thousands of dollars. • AT is the natural logarithm of the current assets in millions. • CHSR is the number of shareholders of ordinary/common capital (in thousands of dollars). • EFFICIENCY is the difference between Tobins Q at time t and Tobins Q at time (t − 1), and it is meant to detect the need of restoring efficiency. • EWRETX equally weighted returns, excluding all dividends. • yr* stands for time dummies for each calendar year from 1999 to 2010. • i is firm fixed effect. I summarize the results of the OLS regression with fixed effects in Table 2. Table 2 provides evidence that backdated options (BD) responded negatively to the introduction of SOX in 2002 and the implementation of the rule in 2006. The model also shows that the probability of backdating (PR) fell with the introduction of SOX in 2002 and the implementation of the rule in 2006. On the other hand, the spring loading responds positively to the introduction of the Sorbonne–Oxley Act. One possible explanation for this pattern is the following. Before 2002, managerial-dominated boards backdated options in order to inflate the options’ value. The introduction of the law in 2002 effectively reduced the phenomenon but led CEOs/boards to search for alternative strategies, such as spring loading. In 2006, the introduction of a new law did nothing to discourage the already tested practice of spring loading, so more boards chose this strategy to enrich their executives. The SOX law thus produced the desired effect of eliminating the backdating phenomenon but reinforced the spring loading strategy. Boards and CEOs shifted from backdating to the riskier but still lucrative forward-looking spring loading.
5. Conclusion This article explored how the introduction of the Sarbanes– Oxley Act in 2002 and its further implementation in 2006 affected the supply of options. I show that the change in the law succeeded in reducing the practice of backdating options but it led some boards to substitute the practice of backdating options with a more forward-looking strategy of granting options around analysts’ price target announcements (spring loading). It also showed that price targets have become increasingly linked to the options granting process. The number of options granted independently from analysts’ price target announcements declined sharply from 1996 to 2010. The new evidence fits better with the managerial power model, which assumes that the board and managers aim to maximize executives’ compensation within the constraints imposed by market penalties and social costs than the arm’s length bargaining model, which assumes that pay results from a negotiation between the managers and a board of directors that seeks to maximize shareholders’ wealth (Bebchuk & Grinstein, 2005). In closing, this analysis provides further insights into the ongoing debate concerning the relative importance of the two competing views of the determination of executive compensation.
Please cite this article in press as: Bianchi, G. Stock options: From backdating to spring loading. The Quarterly Review of Economics and Finance (2015), http://dx.doi.org/10.1016/j.qref.2015.07.004
G Model QUAECO-866; No. of Pages 7
ARTICLE IN PRESS G. Bianchi / The Quarterly Review of Economics and Finance xxx (2015) xxx–xxx
6
Appendix A: Thomson financial insider filing database From IFDF Table II, I obtain the official grant date, the exercise date, the date at which options expire and the exercise price of insiders’ option grants. I restrict my analysis to transactions occurring under transaction code A during the period from 1996 to 2011. Transaction code A reports award transactions pursuant to Rule 16b-3(C). I obtained the data for my analysis beginning with 3995,282 transactions reported in the IFDF file. I retain information for CEOs only. This reduces my sample by 3599,078 transactions. I also exclude non-option derivatives (62,890), so that I consider Call Options, Options, Non-Qualifying Stock Options, Employee Stock Options, Directors’ Stock Options and Non-Employee Director Stock Options. I eliminated a small number of transactions due to what appears to be data errors in the sample. I found some observations where the strike price was not reported by IFDF (3970); other observations which gave a strike price before the grant date (811); others which misreported the maturity date (4844) or reported the maturity date before the transaction occurred (30). This left me with 323,659 transactions. I use yearly official companies’ tickers and companies’ CUSIP numbers to match companies from the IFDF database to the databases compiled by the Center for Research in Stock Prices (CRSP). I drop companies that were not matched with CRSP. I therefore eliminated 79,406 transactions. CRSP allows me to assign the PERMNO (permanent security identification, that is unchanged over time) for each firm and the closing price for each transaction. To get the stock option transaction data for my analysis I then proceeded as follows: • If the reported exercise price is equal to the stock’s closing price inferred from CRSP, the grant date is equal to the reported one. If not, I checked the previous two days and selected the grant date that minimized the difference in absolute value between the stock’s closing price and the strike price10 . I focus on options granted at the money, eliminating 27,839 observations. I defined an option at the money if the closing price listed in CRSP matched exactly or was within 10% of the exercise price. That is, the share price at the grant date is between 90% and 110% of the option’s strike price. I estimated that approximately 93% options were granted at the money. • I then obtained the daily stock price over a window of 41 days (the inferred grant date plus 20 days before and after). If no share price was reported (e.g. because the stock exchange was closed), I report a missing value. I defined an option as backdated if the stock’s closing price on the grant date was the lowest over the 41-day window. Appendix B: Do price targets contain extra information? Bilinski, Lyssimachou, and Walker (2013) show that IBES price targets are reliable predictors of future share prices. Womack (1996) studied the market reaction to analysts’ recommendations and found that positive recommendations are associated with positive returns in the three-day window around the announcement. Asquith, Mikhail, and Au (2005) also finds that the market reacts quickly to price target announcements. Establishing the causality between the price target and share price is not straightforward since the factors that cause a positive or negative price target announcement are likely to be the
Table 3 Testing price targets’ effect on share price. Estimates are reported. * , ** , *** indicate significance at 10%, 5% and 1% levels respectively. The t-statistic are reported in parentheses.
ptt
(1) spt
(2) spt
0.789*** (1049.20)
0.0509*** (92.80) 0.949*** (1649.66)
sp(t−3) ptt
COMPANY DUMMY YEAR DUMMIES CONSTANT N
0.0265*** (25.54)
0.0190*** (19.50) 0.159*** (121.22) Yes Yes 0.00364** (2.39) 80165
Yes Yes 0.680*** (233.47) 545503
Yes Yes 0.0168*** (11.30) 309969
Yes Yes 0.00444*** (2.72) 82759
same factors that lead to a rise or a drop of the share price. The key issue, however, is not the underlying cause for the relation but whether the announcement alters market views and serves as the “conduit” for the information. Accordingly, I follow the analyses of Womack (1996) and Asquith et al. (2005) and examine how the market reacted to price target announcements in the IBES data set. I use a difference-in-differences approach, regressing the change in share price on the change in the price target. That is, I take as my dependent variable the stock’s closing price, after a new price target announcement, minus the share price after the previous price target announcement and take as my independent variable the difference between the new price target and the previous price target11 . Table 3 shows the results of my analysis. In column 1, I regress the log of the share price at the announcement date on the natural logarithms of the price target12 . Column 2 shows the results of regressing the natural logarithm of share price on the natural logarithms of the price target and the natural logarithm of the closing share price three days before a new price target is announced. Column 3 regresses the difference of the natural logarithms of the stock’s closing price at the announcement of the price target with the natural logarithms of the share price of the previous stock announcement on the difference of the logarithms of price targets with the log of the previous price target. sp(t−15) is defined as the logarithms of the stock’s closing price at the announcement date less the logarithms of shares’ closing price 15 days before the announcement. Intuitively, columns 3 and 4 test if a change in the price target produces a change in the share price. The results show that as soon as analysts release a price target, which adds new information to a previous price target, the market reacts. The coefficient is positive and significantly different from zero. Therefore, I conclude that a positive price target announcement leads to an increase the share in price.
