On the importance of golden parachutes

Transcription

1 On the importance of golden parachutes Eliezer M. Fich, Anh L. Tran, Ralph A. Walkling* June 8 th, 2009 Abstract While the previous literature addresses the existence of golden parachutes, it does not consider their relative importance to CEOs. This importance is critical because it correlates with the moral hazard problem facing the firm s top decision maker. When firms become acquisition targets, their CEO has important influence over the deal s outcome. In these cases, an incorrectly sized parachute relative to other executive compensation could produce either a rush to sale or unyielding resistance despite the acquisition price offered. Either of these attitudes might harm target shareholders. We examine 851 acquisition offers from to investigate whether golden parachutes benefit the executives receiving them, the shareholders in the firms that grant them, or both. Although we do find evidence of incentive alignment and ex-post settling up, our primary results are consistent with rent extraction. JEL classification: D82; G34; J33 Keywords: Golden Parachutes; Acquisitions; Moral Hazard * The authors are, respectively, Associate Professor of Finance, Ph.D. candidate, and Stratakis Chair in Corporate Governance, Center for Corporate Governance. All authors are at the LeBow College of Business, Drexel University, 3141 Chestnut Street, Philadelphia, PA Contact information: , efich@drexel.edu (E.M. Fich); , anh@drexel.edu (A.L. Tran); , RW@drexel.edu (R.A. Walkling)

2 Companies receiving federal aid are going to have to disclose publicly all the perks and luxuries bestowed upon senior executives, and provide an explanation to the taxpayers and to shareholders as to why these expenses are justified. And we're putting a stop to these kinds of massive severance packages we've all read about with disgust; we're taking the air out of golden parachutes. President Barack Obama February 4, Introduction Golden parachutes are more controversial today than when they first appeared over twenty years ago. Advocates argue that parachutes are a necessary part of a competitive pay package required to attract and retain talented executives. Moreover, golden parachutes are beneficial to shareholders since they induce senior managers to do the right thing in the event of an acquisition attempt. Opponents object to parachutes because they are linked to a change in control of a company, not to its continuing performance. Detractors portray parachutes as guaranteeing managers pay-for-failure, regardless of shareholder returns. Headlines from the popular press often criticize golden parachutes and express widespread concern about managerial excess and the lack of pay-for-performance related to parachute payments. Government actions with regard to parachutes mirror the controversy. Recent regulation has paved the way for boards of target firms to either award or augment golden parachutes at the onset of an acquisition. 2 In contrast, companies receiving assistance from the Troubled Assets Relief Program (TARP) are prohibited from making golden parachute payments to their senior executives. 3 Taken together, all of these items suggest that the controversy surrounding golden parachutes is alive and well. In this paper, we study a sample of 851 acquisition offers during to investigate whether golden parachutes benefit the executives receiving them, the shareholders in the firms that grant them, or both. From an academic perspective, these issues are similar to those addressed in the early literature: 1 The full speech by president Obama can be viewed at: 2 On October 18, 2006, the Securities and Exchange Commission (SEC) adopted amendments to Rule 14d-10(a)(2) of the 1934 Securities Act to provide a safe harbor enabling the compensation committee of a target's board of directors to provide employee benefits, severance, golden parachutes or other compensation arrangements for its executives during a tender offer negotiation. These arrangements often include stay bonuses, non-compete payments, and other cash and equity compensation. 3 See: 1

3 incentive alignment, rent extraction, and ex post settling up. 4 Although these hypotheses are well known, our implementation differs from prior work in several important respects. While the previous literature addresses the existence of golden parachutes, it does not consider their relative importance to the CEO. Recognizing this importance is critical for understanding the incentives of golden parachutes: an improper balance in an executive s pay package could produce either a rush to sale or unyielding resistance in the face of an acquisition attempt, regardless of the price offered. Neither of these attitudes is likely to benefit shareholders. The attitude of the target CEO is the most important factor in determining whether a target firm is acquired (see Walkling, 1985, Comment and Schwert, 1995, and Schwert, 2000). Thus, when a firm becomes a takeover target, a moral hazard problem exists: the CEOs have direct influence over actions that provide personal benefit at the possible expense of their shareholders. Because of this moral hazard, it is the relative importance of golden parachutes, not their mere presence, that must be recognized. In addition to recognizing the relative importance of the parachute, we examine inferences drawn from the controversy surrounding these payments. It is argued that golden parachutes can prevent an executive from derailing an acquisition; this has implications for deal negotiations and completion. It is also argued that parachutes can strengthen the CEO s ability to bargain for the firm; this has implications for the premia offered in acquisitions. We expand upon previous work by exploring these topics in the context of the relative importance of the parachute, not just its existence. Our research design consists of two stages. In the first stage, we conduct several tests to study whether the relative importance of parachutes influences components of merger negotiations. Specifically, we examine whether parachutes cause deals (i) to be hostile or friendly, (ii) to involve an auction, (iii) to have acquirer termination fees, and (iv) to be completed or not. All of these elements are important in acquisitions and all can be heavily influenced by the CEO. However, even if parachutes affect merger negotiations, it does not necessarily follow that shareholders benefit. As we have noted, the importance of 4 The incentive alignment and rent extraction hypotheses are often studied in settings susceptible to agency problems; [See, for example, Jensen and Meckling (1976) and Lambert and Larcker (1985)]. The ex post settling up hypothesis is due to Fama (1980). 2

4 the parachute to the CEOs can provide incentives to sell or block an acquisition attempt regardless of price. Consequently, the second stage of our empirical analyses directly examines the wealth effects triggered by golden parachutes. We do this by contrasting actual and expected premia, the latter determined by a baseline regression controlling for firm and deal characteristics. We note that parachutes are endogenously chosen, that is, the decision to offer parachutes has its own determinants. To address this issue, our multivariate tests control for potential endogeneity biases arising from self-selection by using the Heckman (1979) approach. In addition, to control for the possibility of omitted variables bias, our multivariate analyses consist of year- and industry-fixed-effects regressions. 5 The results of our first stage tests reveal that the importance of parachutes relative to measures of target CEO wealth materially affects the way in which mergers are negotiated. We show that as the relative importance of the parachute increases, deals are less likely to be hostile, more likely to involve an auction, likely to have lower acquirer termination fees, and more likely to be completed. These results suggest that the willingness to sell a target increases with the relative importance of golden parachutes. On the surface, these findings appear to support the incentive alignment hypotheses. However, because measurable wealth effects are absent from these negotiation-related tests, the rent extraction alternative cannot be dismissed. Our second stage tests indicate that golden parachutes affect the wealth of target CEOs and target shareholders in a non-trivial manner. We estimate that, on average, parachute payments account for approximately 30.5% of the total merger pay package target CEOs receive. These executives cash in about $4.9 million from golden parachutes when their firms are sold. In contrast, we find that as parachute importance increases, targets earn premia that fall short from expected values. We estimate that a 10% increase in parachute importance relative to the merger pay package is associated with a shortfall in premia of about 3.97 percentage points. To put this result in perspective, this shortfall implies a reduction of $196.7 million in deal value for the average acquisition in our sample. These findings indicate that when CEOs are given stronger incentives to sell their firms, vis-à-vis a larger parachute, they appear to do so even at the detriment of their shareholders. These results are consistent with rent extraction. 5 Hausman and Taylor (1981) argue that the fixed-effects specification provides a common, unbiased method to control for omitted variables in a panel data set. 3