Appendix C: Robustness In this appendix, I test the robustness of my definition of spring loaded options. In particular, I consider options as spring loaded if they were granted specifically at the announcement date. The
11
Using this methodology, I found that 356 transactions are officially reported in a calendar year, which is different from the inferred calendar year. When using the calendar year I continue to refer to the official reported calendar year.
(4) spt
sp(t−15)
The results hold also assuming that errors are uncorrelated and adopting the
following specification: ln spt,i /spt−1,i 10
(3) spt
= ˛ + ˇ1 ln ptt,i /spt−1,i
+ ∂yr ∗ +i + εt,i
12
I obtain similar results by regressing the log of a stock’s closing price 15 days and 30 days after the announcement respectively on the natural logarithms of a price target.
Please cite this article in press as: Bianchi, G. Stock options: From backdating to spring loading. The Quarterly Review of Economics and Finance (2015), http://dx.doi.org/10.1016/j.qref.2015.07.004
G Model
ARTICLE IN PRESS
QUAECO-866; No. of Pages 7
G. Bianchi / The Quarterly Review of Economics and Finance xxx (2015) xxx–xxx Table 4 Multivariate analysis of the law’s effects on timing strategy of granting options from 1999 to 2010. Estimates are reported. * , ** , *** indicate significance at 10%, 5% and 1% levels respectively. The t-statistic are reported in parentheses.
SOX SEC
(1) SL
(2) SL
(3) SL
0.119*** (8.67) 0.120*** (7.94)
0.156*** (10.58) 0.116*** (6.60) −0.000804 (−0.63) −0.160 (−0.61) −0.00000248 (−0.10) 0.101*** Yes Yes −0.495*** (−10.20) 22118
0.186*** (12.39) 0.0919*** (5.72)
EFFICIENCY EWRETX CSHO AT Company dummy Year dummies Constant N
Yes Yes 0.218*** (21.68) 25976
−0.213 (−0.84) −0.00000536 (−0.23) 0.0996*** Yes Yes −0.497*** (−10.70) 23595
results reported in Table 4 reveal that my findings hold for different definitions of spring loaded options. References Aboody, D., & Kasznik, R. (2000). CEO stock option awards and the timing of corporate voluntary disclosures. Journal of Accounting and Economics, 29(1), 73–100. Asquith, P., Mikhail, M. B., & Au, A. S. (2005). Information content of equity analyst reports. Journal of Financial Economics, 75(2), 245–282. Bebchuk, L., & Fried, J. (2004). Pay without performance: The unfulfilled promise of executive compensation. Cambridge and London: Harvard University Press. Bebchuk, L., & Grinstein, Y. (2005). The growth of executive pay. Oxford Review of Economic Policy, 21(2), 283–303. Bebchuk, L. A., Grinstein, Y., & Peyer, U. (2010). Lucky CEOs and lucky directors. The Journal of Finance, 65(6), 2363–2401.
7
Bickley, J. M., & Shorter L G. (2008). Stock options: The backdating issue. In No. RL33926. Washington, DC: Congress Research Service. Bilinski, P., Lyssimachou, D., & Walker, M. (2013). Target price accuracy: International evidence. The Accounting Review, 88(3), 825–851. Brochet, F. (2010). Information content of insider trades before and after the Sarbanes–Oxley Act. Accounting Review, 85(2), 419–446. Chauvin, K. W., & Shenoy, C. (2001). Stock price decreases prior to executive stock option grants. Journal of Corporate Finance, 7(1), 53–76. Collins, D. W., Gong, G., & Li, H. (2009). Corporate governance and backdating of executive stock options. Contemporary Accounting Research, 26(2), 403–445. Edelson, R., & Whisenant, S. (2009). A study of companies with abnormally favorable patterns of executive stock option grant timing. In Working paper. Houston, TX: University of Houston—C.T. Bauer College of Business, University of Kansas. http://www.learningace.com/doc/2281499/14a9d7290ad2166a92f1e435dead fdf2/a-study-of-companies-with-abnormally-favorable. Eisenhardt, K. M. (1989). Agency theory: An Assessment and review. The Academy of Management Review, 14(1), 57–74. Forelle, C., Bandler, J., & Anders, G. (2006). The perfect payday. Wall Street Journal—Eastern Edition, 247(64), A1–A5 (cover story). Frydman, C., & Jenter, D. (2010). CEO compensation. Annual Review of Financial Economics, 2(1), 75–102. Glushkov, D. (2009). Overview of IBES on WRDS: Research and data issues. In Data manuals. Wharton Research Data Services. University of Pennsylvania. Heron, R. A., & Lie, E. (2007). Does backdating explain the stock price pattern around executive stock option grants? Journal of Financial Economics, 83(2), 271–295. Hossain, M., Mitra, S., Rezaee, Z., & Sarath, B. (2011). Corporate governance and earnings management in the pre- and post-Sarbanes–Oxley Act regimes: evidence from implicated option backdating firms. Journal of Accounting, Auditing & Finance, 26(2), 279–315. Lie, E. (2005). On the timing of CEO stock option awards. Management Science, 51(5), 802–812. Mahmudi, H., & Gao, H. (2013). Backdating executive stock option grants: Is it all agency? In Working paper. Available at: http://www3.ntu.edu.sg/home/hsgao/ Backdate20140416 final.pdf. Tang, H. (2014). Are CEO stock option grants optimal? Evidence from family firms and non-family firms around the Sarbanes–Oxley Act. Review of Quantitative Finance and Accounting, 42(2), 251–292. Womack, K. L. (1996). Do brokerage analysts’ recommendations have investment value? Journal of Finance, 51(1), 137–167. Yermack, D. (1997). Good timing: CEO stock option awards and company news announcements. Journal of Finance, 52(2), 449–476.
Please cite this article in press as: Bianchi, G. Stock options: From backdating to spring loading. The Quarterly Review of Economics and Finance (2015), http://dx.doi.org/10.1016/j.qref.2015.07.004
ARTICLE IN PRESS
QUAECO-866; No. of Pages 7
The Quarterly Review of Economics and Finance xxx (2015) xxx–xxx
Contents lists available at ScienceDirect
The Quarterly Review of Economics and Finance journal homepage: www.elsevier.com/locate/qref
Stock options: From backdating to spring loading Giuliano Bianchi ∗ Ecole hôtelière de Lausanne, HES-SO // University of Applied Sciences Western Switzerland, Route de Cojonnex 18, Lausanne 1000, Switzerland
a r t i c l e
i n f o
Article history: Received 10 June 2014 Received in revised form 11 June 2015 Accepted 23 July 2015 Available online xxx Keywords: CEO pay Stock options Price targets Backdating Spring loading
a b s t r a c t In this article I explore the impact of the introduction of the Sarbanes–Oxley Act (SOX) in 2002 and the Securities and Exchange Commission’s implementation of the Act in 2006 on the options granting process. I show that after the introduction of the SOX and its implementation the practice of backdating options was substituted with the practice of “spring loading” options around analysts’ price targets announcements. © 2015 The Board of Trustees of the University of Illinois. Published by Elsevier B.V. All rights reserved.
1. Introduction Managers are contracted to act in the best interests of shareholders and to maximize shareholders’ wealth. Ideally, managers’ pay is set according to their effort and productivity, a solution known as the first best solution. However, managers’ interests are not fully aligned with those of shareholders. Plus, the board of directors cannot completely monitor managers’ work, and it does not possess all of the information that is available to managers. Thus, managers might engage in rent-seeking activities at the shareholders’ expense. These types of conflicts lead to ‘agency problems’. Under such circumstances, the first best solution is not possible and agency theory suggests adopting the “second best solution”1 . The second best solution consists of sharing part of the company’s ownership with managers in order to realign the interests of the agent with those of the principal. Stock options are among the many financial tools used to link CEOs’ compensation to companies’ performance. Stock options became very popular in the United States, especially in 2000–2001 when they were the largest single component of CEO compensation (as shown in Fig. 12 ).
∗ Tel.: +0041 217 851 456. E-mail address: [email protected] 1 For a comprehensive review of this concept see Eisenhardt (1989). 2 Data on the composition of CEOs compensation by fiscal year are collected from Execucomp. I retain only CEOs data, using the variable CEOANN that indicates whether the officer served as CEO for the entire or most of the fiscal year or not. The estimated value of the stock options granted to CEOs is approximated by Execucomp
Stock options give the option holder the right to buy company stock at a pre-established strike price (also referred to as the exercise price). In the U.S., stock options are usually granted “at the money”, i.e. the strike price is the share price at the grant date. As the Wall Street Journal (Forelle, Bandler, & Anders, 2006) pointed out, granting options “in the money” (i.e. when the strike price is below the share price) is not illegal per se but granting options in the money is a cost in terms of accounting rules, which reduces the profit and thus must be booked in the company’s accounts. Besides, in most cases the shareholder-approved stock option plan explicitly requires that options be set at the money. Violating this condition could result in allegations of securities fraud (Forelle et al., 2006). When options are granted at the money, executives can opportunistically manipulate the timing of the transaction in order to inflate the value of the options. For instance, a CEO might withhold good news or release bad news before she/he is awarded options—a practice known as spring loading. The subsequent fall in the share price will guarantee a lower strike price for the person receiving the options. Subsequently, a CEO might release positive information right before exercising her/his options. Although these forms of timing options/manipulating news announcements would be considered illegal, and analogous to insider trading if the person was to purchase or sell shares, it is viewed as legal in the granting or selling of options, though many regard it as amoral. An alternative method used to inflate the value of stock options is backdating. Backdating occurs when managers choose a past date as the grant
until 2006, when the FAS123R changed the reporting rules requiring companies to report the fair value of the stock options in the proxy statement.
http://dx.doi.org/10.1016/j.qref.2015.07.004 1062-9769/© 2015 The Board of Trustees of the University of Illinois. Published by Elsevier B.V. All rights reserved.
Please cite this article in press as: Bianchi, G. Stock options: From backdating to spring loading. The Quarterly Review of Economics and Finance (2015), http://dx.doi.org/10.1016/j.qref.2015.07.004
G Model QUAECO-866; No. of Pages 7
ARTICLE IN PRESS G. Bianchi / The Quarterly Review of Economics and Finance xxx (2015) xxx–xxx
0
Percent of Total CEO Pay by Year 20 40 100 60 80
2
96 97 98 99 00 01 02 03 04 05 06 07 08 09 10 11 Salary Options LTIP Others
Bonus Stock Deferred Non Equity Incentive
Fig. 1. Percentage of total CEO pay by year.