5 We also study the creation and adjustments to parachutes during the merger negotiation period. We find target CEOs earning lower (higher) excess pay in the year prior to the merger are more likely to have their parachutes augmented (reduced) during the merger negotiation period. We also find parachute augmentations more likely for CEOs who expect large pay losses when their firms are acquired. This evidence appears consistent with the idea that altering the importance of golden parachutes works as a way of ex post settling up [Fama (1980)]. We do not want our paper to be viewed as opposing golden parachutes. It is possible that in many cases, parachutes induce managers to pursue deals that are in their shareholders best interest. Moreover, we are silent on the potential importance of parachutes in hiring and retaining highly qualified personnel. However, our study has several implications related to the current debate on golden parachutes. First, we find that the importance of parachutes relative to the wealth of the CEO affects the way in which deals are negotiated and firms are sold. In addition, we find that in some instances, parachute agreements are amended once merger talks begin. These results have repercussions on the disclosure of top management compensation because, in many instances, shareholders do not learn of such amendments until after deals are completed. Second, we show that the relative importance of parachutes impacts takeover premia. Given the moral hazard problem that arises during takeovers, this finding has direct implications for the way topmanagement compensation contracts are structured. Third, because payments under parachute agreements are dispensed to target CEOs even when shareholders appear to lose, such agreements do not always align the incentives of target CEOs and shareholders. Consequently, as a matter of public policy, tying parachute payments more closely to shareholder wealth might not only achieve such incentive alignment, but could also make golden parachutes more palatable to regulators, corporate governance activists, and executive compensation watch groups. The paper proceeds as follows. Section 2 provides a background on golden parachutes and develops our hypotheses. Section 3 describes our sample selection and data. Section 4 investigates the determinants of golden parachutes. Section 5 tests whether parachutes affect merger negotiations. Section 6 examines the 4

6 wealth effects of golden parachutes. Section 7 addresses a number of robustness issues. Section 8 presents our conclusions. 2. Background on golden parachutes and hypotheses development 2.1. Golden parachutes: Nonacademic evidence In recent years, golden parachutes have prompted significant public attention and caused some investors concern. Such concern is understandable since, under golden parachute agreements, executives are assured compensation benefits well into the millions without requiring successful performance in order to collect these awards. A study of 137 large U.S. corporations by Equilar Inc. in 2007 finds that 82% of these firms have golden parachute provisions in place for their chief executives and that the median package is worth $29 million. Recently, the business press documents many incidences of lucrative severance packages for top executives that have rankled investors, irked pension funds, and outraged corporate governance activists. In some high-profile instances, executives such as Carly Fiorina of Hewlett- Packard, Robert Nardelli of Home-Depot, Stan O Neal of Merrill Lynch, and Charles Prince of Citigroup, reportedly cashed in hefty golden parachutes while the companies under their stewardship lost millions of dollars and thousands of workers were laid off. 6 In response to such incidences, law makers have asked regulators to reduce these payments to avoid rich severance packages for failed executives. As a result, under the new rules, U.S. banks and car manufacturers receiving funds from the Troubled Assets Relief Program (TARP) can only give their top executives golden parachute payments up to one time the executives previous cash compensation. This provision lowers the commonly used parachute multiple of three customary in these industries. While these limits will encumbrance state-aid recipients in the financial and automotive sectors, proposals to curb golden parachutes might follow in other industries where pay might be deemed excessive. 7 While some prominent firms (such as Nabors Industries and Textron) oppose such proposals arguing that these 6 Dvorak (2008a) and Herbert (2008). 7 Farrell and Ward (2009). 5

7 payments help keep their compensation at competitive levels, others (such as Colgate-Palmolive, 3M, and Chevron) are capping golden parachute payments for their executives Golden parachutes in acquisitions: Academic evidence A golden parachute is a clause in an executive's employment contract specifying benefits that s/he will receive in the event that the company is acquired and/or the executive's employment is terminated. 9 These benefits, which are provided to reduce perverse incentives such as derailing a profitable acquisition, are often based on the regular cash and bonus paid to the executives. Previous research studies the reasons why firms adopt these plans. Knoeber (1986) and Berkovitch and Khanna (1991) view golden parachutes as implicit deferred compensation, already earned but not yet received, that promotes managerial human capital investment in the firm. Their view is consistent with the ex-post settling up theory of managerial compensation advanced by Fama (1980). 10 Jensen (1988) states that severance packages are used to compensate managers for the loss of their jobs in the event of an employment termination. In this context, he argues that if correctly implemented, golden parachutes may help reduce the conflict of interest between shareholders and managers. Almazan and Suarez (2003) theorize that severance pay may benefit managers and shareholders depending on the quality of board monitoring. Specifically, they argue that under strong boards, optimal severance pay helps protect managers from repeated replacement decisions from shareholders while under weak boards, it protects shareholders by reducing excessive resistance from managers who should be replaced Do parachutes induce/deter takeovers? In the context of corporate acquisitions, one common view is that firms adopt golden parachutes as an anti-takeover protection mechanism. In fact, the presence of a golden parachute is one of the 24 antitakeover provisions tracked by the IRRC and indexed by Gompers, Ishii, and Metrick (2003). While such 8 Dvorak (2008b). 9 Employment might cease under a change in control. Instances may include a consolidation or an acquisition of the firm in which the company is not the continuing or surviving corporation, or a sale, lease, exchange or transfer of all or substantially all the assets of the company. 10 In Fama s model the manager s wage is periodically revised to account for his performance and personal consumption of company resources. Consequently, Fama s model implies that perk consumption, and perhaps excess compensation, motivates managers to work harder and increase firm value. 6