date, i.e. when the share price was particularly low in order to have a favorable strike price. Specifically, officers that backdate options look back through the share price history for a date when the share price was particularly low. Once found, they announce the options as granted at the money that day. As a result, they are able to obtain an option that appears granted at the money when in fact it is awarded in the money. Opportunistic option timing is found to be associated with weaker corporate governance. Indeed, practices such as backdating and spring loading raise governance concerns. When governance is weak, managers have more influence over the board of directors in contracting the terms of their pay (Bebchuk, Grinstein, & Peyer, 2010; Collins, Gong, & Li, 2009). Eventually, the opportunistic option timing casts doubt on the efficacy of incentives to address the principal agent models. Bebchuk and Fried (2004) divide scholars into two contrasting views. In one corner there are those who embrace the so-called “arm’s-length bargaining model” and in the other corner there are those who are in favor of the “managerial power view” (Bebchuk & Fried, 2004). According to the arm’s-length bargaining model, the second best solution is efficient and the board of directors sets optimally managers’ compensation solving the disalignement of interest between shareholders and managers. Scholars that adhere to the managerial power theory sustain instead that the board of directors is not independent from managers and the compensation package is incapable of solving the agency problem. As option timing is more likely to happen under the managerial power model, practices such as backdating and spring loading might shed light into which one of the two models is a better indicator of the compensation of top executives/senior managers. The relative importance of the two opposing views returned to the fore in 2002 when the U.S. Senate introduced the Sarbanes–Oxley Act (SOX) in an attempt to restore the public’s trust in the market, which the Enron debacle – among other corporate and accounting scandals – had destroyed along with millions of dollars of shareholder wealth. Among other countermeasures, the SOX required firms to report to the SEC options granted to company insiders within two business days. If enforced, the reform would have eliminated backdating but not spring loading. Before August 2002, insiders had to file Form 4 and submit it to the SEC “within ten days after the close of the calendar month in which the transaction had occurred” (Brochet, 2010, p. 420). In 2006, the SEC implemented the rules to increase the transparency of granting options. They required top officers to disclose if they were “timing options grants to make them more lucrative to executives and other
employees” (Bickley & Shorter, 2008, p. 15). Firms had to report to the SEC the share price on the grant day, the grant date, if the strike price was lower than the share price as well as justification for having chosen a particular date as a grant date (Bickley and Shorter, 2008). This article explores the impact of the introduction of the Sarbanes–Oxley Act (SOX) in 2002 and its further implementation in 2006 on the options granting process. Consistent with previous literature (Collins et al., 2009; Heron & Lie, 2007; Hossain, Mitra, Rezaee, & Sarath, 2011; Tang, 2014; Lie, 2005), I found that the SOX affected backdating. I show that the introduction of SOX and its further implementation in 2006 effectively reduced and eventually eliminated the practice of backdating options (granting options at time “t” but dating them at some earlier time when the share price was lower). However, in contrast with previous literature, I show that the SOX is also associated with greater “spring loading” of option granting, to time them with analysts’ price target announcements. In particular, I show that springloading options substituted the practice of backdating options in 2002 and that price targets have increasingly played a crucial role in the options granting process. I conclude that the backdating phenomenon ended but it was substituted with the practice of spring loading. The new evidence fits better with the managerial power model than the arm’s length bargaining model. Thus, this study might provide further insight into the current debate regarding managerial compensation. To the best of my knowledge, analysts’ price targets constitute a new indicator to measure spring loading. The remainder of this paper is organized as follows. In the next section, I provide a literature review. In Section 3, I discuss the data and methodology. Section 4 discusses the results. Section 5 is a conclusion.
2. Literature review The economic literature has long investigated the opportunistic timing of executives’ stock options, including the two main techniques: backdating and spring-loading. Yermack (1997) conducted the first study on the opportunistic timing of option granting. Yermack (1997) showed that companies’ share prices tended to drop before options were granted and to increase afterward. Aboody and Kasznik (2000) and Chauvin and Shenoy (2001) argued that these abnormal patterns could be due to a manipulation of information by officers, a practice known as spring loading. Frydman and Jenter (2010) showed that spring loading alone cannot fully explain the abnormal return of option grants nor can “luck” (Lie, 2005). Specifically, Lie (2005) shows that the most likely reason executives have historically made abnormal returns from options is that the grant date was opportunistically chosen retroactively—a process known as backdating. Lie (2005) and Heron and Lie (2007) used the abnormally low prices of shares at the dates when firms grant stock options to show that backdating was frequent in the 1990s. Bebchuk et al. (2010) showed that opportunistic timing is coupled with weaker corporate governance. Similarly, Collins et al. (2009) found that backdating companies “have a higher proportion of inside and gray directors on the board, a higher proportion of outside directors appointed by the incumbent CEO, and a higher incidence of the CEO serving as chair of the board, relative to a control sample of nonbackdating firms” (Collins et al., 405, 2009). After the introduction of the SOX, however, backdating became more difficult (Heron and Lie, 2007; Hossain et al., 2011; Tang, 2014) but was still practiced. In this regard, Collins et al. (2009) revealed that when CEOs’ influence over the board is higher, there is also a higher tendency to violate the SOX’s two-day reporting rule.
Please cite this article in press as: Bianchi, G. Stock options: From backdating to spring loading. The Quarterly Review of Economics and Finance (2015), http://dx.doi.org/10.1016/j.qref.2015.07.004
G Model QUAECO-866; No. of Pages 7
ARTICLE IN PRESS G. Bianchi / The Quarterly Review of Economics and Finance xxx (2015) xxx–xxx
Although, spring loading and backdating are generally recognized as ethically and legally questionable, some scholars propose an alternative point of view arguing that opportunistic timing can be explained via the arm’s-length bargaining model. Accordingly, granting options at or out of the money has tax benefits. In particular, firms can take advantage of section 162 m of the tax code and deduct incentive pay above $1 million as a business expense, but they cannot deduct salaries above $1 million. Thus, the board of directors that opportunistically times stock options would be able to set CEO compensation at an optimal level, while taking advantage of tax benefits and thus saving shareholders from bearing unneeded expenses (Mahmudi & Gao, 2013). According to this view, timing grants is efficient and economically ideal (despite being ethically and legally dubious). In contrast with this interpretation, Bebchuk et al. (2010) provided evidence that gains from opportunistically-timed options are not a substitute for other forms of compensation (like salaries) and that “lucky” CEOs simply tend to have higher compensation. My study contributes to the ongoing debate about the opportunistic timing of executive stock options. First of all, I show that the SOX and its further implementation in 2006 ended this form of option granting. I then examine the extent to which firms substituted backdating with spring loading strategies, i.e. strategically timing their options granting around price target announcements, in order to ensure that executives profited from options. These results provide insight into the relative importance of the two main models (the arm’s-length bargaining model and the managerial power model) in explaining executive compensation and suggest that the new evidence fits better with the managerial power view. 3. Data and methodology To obtain detailed data on stock options granted to executives I used the Thomson–Reuters data set, and in particular the Insider Filing Data Feed (IFDF) Table II, which collects details on stock options and other derivatives paid to executives that firms report to the Securities and Exchange Commission. From this table, I obtain the official grant date, the exercise date, the expiry date and the strike price of stock options granted to corporate ‘insiders’, which allows me to infer if the options were backdated or if they were granted at the money, out of the money, or in the money. In particular, I obtain the daily stock price over a window of 41 days (the grant date plus 20 days before and after3 ) from CRSP. I define an option as backdated if the stock’s closing price on the grant date is equal to the minimum over the 41-day window. That is, backdated options are those set at the minimum of a share price valley. I focus on transactions involving CEOs and stock options granted at the money for fiscal years from 1996 to 2011. In Appendix A, I provide details regarding the methodology adopted to construct my dataset. I obtained price target data from the IBES Detail History Price Target file. Price targets are defined as “the projected price level[s] forecasted by an analyst within a specific time horizon” (Glushkov, 2009, p. 6). The IBES records the following variables: analysts’ price targets, the name of the analyst, the company he/she works for, company for which he/she issues the target price, the horizonperiod, the day the price target was announced and when it became active in the IBES data file, the company currency and whether or not the company is a U.S. firm. IBES has reported price targets since March 1999. I retain only price targets expressed in USD and I only take into account those that are forecasted for a 12-month horizon, which constitutes more than 90% of the entire dataset. First, I
3 I look at the stock’s closing price 20 days before and 20 days after, not trading days. Thus, for non-working days I have missing data.
3
Table 1 Characteristics of options granted, 1996–2011. Year
Firms
Transactions
Backdated
%
Spring loaded
%
1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011
792 1093 1262 1395 1533 1797 1767 1837 1888 1848 1621 1673 1768 1636 1238 1436
1419 2283 2892 3873 5789 15,824 17,747 18,930 21,606 19,260 17,914 19,630 21,412 18,184 13,014 16,637
152 130 208 220 470 1582 1068 873 958 834 427 551 524 558 317 273
11 6 7 6 8 10 6 5 4 4 2 3 2 3 2 2
– – – 473 1366 3346 4854 5294 7056 6375 5999 6459 8068 7247 5792 6955
– – – 12 24 21 27 28 33 33 33 33 38 40 45 42
merge the Detail History price target file with CRSP. I am thus able to assign a PERMNO code for approximately 80% of observations. Then, I merge IFDF information with the IBES Detail History Price Target file. For every option granted, I determine whether IBES reports a price target announcement seven days before or after the grant, i.e. a 14-day window4 . If more than one announcement was made within this window, I use the announcement closest to the grant date. Specifically, I look to see if any price target was announced at the grant date, if not, I look to see if a price target was announced one day before or after the grant date. In some cases, price targets were announced both one day before and one day after the options were granted5 : I retain as a price target the announcement made before the option was granted. This method allows me to work under the null hypothesis of an ideal board (that grants options when the information is disclosed and available to the market). Besides, this approach favors disclosed – rather than undisclosed – information. If no price target was announced at the grant date or one day before/after I proceed as before. Namely, I look to see if any price target was announced two days before/after. . .and so on. I report the announced price target and the announcement date. I create a dummy (SL) taking the value of one if the option was granted seven days before or seven days after or at the announcement date and zero otherwise. SL is meant to identify the spring loading phenomenon. Spring loading occurs when options are granted at specific moments when pertinent information is disclosed. In this article, we argue that granting options around price target announcements is a possible way (i.e. by using analysts’ information) to inflate the value of options. In fact, as we will prove in Appendix B, price target announcements contain reliable information. Thus, after a positive (negative) announcement, the share price tends to go up (or down), and CEOs might benefit from such disclosures. Indeed, if a manager receives an option at the money before a positive announcement, she/he will most likely have an option in the money after the announcement and vice-versa (if a negative announcement is about to be announced, a CEO might wait for the announcement to be made before receiving the option). Table 1 summarizes statistics for the granting of options to CEOs from the IFDF Table II. The first three columns show the calendar year of the transaction, the number of firms in the data set, and the number of transactions. The remaining columns show the characteristics of the options granted. Column 4 reports the number of transactions
4 Again, I look at announcements that occurred days before and after and not working days. In Appendix C I provide evidence for a window of 0 days. 5 Which would imply that the chance that an option would be granted at the lowest value period within the option is 1/41 (2.44%).
Please cite this article in press as: Bianchi, G. Stock options: From backdating to spring loading. The Quarterly Review of Economics and Finance (2015), http://dx.doi.org/10.1016/j.qref.2015.07.004
G Model QUAECO-866; No. of Pages 7
ARTICLE IN PRESS G. Bianchi / The Quarterly Review of Economics and Finance xxx (2015) xxx–xxx
4
that I infer were backdated. I explain in detail how I estimated the number of backdated options in Appendix A. The basic idea is that whenever firms report an excessively large number of options on days when the share price is especially low, they are likely to be backdating. If, for instance, firms have the possibility of granting options on any one of three particular days, it is reasonable to expect that one-third will be granted each day. If all the options are granted on the day with the lowest share price, that would be a sign of potential backdating. Column 4 shows a drop in inferred backdating after 2002. The most likely reason for this is the implementation of the Sarbanes–Oxley Act of 2002, which forced companies to notify, within two business days, any change in ownership of officers’ options. There is another drop in the percentage of backdated options in 2006 and 2007. The likely reason for this is that in 2006 the SEC strengthened the 2-day rule in an attempt to further reduce backdating (Bickley and Shorter, 2008). In my analysis using a window of 41 days6 , which are on average 28 trading days, the chance that an option would be granted at the lowest value period is 1/28 (3.6%). The average “% estimated backdating” from 2006 to 2010 is 2.33%, so I conclude that backdating has essentially ended.I found that approximately 76% of price targets reported by analysts in my IBES subsample foresee an increase in the share price (greater than or equal to zero). Once the IBES subsample is crossed with Thomson IFDF, I find that the number of positive price target announcements around stock awards is about 82%.To isolate the effect of the change in the law on the relationship between price target announcements and granting options, I generated a dummy variable called “SOX” that assumes the value of 0 if the transaction occurred before 2002 and 1 if it was granted during 2002 or later. This variable is designed to detect t the effect of change in the reporting rules pursuant to the new law that went into effect in 2002. Likewise, I also create a variable I have named “SEC” that takes the value of the unity if the options were granted after 2006, when the SEC strengthened the SOX criteria. 4. Results This section tests the effect of the change in reporting rules in preventing backdating in the options granting process, and if this had an unanticipated effect in reinforcing spring loading. The evidence tells a clear story. The introduction of the SOX in 2002 and its further implementation in 2006 reduced backdating. Before 2002, boards frequently backdated options by strategically picking a grant date in the past when the share price was particularly low. This retroactive strategy allowed CEOs to report a favorably low strike price, and in doing so, inflate the option’s value. After 2002, when the SEC required that insiders report the options they were granted within two days, backdating disappeared. But firms found another way to “inflate” the value of executives’ options by using analysts’ price targets, a forward-looking strategy. 4.1. Multivariate analysis To estimate the effect of the change in the law on optiongranting behavior, I run the OLS regression with fixed effects. In the panel database, I use the PERMNO code as the company identifier (id) and the transaction date as a time variable. I adopt the following specification: DEP VARt,i = ˛ + ˇ1 SOXt,i + ˇ2 SECt,i + ˇ3 Xt,i + ıyr∗t + i + εi,t (1)
6 I report a missing value if no transaction took place over the course of a specific day (e.g. when the stock exchange was closed).