8 provisions may increase a firm s ability to defeat a takeover offer (Malatesta and Walkling, 1988) they could also enhance the target s bargaining position with the bidder (Comment and Schwert, 1995). Lambert and Larcker (1985) find that the market views golden parachute adoptions as a signal of possible takeover attempts. However, the empirical evidence related to the parachutes effect on takeover probability is mixed. Machlin, Choe, and Miles (1993) show that the adoption of a golden parachute is associated with a greater likelihood of a successful takeover. In contrast, Cotter and Zenner (1994) find that potential wealth increase from managers equity ownership, rather than golden parachute payouts, affects the probability of an acquisition Golden parachutes and target shareholder wealth Harris (1990) models the role of golden parachutes in negotiations between target management and bidders. She theorizes that by awarding the manager a golden parachute of the optimal size, target shareholders can maximize their takeover gains. Without such proper incentives, managers may otherwise reject bids that increase shareholder value due to their potential losses in compensation and other executive benefits. The existing empirical evidence supports Harris s predictions. Walkling and Long (1984) empirically show that managers resistance to takeover bids is related to the effect of the acquisition on their personal wealth. Golden parachutes are designed to reduce such resistance. Related to this, Lambert and Larcker (1985) find positive investor reactions upon the announcement of golden parachute adoptions during They interpret this finding as evidence that golden parachutes align the incentives of top managers and shareholders. Perhaps to gauge the parachute s optimal size conjectured by Harris, some empirical papers use relative measures of golden parachutes. For example, Cotter and Zenner (1994) find that the size of golden parachute payments relative to the targets market capitalization is unrelated to the acquisition premia. In contrast, Machlin, Choe, and Miles (1993) find the opposite result. Lambert and Larcker (1985) argue that the parachute size relative to the target size variable likely measures the increased acquisition costs arising from the parachute payment. Nevertheless, it is unclear whether the relative parachute size used in those studies captures the incentives to top-managers receiving these payments. This issue, which we view as an 7

9 empirically open research question, is important because golden parachutes are purportedly aimed at aligning the incentives of managers and shareholders Golden parachutes and target CEO wealth Jensen and Meckling (1976) argue that, as managerial ownership decreases, incentives increase for the manager to deviate from policies that maximize firm value. Almazan and Suarez (2003) develop a model in which entrenched managers obtain higher severance pay and are less subject to pay-performance sensitivity. If the size of golden parachutes is not optimal (Harris, 1990) and monitoring is weak, the target CEOs receiving parachutes may deviate from maximizing the wealth of their shareholders when their firms are sold. Under this view, commonly referred to as the rent extraction hypothesis, golden parachutes provide a vehicle for entrenched managers to expropriate shareholders. Previous empirical research documents that target managers may engage in self-dealing at the expense of the target shareholders. Hartzell, Ofek, and Yermack (2004) study 311 acquisition offers during They show that target CEOs may accept lower takeover premia in lieu of exiting cash bonuses or when their parachutes are augmented prior to the deal s completion. These results could be interpreted as consistent with the rent extraction hypothesis and the inadequacy of the size of golden parachutes. 11 However, Hartzell, Ofek, and Yermack (2004) argue that their results can be consistent with Fama s (1980) ex-post settling up theory if the cash bonuses and other payments help in making target CEOs whole for the benefits they lose when their firms are sold Hypotheses The existing literature studies theories of incentive alignment, ex-post settling up, and rent extraction to explain golden parachutes. We believe that a key element to help disentangle these alternatives is the idea, advanced by Harris (1990), of the optimal size of the parachute. However, as discussed earlier and as the literature summary in Table 1 indicates, existing studies do not capture the effect of golden parachutes relative to the wealth of top managers. Parachutes increase the wealth of a CEO of an acquired firm, but 11 These authors argue that, in some circumstances, the evidence related to the cash bonuses to the CEO as a form of ex-post settling up [Fama (1980)], whereby target CEOs are made whole from the benefits they lose when their firms are acquired. 8

10 other elements of CEO utility (i.e., loss of control, loss of future compensation) are negatively impacted. Because of this imbalance, it is the relative importance of the parachute that should be important. Put differently, relative to their total merger pay package and/or their expected personal wealth losses, the same parachute payment might create different incentives for different executives. Consequently, in this paper, we re-examine the incentive alignment, ex-post settling up, and rent extraction hypotheses of golden parachutes on a recent sample with variables that measure the relative impact of the parachute on managerial wealth. Specifically, our variables measure the importance of the golden parachute relative to the target CEO s personal wealth. As described later, we scale the parachute payment by the value of the merger pay package awarded to the target CEO and by the expected wage losses the executive will incur once the firm is sold, respectively. We believe that our empirical strategy is important in recognizing the balance of incentives inherent in the design of golden parachutes Benchmarking parachute payouts Based on the classic principal-agent problem, Raith (2008) theorizes that the structure of compensation is crucial in determining whether agents will engage in self-dealing or will take actions that benefit principals. He proposes that if agents are paid for input activities they are less likely to use their firm specific knowledge to enhance firm value. Consequently, Raith argues that only the output or performance based portion of the pay package will induce agents to increase the principal s wealth. In the context of this paper, we note that the only requirement to collect golden parachute payments is that an acquisition materializes. Otherwise, a golden parachute is not output-based compensation. In contrast, stock and option ownership is output based. This feature underscores the importance of benchmarking golden parachutes against the entire merger pay package to measure the proportion of the package that is not output-based. Under Raith s theories, we would expect target CEOs with properly structured parachutes and merger pay packages to be associated with deals that enhance their shareholders wealth. We also benchmark the parachute payout against the expected income loss by target CEOs when their firms are sold. The goal of this measure is to gauge how parachutes contribute to make the executives 9

11 whole. In this sense, our measure might capture the target boards effort to ex-post settle up CEO compensation and how such effort affects the targets value. 3. Data and sample characteristics 3.1. Sample selection We begin with a base sample of 4,459 mergers and acquisitions announced during and tracked in the Securities Data Company s (SDC) Merger and Acquisition database. We require the target to be a publicly traded U.S. company and exclude spinoffs, recapitalizations, exchange offers, repurchases, self-tenders, privatizations, acquisitions of remaining interest, partial interests or assets, and transactions without disclosed deal value. From this group, we keep 3,567 deals in which target firms have stock return and accounting data available from the Center for Research in Securities Prices (CRSP) and Compustat, respectively. We lose 288 deals because the premium data are missing from SDC and from other sources such as CRSP, LexisNexis, or Factiva. We require corporate governance data for target firms from RiskMetrics (formerly the Investor Responsibility Research Center). After filtering out deals in which governance data for target firms are not available, our final sample consists of 851 offers Target and deal characteristics We report the offer characteristics for our sample in Panel A of Table 2. Among the 851 transactions, 155 (or about 18%) are tender offers and 59 (almost 7%) are hostile takeovers. These statistics compare favorably to those in Officer (2003). His sample of acquisitions during consists of about 20% tender offers and 8% hostile deals. Similar to Moeller, Schlingemann and Stulz (2005), 55% of the transactions in our sample are paid in cash. We also read proxies including S-4, DEFM14A, SC-TO, and DEF14A filed with the Securities and Exchange Commission (SEC) by the target and/or acquiring firms to obtain information on the sale procedure, the party that initiates the deal, and the time of initiation. We find that 34% of the transactions are conducted using auctions and in over 39% of all deals the target firm initiates the sale. Deals in our sample have a completion rate of 88%, which is similar to that in Song and Walkling (2007). They report a completion rate of 86% in their sample of acquisitions during