where DEP VARi,t stands for dependent variables. The dependent variables are set as follows: • BD is a dummy taking the value of one if options were backdated or zero otherwise. An option is defined as backdated if the share price at the grant date is the lowest in a window of 40 days7 . As described in Appendix A, I define the transaction share price (and day) as the one that minimizes in absolute value the difference between the strike price and the share’s closing price at the reported transaction date or within two days before the transaction (because the SOX required transactions to be reported to the SEC within two business days)8 . • PR is a continuous variable measuring the probability that an option was backdated (or spring-loaded). To obtain it, I construct a symmetric window of 20 days before and after the transaction. For each of these 40 days, I record either the share price or a missing value if no transaction took place (i.e. when the stock exchange was closed). I count how many times the share price at the grant date is actually equal to or less than any share price reported in the 40-day window9 . For instance, according to the previous definition, backdated options are those set at the minimum of a share price valley. In this case, the number of times the stock’s closing price at the grant date is equal to or less than the stock’s closing price in the window of 40 days (20 days before and 20 days later) is 100%. On average, I expect that the closing price at the grant date will be 50% higher and 50% lower than the share prices of the transactions occurring in the window of 40 days. This variable is meant to detect all those transactions that are not necessarily made at the minimum of a share price valley but still were granted at a favorably lower share price. For instance, a board might decide to set the options at the second-lowest share price in order to appear less suspicious. It is more difficult to use spring loading to grant stock options at the bottom of the V share price valley. A share price that is falling might suddenly increase before falling again (whisk of the tail), so that the CEO/board that intends to spring load might grant the options too early or too late and not catch the exact minimum. Nonetheless, despite the fact that options are not at the lowest share price, the return is still favorable. • SL is a dummy taking the value of one if the option was granted seven days before or seven days after or at the announcement date, zero otherwise. The independent variables are: • SOX is a dummy that assumes the value of 0 if the transaction occurred before 2002 and it takes the value of 1 if it was granted during 2002 or later.
7 I report a missing value if no transaction took place over the course of a specific day (e.g. when the stock exchange was closed). 8 Since in one single day more than one transaction may occur and each transaction may report a different strike price, it might be possible that a transaction is assigned to a day (with its related share price) while another transaction is assigned to a different day. Therefore, it is possible that one or more transactions reported by a company are considered backdated, while others reported by the same company are not considered backdated even if these transactions occur on the same day. For those transactions, I retain the daily average of the true/false variable defining a transaction as backdated. The same issue also emerges on the variable PR and therefore I adopt the same methodology by retaining only the daily average. 9 The procedure I adopt is similar but different to the one described by Edelson and Whisenant (2009) Instead of considering whether the stock’s closing price at the grant date is lower or higher than the stock’s closing price of the symmetric window of trading days around the transaction, the authors ranked the stock’s closing price of a symmetric window around the grant day and test whether the stock price at the ¨ grant date “has equal probability of receiving any rank within the window(Edelson
and Whisenant, 2009, p. 3).
Please cite this article in press as: Bianchi, G. Stock options: From backdating to spring loading. The Quarterly Review of Economics and Finance (2015), http://dx.doi.org/10.1016/j.qref.2015.07.004
G Model
ARTICLE IN PRESS
QUAECO-866; No. of Pages 7
−0.213 (−0.85) −0.00000628 (−0.27) 0.101*** (13.43) Yes Yes −0.504*** (−10.83) 23595
(9) SL
0.186*** (12.40) 0.0926*** (5.76)
5
• SEC is a dummy that takes the value of the unity if the options were granted after 2006, when the SEC strengthened the SOX criteria, and zero otherwise.
(8) SL
0.156*** (10.57) 0.117*** (6.68) −0.00121 (−0.94) −0.160 (−0.61) −0.00000337 (−0.14) 0.101*** (12.83) Yes Yes −0.498*** (−10.27) 22118
(7) SL
0.120*** (8.71) 0.121*** (7.98)
Yes Yes 0.218*** (21.67) 25976 −2.523*** (−16.79) −0.00000404 (−0.52) 0.00159 (0.41) Yes Yes 0.609*** (25.54) 27513
(6) PR
−0.0563*** (−5.25) −0.0532*** (−5.73)
Yes Yes 0.614*** (68.93) 29959 −1.209*** (−9.36) 0.0000122* (1.83) −0.00566* (−1.69) Yes Yes 0.140*** (6.82) 27513
(5) PR (4) PR
N
Yes Yes 0.104*** (13.80) 29959 COMPANY DUMMY YEAR DUMMIES CONSTANT
AT
CSHO
EWRETX
EFFICIENCY
SEC
−0.0370*** (−2.65) −0.0208** (−2.31) −0.000877 (−1.35) −1.069*** (−8.04) 0.00000759 (0.63) −0.00494 (−1.24) Yes Yes 0.120*** (4.49) 22433 −0.0616*** (−6.81) −0.0138* (−1.77)
−0.0610*** (−6.23) −0.0108 (−1.25)
(2) BD (1) BD
(3) BD
−0.0452*** (−2.66) −0.0267** (−2.44) −0.000802 (−1.01) −2.564*** (−15.86) −0.00000156 (−0.11) 0.00550 (1.13) Yes Yes 0.597*** (18.35) 22433
−0.0601*** (−5.27) 0.00269 (0.27)
Xi,t is a vector of observed individual characteristics:
SOX
Table 2 Multivariate analysis of the effects of the law on timing strategy of granting options from 1999 to 2010. Estimates are reported. * , ** , *** indicate significance at 10%, 5% and 1% levels respectively. The t-statistics are reported in parentheses.