12 Panel B of Table 2 contains key financial characteristics of the target firms in our sample. For these companies, the average (median) market value of target equity is $3.302 billion ($0.991 billion) and leverage accounts for 26% (25%) of the total assets. These statistics are comparable to those of Boone and Mulherin (2007) who report a mean market capitalization of $2.7 billion and Bates and Lemmon (2003) who report an average leverage of 23.3%. In addition, targets in our sample have a median market-to-book equity ratio of 1.42, which is close to the ratio of 1.69 reported by Officer (2003). Grinstein and Hribar (2004) report a mean deal value of $4.7 billion for transactions which is similar to the $4.8 billion value in our sample Target CEO characteristics In Panel C of Table 2, we report the target CEO s characteristics. On average, 57% of all CEOs are also chairmen of the board and 13% are firm founders. The average (median) CEO is 54 (55) years old, owns 4.6% (1.8%) of the firm s common equity, and has been the chief executive for about 7 (5) years. These characteristics are in line with those in Hartzell, Ofek, and Yermack (2004) who report the following CEO characteristics: median tenure of 5 years, mean age of 54, an incidence of CEO/chairman duality of 70%, and average equity ownership of 3.6%. During our sample period, the average CEO earns $5.4 million in annual total compensation, 33% of which comes from stock-option pay. These values are consistent with the evidence in Bebchuk and Grinstein (2005) who report an average of $5.01 million in total CEO compensation. 12 We calculate the present value of the expected lost compensation for the target CEOs once their firms are sold. To do so, we use information on salary, bonus, other annual compensation, long-term incentive payout and the value of restricted stock and option awards as reported in proxy statements before the merger announcement. 13 Fich 12 Bebchuk and Grinstein (2005) report the average total compensation of $9.41 million for CEOs of S&P500 firms, $3.94 million for CEOs of MidCap400 firms, and $2.05 million for CEOs of SmallCap600 firms during We make a number of assumptions to estimate the expected lost compensation. First, following Hartzell, Ofek, and Yermack (2004), we assume that all CEOs retire by age 65 and that CEOs who are at least 65 years old expect to stay in office one more year before retiring. Second, following Yermack (2004), we assume that the probability of departure increases by 4% each year due to acquisitions, delistings, or other turnover reasons. Third, we assume that salary and bonus would increase by 2% from that received during the year prior to acquisition when firm performance is above the Fama and French (1997) median industry ROA. This assumption follows Bebchuk and Grinstein (2005), who report a 40% increase in salary and bonus for the period Fourth, we assume that the probability of 11

13 and Shivdasani (2007) estimate that the present value of lost income for CEOs expected to remain in office for another seven years is $45.5 million. We estimate that, on average, the present value of the expected lost compensation for target CEOs in our sample is $40 million. 14 Consequently, it appears that employment termination due to an acquisition triggers non-trivial wealth losses for target CEOs Golden parachute for target CEOs Many boards of directors provide golden parachutes to their CEOs. We read proxies filed by the target firms with the SEC for the fiscal year before the merger announcement to obtain information on CEO employment contracts and golden parachute provisions. Among the 851 targets, 735 (or about 86%) have a golden parachute in place for their CEOs before the merger negotiation starts. From the target CEO s employment agreement, we are able to estimate the size of golden parachutes. Specifically, when a parachute is provided, the employment agreement often stipulates that the parachute payment is based on a multiple of the executive s regular cash compensation. Panel D of Table 2 shows that the mean (median) parachute payment is $4.873 million ($2.553 million). Following Hartzell, Ofek, and Yermack (2004), we estimate the target CEO s total merger pay package. This package includes common equity and stock option appreciation, the golden parachute, and, in some instances, a special merger bonus. Chart 1 in Figure 1 displays the distribution of the CEO s merger pay package. Equity based appreciation accounts for more than two thirds of the package and parachutes comprise almost one third of the merger pay package for the average target CEO in our sample. When the deal is being negotiated, some boards vote to revise the value of existing parachutes or to put one in place for their CEOs. Target shareholders learn of these revisions from proxies filed with the SEC after the deal is publicly announced. We read merger proxies including S-4, DEFM14A, SC-TO, and/or 8K filed by the target and/or acquiring firms to obtain information on possible revisions in parachute provisions. Charts 2 and 3 of Figure 1 show the number of parachute revisions for firms with and without departure increases by an additional 2% when firm performance is below the median industry performance. Finally, we use a real rate of 3% to discount cash flows. 14 According to Liebowitz (2003), the wealthiest 1% of U.S. households has an average net worth of $10.2 million. Based on this level of wealth, this loss represents about four times the average net worth of the wealthiest 1% U.S. household in

14 parachutes, respectively. We find that among 735 target CEOs that have a golden parachute before the merger, 30 (or about 4%) experience an increase in the value of their existing parachute. In the remaining 116 cases that do not have a golden parachute for their CEO, 23 boards (or almost 20%) implement one after merger negotiations begin. The incidence of parachute augmentation in our sample is similar to that in Hartzell, Ofek, and Yermack (2004) who report that about half of parachute augmentations happen in cases without an existing prior parachute. Among 53 cases of augmentation in our sample, 23 (or 44%) do not have a parachute in place for the CEO prior to the augmentation. Table 2, Panel D shows that for these 53 cases, target CEOs experience an average increase of $1.687 million in their merger pay package due to a golden parachute. Perhaps more interesting and undocumented in prior research are instances of diminution of existing parachute. We find 9 cases (about 1% of the sample) in which the target firms reduce the golden parachute. In these situations, the mean reduction in the value of the parachute is $1.987 million Measures of parachute importance The stated goal of golden parachutes is incentive alignment. But parachutes are only one part of a CEO s pay package and to truly measure alignment we must consider the relative incentives of the entire package. Some elements of the pay package (equity, options and probably merger related bonuses) increase directly with the size of the premia offered the target. Other elements are likely to decrease with the completion of a deal and are invariant to the size of negotiated premia (e.g. perquisites of control, the loss of future salary). Golden parachutes are triggered by a change in control and are also unrelated to the size of the negotiated premia. Obviously, a trivial parachute relative to a large loss in compensation is unlikely to motivate an executive to aggressively negotiate a deal. Conversely, an overly generous parachute relative to other elements of the pay package could induce a rush to sale without adequate concern for the premia offered. In order to truly gauge incentive alignment, it is necessary to consider parachutes relative to the various elements of the entire pay package. Parachute payments mitigate any wealth losses target CEOs incur when their firms are sold. These losses arise because, in most cases, targets CEOs lose their jobs as their firms cease to exist as a standalone entity. As a result, parachutes might be particularly important for CEOs expected to remain in office if their 13

15 firms were not sold. Income losses are likely to be more acute for these executives. This phenomenon provides an additional rationale for examining the importance of golden parachutes relative to the expected value of lost income for target CEOs if their companies are acquired. As Figure 1 shows, golden parachutes play an important role in the welfare of target CEOs when their firms are sold. In this paper, we propose two measures of their relative importance. The first is the proportion of the parachute value in the total merger pay package. Raith (2008) theorizes that only the output or performance based portion of the pay package will induce agents to increase the principal s wealth. In the context of mergers, a golden parachute is not output-based compensation but will be collected together with other output-based pay components such as common equity, stock options, and special merger bonus once the deals materialize. Therefore, we benchmark the value golden parachutes against the merger pay package to measure whether the non-output-based proportion of the package affects the way target CEOs negotiate acquisition offers. Our second measure is the value of golden parachute divided by the compensation the CEOs expect to forego when their firms are sold. Hartzell, Ofek, and Yermack (2004) find that last-minute special merger bonuses to target CEOs act as a form of ex post settling up in which target CEOs are made whole for the benefits they lose when firms are sold. In our context, golden parachutes will mitigate wealth losses target CEOs incur as a result of the merger, especially for CEOs expected to remain in office if their firms were not sold. Therefore, this measure not only helps gauge how parachutes contribute to make the executives whole, but also examines whether the target boards ex-post settle up CEO compensation with golden parachutes. Table 3, Panel A shows summary statistics for our two measures of parachute importance. The average (median) golden parachute covers 30.5% (23.1%) of the total merger package to target CEOs when the deal is completed. The average (median) golden parachute represents 25.2% (12.3%) of the expected lost compensation. In Panel B and Panel C of Table 3, we show the variation of parachute importance together with the distribution of our sample over time and across industries, respectively. The two measures are generally stable over time, albeit slightly larger in The annual number of mergers is higher at the beginning and the end of our sample period, which coincides with periods of economic expansion when the 14

16 stock market valuation is higher. Shleifer and Vishny (2003) and Rhodes-Kropf and Viswanathan (2004) show that stock market health drives merger activity. The temporal distribution of our sample appears in line with the merger activity reported in these studies. We do not find any significant variation in the parachute importance across industries, except that parachutes seem to be particularly important in the utilities industry. The industrial distribution of our sample is also similar to that in other research in merger and acquisitions and to the actual distribution in the base sample from SDC. For example, Officer (2003) reports that 2.03% of his sample are firms in durable consumer goods, 17.40% in business equipment, 7.81% in shops, and 4.62% in chemicals. The percentage of targets in our sample that belong to those same industries is quite similar: 2.70%, 20.09%, 9.99%, and 2.12%, respectively. In addition, the base acquisition sample from SDC has 22.59% of targets in business equipment, 3.84% in telecommunication, and 8.86% in the healthcare industry. Analogously, the incidence in our final sample is 20.09%, 4.00%, and 8.93% for those same industries, respectively. 4. Determinants of golden parachutes In Table 4, we run a set of four tobit models to study the importance of golden parachutes for target CEOs in our sample. 15 All models control for year and industry-fixed effects. In the first two columns of the table, we use the size of the golden parachute relative to the entire merger pay package as the dependent variable. In the last two columns, we use the ratio of the parachute s size to the present value of lost compensation to the CEO as the dependent variable. The regressions also control for various specifications of target, CEO, and governance characteristics, which could affect the relative importance of parachutes; these are defined in the legend accompanying Table 4. Our results indicate that parachutes are less important relative to all merger-based-pay in firms that are larger, have higher levels of Tobin s Q, and in situations where the CEO is a founder, or when the firm is a family firm. Thus, the materiality of parachutes decreases: (a) in larger firms where, as previous research 15 We run tobit models because our measures of the relative importance of parachutes are left-hand censored. 15

17 shows, 16 CEOs often get paid more; (b) in companies with high growth opportunities where equity-based pay might be more valuable; and (c) in organizations in which founding families benefit from large ownership stakes. The importance of parachutes is also an inverse function of the tenure of the CEO. This result appears consistent with the idea that parachutes might be more important for CEOs expected to remain in office for several years. The relative importance of parachutes increases when the CEO is also chairman. In terms of the marginal effect implied by the coefficients in models (1) and (2) of Table 4, the importance of parachutes increases by 2 percentage points when the CEO is also the chairman. This result appears consistent with the conjecture that more powerful CEOs negotiate larger parachutes. The estimates also indicate that parachute importance increases by about 3 percentage points with a one standard deviation increase in the G index: firms with greater takeover defenses are more likely to have greater parachutes as part of the merger-related-pay. The results in Table 4 suggest that parachutes are particularly important when governance structures are weak and/or top managers are entrenched. 5. Do golden parachutes affect merger negotiations? Not all CEOs are offered a golden parachute as part of their compensation. As Chart 2 of Figure 1 shows, 116 out of 851 target CEOs (about 14% of our sample) do not have a golden parachute prior to the start of merger negotiations. Moreover, Chart 3 reveals that once merger talks begin, 39 of the 735 firms that offer parachutes modify these agreements while 23 of the 116 firms that do not offer parachutes put one in place. The incidence and revisions of parachutes in our sample suggest that many factors affect the way CEO compensation contracts are structured. Because parachutes and compensation contracts are endogenously determined, our analyses must account for this fact. Thus, when appropriate, our tests control for selfselection by using the Heckman (1979) inverse Mill s ratio (λ). The first stage of this method requires the 16 See, for example, Baker (1939), Murphy (1998), and Frydman and Saks (2007). 16

18 estimation of probit models of the determinants of golden parachutes. 17 For reference, we report these models in Table A1 of the appendix. The frequency and amendments to parachutes are also important in other dimensions. First, as Raith (2008) theorizes, pay structure ultimately determines whether CEOs will engage in self-dealing or use their firm-specific knowledge to improve firm value. Second, the revisions of parachute agreements once merger talks start might reflect compensation adjustments consistent with Fama s (1980) ex-post settling up. Third, Harris (1990) argues that the without an optimally sized golden parachute, target managers may reject bids that increase their shareholders wealth. These theories suggest that the relative importance of golden parachutes may affect deal negotiations. To investigate if this is the case, we examine the effect of parachute importance on negotiation-related deal characteristics such as attitude, sale procedure, termination fees, and deal completion. We believe that tests of whether parachutes affect merger negotiations might provide helpful clues to differentiate among our hypotheses. These tests would imply an effect of parachutes on the wealth of target CEOs and shareholders; however, the tests will not reveal the magnitude, direction or significance of the wealth effect. Later, we present a series of tests in which we measure the potential wealth effects to target CEOs and shareholders induced by parachutes Deal attitude As noted earlier, the relative importance of the parachute might affect a CEOs willingness and stance towards an acquisition offer. Consequently, we examine the relation between deal attitude and the importance of the parachute. In Table 5, we run a set of logit regressions in which the dependent variable takes the value of 1 is the deal attitude is characterized as hostile in the SDC database, and is 0 otherwise. The key independent variable in these tests is the size of the parachute relative to the merger pay package (models 1 and 2) and relative to the target CEO s lost compensation (models 3 and 4). Our regressions control for target and deal characteristics similar to those in Schwert (2000). These variables 17 The probit models are estimated for the universe of 25,047 firm-years that have data available from CRSP, Compustat, and the IRRC database, respectively during In the second stage, the inverse Mill's ratio from the probit model is included in the estimation as a variable to control for self-selection. 17