G. Bianchi / The Quarterly Review of Economics and Finance xxx (2015) xxx–xxx
• SALE is the natural logarithm of the gross sales or of the amount of billing for regular sales in thousands of dollars. • AT is the natural logarithm of the current assets in millions. • CHSR is the number of shareholders of ordinary/common capital (in thousands of dollars). • EFFICIENCY is the difference between Tobins Q at time t and Tobins Q at time (t − 1), and it is meant to detect the need of restoring efficiency. • EWRETX equally weighted returns, excluding all dividends. • yr* stands for time dummies for each calendar year from 1999 to 2010. • i is firm fixed effect. I summarize the results of the OLS regression with fixed effects in Table 2. Table 2 provides evidence that backdated options (BD) responded negatively to the introduction of SOX in 2002 and the implementation of the rule in 2006. The model also shows that the probability of backdating (PR) fell with the introduction of SOX in 2002 and the implementation of the rule in 2006. On the other hand, the spring loading responds positively to the introduction of the Sorbonne–Oxley Act. One possible explanation for this pattern is the following. Before 2002, managerial-dominated boards backdated options in order to inflate the options’ value. The introduction of the law in 2002 effectively reduced the phenomenon but led CEOs/boards to search for alternative strategies, such as spring loading. In 2006, the introduction of a new law did nothing to discourage the already tested practice of spring loading, so more boards chose this strategy to enrich their executives. The SOX law thus produced the desired effect of eliminating the backdating phenomenon but reinforced the spring loading strategy. Boards and CEOs shifted from backdating to the riskier but still lucrative forward-looking spring loading.
5. Conclusion This article explored how the introduction of the Sarbanes– Oxley Act in 2002 and its further implementation in 2006 affected the supply of options. I show that the change in the law succeeded in reducing the practice of backdating options but it led some boards to substitute the practice of backdating options with a more forward-looking strategy of granting options around analysts’ price target announcements (spring loading). It also showed that price targets have become increasingly linked to the options granting process. The number of options granted independently from analysts’ price target announcements declined sharply from 1996 to 2010. The new evidence fits better with the managerial power model, which assumes that the board and managers aim to maximize executives’ compensation within the constraints imposed by market penalties and social costs than the arm’s length bargaining model, which assumes that pay results from a negotiation between the managers and a board of directors that seeks to maximize shareholders’ wealth (Bebchuk & Grinstein, 2005). In closing, this analysis provides further insights into the ongoing debate concerning the relative importance of the two competing views of the determination of executive compensation.
Please cite this article in press as: Bianchi, G. Stock options: From backdating to spring loading. The Quarterly Review of Economics and Finance (2015), http://dx.doi.org/10.1016/j.qref.2015.07.004
G Model QUAECO-866; No. of Pages 7
ARTICLE IN PRESS G. Bianchi / The Quarterly Review of Economics and Finance xxx (2015) xxx–xxx
6
Appendix A: Thomson financial insider filing database From IFDF Table II, I obtain the official grant date, the exercise date, the date at which options expire and the exercise price of insiders’ option grants. I restrict my analysis to transactions occurring under transaction code A during the period from 1996 to 2011. Transaction code A reports award transactions pursuant to Rule 16b-3(C). I obtained the data for my analysis beginning with 3995,282 transactions reported in the IFDF file. I retain information for CEOs only. This reduces my sample by 3599,078 transactions. I also exclude non-option derivatives (62,890), so that I consider Call Options, Options, Non-Qualifying Stock Options, Employee Stock Options, Directors’ Stock Options and Non-Employee Director Stock Options. I eliminated a small number of transactions due to what appears to be data errors in the sample. I found some observations where the strike price was not reported by IFDF (3970); other observations which gave a strike price before the grant date (811); others which misreported the maturity date (4844) or reported the maturity date before the transaction occurred (30). This left me with 323,659 transactions. I use yearly official companies’ tickers and companies’ CUSIP numbers to match companies from the IFDF database to the databases compiled by the Center for Research in Stock Prices (CRSP). I drop companies that were not matched with CRSP. I therefore eliminated 79,406 transactions. CRSP allows me to assign the PERMNO (permanent security identification, that is unchanged over time) for each firm and the closing price for each transaction. To get the stock option transaction data for my analysis I then proceeded as follows: • If the reported exercise price is equal to the stock’s closing price inferred from CRSP, the grant date is equal to the reported one. If not, I checked the previous two days and selected the grant date that minimized the difference in absolute value between the stock’s closing price and the strike price10 . I focus on options granted at the money, eliminating 27,839 observations. I defined an option at the money if the closing price listed in CRSP matched exactly or was within 10% of the exercise price. That is, the share price at the grant date is between 90% and 110% of the option’s strike price. I estimated that approximately 93% options were granted at the money. • I then obtained the daily stock price over a window of 41 days (the inferred grant date plus 20 days before and after). If no share price was reported (e.g. because the stock exchange was closed), I report a missing value. I defined an option as backdated if the stock’s closing price on the grant date was the lowest over the 41-day window. Appendix B: Do price targets contain extra information? Bilinski, Lyssimachou, and Walker (2013) show that IBES price targets are reliable predictors of future share prices. Womack (1996) studied the market reaction to analysts’ recommendations and found that positive recommendations are associated with positive returns in the three-day window around the announcement. Asquith, Mikhail, and Au (2005) also finds that the market reacts quickly to price target announcements. Establishing the causality between the price target and share price is not straightforward since the factors that cause a positive or negative price target announcement are likely to be the
Table 3 Testing price targets’ effect on share price. Estimates are reported. * , ** , *** indicate significance at 10%, 5% and 1% levels respectively. The t-statistic are reported in parentheses.
ptt
(1) spt
(2) spt
0.789*** (1049.20)
0.0509*** (92.80) 0.949*** (1649.66)
sp(t−3) ptt
COMPANY DUMMY YEAR DUMMIES CONSTANT N
0.0265*** (25.54)
0.0190*** (19.50) 0.159*** (121.22) Yes Yes 0.00364** (2.39) 80165
Yes Yes 0.680*** (233.47) 545503
Yes Yes 0.0168*** (11.30) 309969
Yes Yes 0.00444*** (2.72) 82759
same factors that lead to a rise or a drop of the share price. The key issue, however, is not the underlying cause for the relation but whether the announcement alters market views and serves as the “conduit” for the information. Accordingly, I follow the analyses of Womack (1996) and Asquith et al. (2005) and examine how the market reacted to price target announcements in the IBES data set. I use a difference-in-differences approach, regressing the change in share price on the change in the price target. That is, I take as my dependent variable the stock’s closing price, after a new price target announcement, minus the share price after the previous price target announcement and take as my independent variable the difference between the new price target and the previous price target11 . Table 3 shows the results of my analysis. In column 1, I regress the log of the share price at the announcement date on the natural logarithms of the price target12 . Column 2 shows the results of regressing the natural logarithm of share price on the natural logarithms of the price target and the natural logarithm of the closing share price three days before a new price target is announced. Column 3 regresses the difference of the natural logarithms of the stock’s closing price at the announcement of the price target with the natural logarithms of the share price of the previous stock announcement on the difference of the logarithms of price targets with the log of the previous price target. sp(t−15) is defined as the logarithms of the stock’s closing price at the announcement date less the logarithms of shares’ closing price 15 days before the announcement. Intuitively, columns 3 and 4 test if a change in the price target produces a change in the share price. The results show that as soon as analysts release a price target, which adds new information to a previous price target, the market reacts. The coefficient is positive and significantly different from zero. Therefore, I conclude that a positive price target announcement leads to an increase the share in price.
Appendix C: Robustness In this appendix, I test the robustness of my definition of spring loaded options. In particular, I consider options as spring loaded if they were granted specifically at the announcement date. The
11
Using this methodology, I found that 356 transactions are officially reported in a calendar year, which is different from the inferred calendar year. When using the calendar year I continue to refer to the official reported calendar year.
(4) spt
sp(t−15)
The results hold also assuming that errors are uncorrelated and adopting the
following specification: ln spt,i /spt−1,i 10
(3) spt
= ˛ + ˇ1 ln ptt,i /spt−1,i
+ ∂yr ∗ +i + εt,i
12
I obtain similar results by regressing the log of a stock’s closing price 15 days and 30 days after the announcement respectively on the natural logarithms of a price target.
Please cite this article in press as: Bianchi, G. Stock options: From backdating to spring loading. The Quarterly Review of Economics and Finance (2015), http://dx.doi.org/10.1016/j.qref.2015.07.004
G Model
ARTICLE IN PRESS
QUAECO-866; No. of Pages 7
G. Bianchi / The Quarterly Review of Economics and Finance xxx (2015) xxx–xxx Table 4 Multivariate analysis of the law’s effects on timing strategy of granting options from 1999 to 2010. Estimates are reported. * , ** , *** indicate significance at 10%, 5% and 1% levels respectively. The t-statistic are reported in parentheses.
SOX SEC
(1) SL
(2) SL
(3) SL
0.119*** (8.67) 0.120*** (7.94)
0.156*** (10.58) 0.116*** (6.60) −0.000804 (−0.63) −0.160 (−0.61) −0.00000248 (−0.10) 0.101*** Yes Yes −0.495*** (−10.20) 22118
0.186*** (12.39) 0.0919*** (5.72)
EFFICIENCY EWRETX CSHO AT Company dummy Year dummies Constant N
Yes Yes 0.218*** (21.68) 25976
−0.213 (−0.84) −0.00000536 (−0.23) 0.0996*** Yes Yes −0.497*** (−10.70) 23595
results reported in Table 4 reveal that my findings hold for different definitions of spring loaded options. References Aboody, D., & Kasznik, R. (2000). CEO stock option awards and the timing of corporate voluntary disclosures. Journal of Accounting and Economics, 29(1), 73–100. Asquith, P., Mikhail, M. B., & Au, A. S. (2005). Information content of equity analyst reports. Journal of Financial Economics, 75(2), 245–282. Bebchuk, L., & Fried, J. (2004). Pay without performance: The unfulfilled promise of executive compensation. Cambridge and London: Harvard University Press. Bebchuk, L., & Grinstein, Y. (2005). The growth of executive pay. Oxford Review of Economic Policy, 21(2), 283–303. Bebchuk, L. A., Grinstein, Y., & Peyer, U. (2010). Lucky CEOs and lucky directors. The Journal of Finance, 65(6), 2363–2401.
7
Bickley, J. M., & Shorter L G. (2008). Stock options: The backdating issue. In No. RL33926. Washington, DC: Congress Research Service. Bilinski, P., Lyssimachou, D., & Walker, M. (2013). Target price accuracy: International evidence. The Accounting Review, 88(3), 825–851. Brochet, F. (2010). Information content of insider trades before and after the Sarbanes–Oxley Act. Accounting Review, 85(2), 419–446. Chauvin, K. W., & Shenoy, C. (2001). Stock price decreases prior to executive stock option grants. Journal of Corporate Finance, 7(1), 53–76. Collins, D. W., Gong, G., & Li, H. (2009). Corporate governance and backdating of executive stock options. Contemporary Accounting Research, 26(2), 403–445. Edelson, R., & Whisenant, S. (2009). A study of companies with abnormally favorable patterns of executive stock option grant timing. In Working paper. Houston, TX: University of Houston—C.T. Bauer College of Business, University of Kansas. http://www.learningace.com/doc/2281499/14a9d7290ad2166a92f1e435dead fdf2/a-study-of-companies-with-abnormally-favorable. Eisenhardt, K. M. (1989). Agency theory: An Assessment and review. The Academy of Management Review, 14(1), 57–74. Forelle, C., Bandler, J., & Anders, G. (2006). The perfect payday. Wall Street Journal—Eastern Edition, 247(64), A1–A5 (cover story). Frydman, C., & Jenter, D. (2010). CEO compensation. Annual Review of Financial Economics, 2(1), 75–102. Glushkov, D. (2009). Overview of IBES on WRDS: Research and data issues. In Data manuals. Wharton Research Data Services. University of Pennsylvania. Heron, R. A., & Lie, E. (2007). Does backdating explain the stock price pattern around executive stock option grants? Journal of Financial Economics, 83(2), 271–295. Hossain, M., Mitra, S., Rezaee, Z., & Sarath, B. (2011). Corporate governance and earnings management in the pre- and post-Sarbanes–Oxley Act regimes: evidence from implicated option backdating firms. Journal of Accounting, Auditing & Finance, 26(2), 279–315. Lie, E. (2005). On the timing of CEO stock option awards. Management Science, 51(5), 802–812. Mahmudi, H., & Gao, H. (2013). Backdating executive stock option grants: Is it all agency? In Working paper. Available at: http://www3.ntu.edu.sg/home/hsgao/ Backdate20140416 final.pdf. Tang, H. (2014). Are CEO stock option grants optimal? Evidence from family firms and non-family firms around the Sarbanes–Oxley Act. Review of Quantitative Finance and Accounting, 42(2), 251–292. Womack, K. L. (1996). Do brokerage analysts’ recommendations have investment value? Journal of Finance, 51(1), 137–167. Yermack, D. (1997). Good timing: CEO stock option awards and company news announcements. Journal of Finance, 52(2), 449–476.
Please cite this article in press as: Bianchi, G. Stock options: From backdating to spring loading. The Quarterly Review of Economics and Finance (2015), http://dx.doi.org/10.1016/j.qref.2015.07.004