19 are defined in the legend accompanying Table 5. In addition, all models control for year- and industryeffects, and in models (2) and (4) we control for self-selectivity. In all models of Table 5, we find that the probability of a hostile deal declines with the importance of the parachute. 18 According to our estimates, a one standard deviation in parachute importance decreases the probability of hostility by 9.6 percentage points in models (1) and (2) and by about 11.6 percentage points in models (3) and (4). These findings could be consistent with theories of incentive alignment in that parachutes reduce managerial resistance to takeover bids. However, hostility might be desirable if it increases the target s bargaining position (Comment and Schwert, 1995). Under this possibility, hostility may either enhance an acquisition offer or defeat an unfavorable one. We study the related wealth effects of parachutes in Section Sale procedure The way in which targets are sold provides another alternative to test whether the importance of golden parachutes affects the way in which mergers are negotiated. Accordingly, we examine the relation between target characteristics, deal characteristics, parachute importance and the probability that the firm is sold as an auction. Table 6 reports the estimation of a set of year- and industry-fixed effects logit models. We run four logit regressions. As with our attitude tests, in models (1) and (2) the importance of the parachute is relative to the merger pay package, while in models (3) and (4) the importance is relative to the target CEO s lost compensation. Our control variables, described in the legend of Table 6, are similar to those in Boone and Mulherin (2008). To control for self-selection, we include the inverse Mill s ratio in models (2) and (4). We find that the importance of the parachute is significantly related to the probability that the firm will be auctioned. The marginal effect we estimate in model (2) indicates that one standard deviation increase in parachute importance increases the probability that the target is auctioned by percentage points. The marginal effect that arises from model (4) is 7.86 percentage points. These results suggest that golden 18 We note that, as in Schwert (2000), in all models we also find a positive association between target size and the probability that deals are characterized as hostile. 18

20 parachutes induce managers to avoid opposing or derailing deals. Further, to the extent that the auction could result in a higher offer, this could be evidence in support of the incentive alignment hypotheses. Alternatively, managers motivated to sell the target because they want to cash in on their parachutes, conduct an auction to ensure that their firm is sold. Results from our control variables appear in line with those in the existing literature. We find that the sale procedure is more likely to be an auction if the target initiates the deal. Auctions are also more likely as the potential depth of the bidder pool increases. This result is consistent with the work of Klemperer (2002) and Boone and Mulherin (2008). Auctions are about 12 percentage points less likely when the bidder has a toehold and approximately 14 percentage points less likely when payment is in stock. The latter result, which is consistent with Hansen (1987), can be interpreted as follows. Given that a bidder in a stock deal has to reveal more information to the target, that bidder would prefer negotiation rather than an auction so as to protect confidential information. In the event of an auction, the probability of winning the deal is reduced for any single bidder. Information shared by the losing bidder is now in the hands of a potential competitor. Also, speed is of greater importance in auctions and speed is facilitated by a cash offer Acquirer termination fees The existence and magnitude of acquiring firm termination fees is a commitment device. Increased fees commit a potential bidder to negotiating diligently but ex ante they could also deter a prospective bidder from seriously contemplating an acquisition. Consequently, the existence and size of acquirer termination fees might be useful in determining (a) whether the target or the acquirer has the upper hand in the deal (i.e. controls the synergies) and/or (b) the willingness of targets to be acquired. To examine acquirer termination fees in our sample, we run the four tobit models in which the dependent variable is the amount of acquirer termination fees. The results are shown in Table 7. In the first two models, the key independent variable is parachute importance relative to the total merger-relatedpay target CEOs get. In the last two models, parachute importance is measured relative to the target CEO s lost compensation. Our control variables, defined in the legend accompanying Table 7, are similar to those 19

21 in the Bates and Lemmon (2003) investigation of acquirer termination fees. All of the tobit models include year and industry indicators and the inverse Mill s ratio is an additional control in models (2) and (4). The results in Table 7 indicate that parachute importance materially affects acquirer termination fees. 19 Based on the estimates in models (1) and (2), a one standard deviation increase in the importance of parachutes is associated with a reduction in fees of about $65 million. Analogously, models (3) and (4) imply a fee reduction of almost $10 million. Again, the absence of measurable wealth effects in these tests restricts our ability to assess how these results affect firm value. We examine that issue in a subsequent section. Nonetheless, the results in Table 7 reaffirm the idea that the relative importance of parachutes affects the way in which firms are sold. We note that findings related to our control variables agree with those documented by others. For instance, the positive and significant coefficient estimates of target size, the market-to-book ratio, and stock payment are similar to those in Bates and Lemmon (2003) Deal completion In Table 8, we examine the relation between parachute importance and deal completion. One presumes that completed deals are beneficial to target shareholders since premia are generally paid and in the case of mergers and tender offers the target shareholders have the option of not approving the deal. Nevertheless, an improperly designed golden parachute could cause an executive to push for deal completion regardless of the premia offered to target shareholders. In Table 8, we report the estimation of four variants of a fixed-effect logit model in which the dependent variable equals 1 for completed deals and 0 for withdrawn deals. Officer (2003), Bates and Lemmon (2003) and Walkling (1985) estimate similar models. Therefore, the control variables in our regressions are similar to theirs. The exceptions, of course, are our proxies of parachute importance. Consistent with the literature, we find that deals are about 9.11 percentage points more likely to be completed if there is a target termination fee. This marginal effect is comparable to that of 11 percentage points in Officer (2003). In addition, tender offers are 4.5 percentage points more likely to go through, as 19 We also study target termination fee provisions. These provisions are unrelated to the relative importance of the parachute. 20

22 are acquisitions in which the parties to the deal are in the same industry. Deals are less likely to be completed if there is prior bidding and if the deal is hostile. These results are similar to those in Bates and Lemmon (2003). Of primary interest is the result that deal completion increases with the importance of the golden parachute. The marginal effects implied by the estimates in all four tests in Table 8 indicate that a one standard deviation increase in parachute importance raises the probability of deal completion by over 6 percentage points. This finding could be consistent with the incentive alignment hypotheses in that executives are less likely to oppose or derail a deal if they have more to gain personally from its completion. Nevertheless, deal completion is necessary for the target CEO to cash in his parachute. Under this view, the positive relation of the golden parachute to deal completion could also be consistent with rent extraction. 6. Wealth effects of golden parachutes So far, our tests document that as the importance of the parachute increases, deals are less likely to be hostile, more likely to be auctions, subject to lower acquirer termination fees, and more likely to be completed. Consequently, these analyses show that the relative importance of parachutes to target CEOs significantly affects the way in which mergers are negotiated. Next, we evaluate the effect of parachutes on the wealth of target CEOs and shareholders Parachute importance in acquisition premia: Univariate tests An important test of the incentive alignment vs. rent extraction hypotheses is the relation of the golden parachute to premia paid in the acquisition. To the extent that the golden parachute aligns owner-manager interests, the wealth paid to shareholders should increase. Proponents of golden parachutes argue that this alignment induces the executive to do the right thing in the face of an acquisition attempt on their firm. Moreover, it is argued that the golden parachute can be used as an effective bargaining tool for increasing the premium. To investigate this issue, we begin by estimating the average and median premium for targets 21

23 in our sample using the four-week premium reported by SDC. 20 Results are shown in Panel A of Table 9. To get some sense of whether parachutes affect takeover offers, we split the mean and median premia by whether the target CEO is offered a parachute. Results indicate that targets that do not offer parachutes receive higher premia. The mean (median) premium associated with targets without parachutes is 44.67% (34.86%). In contrast, the mean (median) premium of 35.2% (30.47%) is associated with targets offering parachutes. The difference in premia is statistically significant at the 1% level. In Panel B of Table 9, we sort observations into quintiles according to the importance of the golden parachute. The incentive alignment hypothesis suggests that premia will increase with the importance of the parachute. The results appear inconsistent with the incentive alignment hypothesis. Shareholders of target firms in the lowest quintiles of parachute importance earn premia of 40-50%. The mean and median premia appear to decrease monotonically across the quintiles, particularly in those tests where parachute importance is related to the merger pay package. Thus, when executives are given a stronger incentive to sell the firm, they seem to do so, even at the cost of lower premia. These results appear consistent with rent extraction hypotheses. Of course, there are sound economic reasons for the premia offered in acquisitions and it is important to control for these factors before passing judgment on the rent extraction hypothesis. That is, there may be economic reasons why premia are lower in particular deals. Moreover, because of the univariate nature of the tests in Table 9, such economic factors are not accounted for. Consequently, we proceed with a set of multivariate analyses Parachute importance in acquisition premia: Multivariate tests Our previous tests show that premia decline as parachute importance increases. At first glance, such result appears consistent with the rent extraction hypothesis because it suggests that as the incentive for executives to complete the sale increases, shareholders lose. Nonetheless, it is possible that targets receive lower premia because those firms are difficult to sell. Given this scenario, the offers these targets get may reflect their true value. Moreover, for these firms a more important parachute might be justified. Under this 20 Following Officer (2003) we restrict this premium measure to 2 (or 200%) to avoid extreme outliers. 22

24 view, the results of our univariate tests are not necessarily inconsistent with the incentive alignment hypothesis. To further examine these ideas, we design a two-step test to directly investigate if the importance of golden parachutes affects whether the premia target firms receive are consistent with expectations. The first step involves estimating the premium that targets should earn based on their characteristics. For this purpose, we use the full sample of 851 deals to run a baseline regression similar to those in Walkling and Edmister (1985) and in Bargeron, Schlingemann, Stulz, and Zutter (2008) to predict the size of the acquisition premium. This regression is reported in Panel A of Table 10. In the second step, we compare the actual premia targets realize with their predicted premia from this baseline regression. In Panel A of Table 10, the dependent variable in the baseline regression is the acquisition premium as reported by SDC, calculated as the offer price as a percentage of the target stock price four weeks prior to the merger announcement. Consistent with the existing literature, we also find that acquisition premia increase with leverage, liquidity, and the existence of cash offers, tender offers, and hostile offers. Bid premia also increase with rumors, prior bidding, and the existence of a target termination fee. Bid premia decrease with the size of the firm, recent excess returns, CEOs near retirement age, and acquisitions by private acquirers. The adjusted R 2 on this model is about 22%. In Figure 2, we plot the actual and predicted premia related to targets in our sample according to quintiles of parachute importance. We observe that while actual premia decrease as parachute importance increases, predicted premia appear fairly stable across quintiles. Moreover, from the graphs in Figure 2, it appears that in some instances targets earn premia higher than those predicted. To investigate this issue in greater detail, we perform the second step of our test. In Panel B of Table 10, we estimate four models of the premium difference for targets in our sample. The dependent variable in all tests is defined as follows. Δ premia = Actual premia as reported in SDC Predicted premia from the baseline regression 23

25 All tests control for year- and industry-fixed effects, self-selection, as well as for other target- and dealspecific variables defined in the legend accompanying Table In model (1), the key independent variable is the importance of the parachute relative to the merger pay package. In model (3), parachute importance is measured relative to the target CEO s lost compensation. Given our univariate results and the premia patterns in Figure 2, in models (2) and (4) we use quintiles of the level of parachute importance. The coefficients in Panel B of Table 10 document an inverse association between parachute importance and premia difference. The marginal effect associated with the estimates in model (1) indicates that a 10 percentage point increase in parachute importance causes target firms to earn premia of 3.97 percentage points lower than predicted. 22 To put this result in perspective, for the average target in our sample the gap between actual and predicted premia implies a shortfall of about $196.7 million in terms of deal value. Consequently, our findings document severe wealth effects for shareholders in these targets. Moreover, since the results in model (2) indicate that the premia shortfall increases with the importance of the parachute, we interpret our results as consistent with the rent extraction hypothesis. Harris (1990) theorizes that an improperly sized golden parachute may lead target managers to accept takeover bids that may not increase their shareholders wealth. The evidence we present appears to support her theories. In model (2) of Table 10, we also note that the two first two quintiles, in which parachute importance is low, do not exhibit statistically significant coefficients. This lack of significance indicates that, in about 340 deals, targets approximately earn their predicted premia. In contrast, in the remaining 511 deals, targets earn premia that falls significantly short from expected values. Consequently, about 60% of targets in our sample appear to have an inadequately sized parachute Parachute importance and ex-post settling up 21 To make sure these fixed effects models are properly specified, in the specification, we include control variables that (1) that happen after the deal announcement (such as time to completion) and (2) are not used in the base line regression in Panel A of Table 10 (such as macroeconomic conditions and CEO employment). We thank Bill Greene for helping us with this matter. 22 In general, the results in models (3) and (4) are in line with those in models (1) and (2). For instance, in model (3), a one standard deviation increase in parachute importance is associated with a marginal effect of just under 2 percentage points. 24

26 Fama (1980) puts forth a model in which the manager s wage is revised to account for his previous performance. In Fama s model the possibility of upward wage revisions motivates managers to work harder and increase firm value. On the flip side, underperforming managers will experience wage decreases in future periods or ex post settling up wage discounts. In our discussion of Chart 3 of Figure 1 we note that almost 20% of the 116 firms without golden parachutes offer them during negotiations. Chart 2 of Figure 1 also shows that about 5% of the 735 firms with parachutes alter them during negotiations; 1% are lowered, 4% are raised. We use these revisions of parachutes in our sample to test Fama s (1980) theory of ex-post settling up. Results are shown in Table 11. We run two logit regressions in which the key explanatory variable is the excess compensation target CEOs receive in the fiscal year before the acquisition. This excess compensation is the residual from a total pay regression estimated for all firms in our sample. 23 In model (1) of Table 11, the dependent variable is a dummy that takes the value of 1 if targets either augment the size of an existing parachute or put one in place. The variable is set to 0 otherwise. The coefficient estimate for this dummy variable, (-0.288, p-value = 0.071), indicates that when CEOs enjoy lower excess pay prior to the merger, their targets increase the importance of the parachute. This result suggests that for these CEOs the augmentation of parachutes acts as a form of ex post settling up, whereby executives wages increase in lieu of prior lower pay. 24 In model (2) of Table 11, we analyze the 735 targets in our sample that offer their CEOs a parachute prior to the start of merger negotiations. The dependent variable in this model is a dummy that equals 1 if the size of the parachute is reduced and equals 0 if it is not. The excess compensation variable exhibits a positive and significant coefficient (2.019, p-value = 0.055). This result suggests that managers that enjoy higher excess pay prior to the deal experience ex post settling up wage discounts in the year of the acquisition. Caution must be applied in this latter case since the number of parachute reductions is small. 23 The excess compensation variable is constructed following Hartzell, Ofek, and Yermack (2004) as the residual from a basic compensation regression model using a panel data of large universe of firms from our sample and Execucomp. We regress salary and bonus against two-digit SIC industry indicator variables, the natural logarithm of market capitalization, year indicators, and the prior stock return. 24 This result is consistent with the findings in Hartzell, Ofek, and Yermack (2004). Their study, which focuses on the special merger bonus for target CEOs, examines parachute augmentations. 25

27 Overall, the results in Table 11 indicate that the augmentation or diminution of the importance of golden parachutes once merger talks begin can help boards of target firms in ex post settling up wages for their CEOs. Another result in Table 11 deserves some attention. The coefficient for the expected lost compensation is positive in model (1). This result also suggests that augmenting the importance of parachutes works as a way to settle up ex post since target CEOs are compensated for the salary they lose when their firms are sold. 7. Robustness tests In this section, we probe our results by addressing a number of issues that could potentially alter our results or the inferences related to our findings. To conserve space, we do not report the empirical analyses related to the issues examined in this section. Instead, we discuss the economic effect related to each test Takeover premia alternatives The estimates presented in Tables 9 and 10 are based on the four-week premium reported by SDC, defined as the offer price divided by the target s stock price four weeks before the announcement date. We re-estimate all premia using the combined premium method in Officer (2003). Following his approach, we first estimate a premium based on component data using the aggregate value of cash, stock, and other securities offered by the bidder to target shareholders as reported by SDC. We then estimate premia based on initial price and final price data based on the initial offer and final offer price, respectively. These prices are also reported by SDC. All three premium measures are deflated by the target s market value 42 trading days prior to the bid announcement. The combined premium is based on the component measure if it is greater than 0 and less than 2; otherwise the premium relies on the initial price measure (or on the final price measure if initial price data are missing). Using the combined premium, the multivariate results in Table 10 Panel B are as follows. In model (1), the coefficient on the importance of the parachute relative to the merger pay package (GP/MPP) is

28 (p-value <0.001). 25 The marginal effect associated with this estimate implies that a 10 percentage point increase in parachute importance causes target firms to earn a premium 4.30 percentage points lower than predicted. This shortfall triggers a decline of $ million in deal value for the average sample target. This result is similar to those tabulated. We also replace the four-week premium with the cumulative abnormal return (CAR) accruing to sample targets running from 20 days before the deal announcement (AD-20) until the day after (AD+1). Using the CAR (AD-20, AD+1) as the premium proxy, the multivariate results in Table 10 Panel B continue to generate inferences similar to those reported. For example, in model (1) of the table, the coefficient on GP/MPP changes to (p-value <0.001). This estimate implies that a 10 percentage point increase in parachute importance reduces the average deal value by $100.1 million Tax regulations and the Sarbanes-Oxley Act On February 19, 2002, the Internal Revenue Service ( IRS ) proposed new regulations to Section 280G of the Internal Revenue Code (IRC). 26 The new regulations provide amendments and clarifications to the regulations issued on May 5, 1989, and apply to golden parachutes payments occurring on or after January 1, Section 280G states that, "if the present value of a change-in-control payment exceeds the safe harbor (three times the average taxable compensation over the five most recent calendar years preceding the change-in-control, less $1), the company loses tax deductions for these excess amounts. Additionally, the executive is required to pay a 20% excise tax on the excess payment." A 2008 study by RiskMetrics finds that the new tax regulations have done little to reduce parachute payments. 27 In particular, the study reports that two-thirds of the companies in the S&P 500 index disclose that they would provide excise tax gross-ups to one or more top executives. The excise tax gross-ups essentially eliminate the executive from personally paying the excise tax on excess parachute payments. The RiskMetrics study shows that excise tax gross-ups are a costly benefit, since it generally takes at least 25 With regards to Table 10, Panel B, model (3), the coefficient on Parachute / Lost compensation is (p-value = 0.053). This estimate implies that a one standard deviation increase in parachute importance causes target firms to earn a premium of 2.18 percentage points lower than predicted. 26 See: REG at 27 See: Gilding Golden Parachutes: the Impact of Excise Tax Gross-Ups by Kosmas Papadopoulos at 27

29 $2.50 and as much as $4 to cover each $1 of excise tax that must be grossed-up. In addition, other companies that do not provide the gross-up benefit may increase parachute payments in order to mitigate the excise tax to their executives. For our purposes, it appears that the new Section 280G rules may have affected the size of golden parachute, and, in turn, the relative importance of these payments. To investigate the potential effect of the new tax rules on parachutes, we revisit the regressions reported in Table 4 related to the relative importance of parachutes. In model (2) of the table, we include a dummy variable for deals initiated after February 19, The estimate for this variable (0.046, p-value = 0.038) indicates that parachute importance relative to the merger pay package increases by about 4.6 percentage points. While this result is consistent with the notion that some companies increase parachutes in response to the new proposed tax rules, alternative explanations are possible. For instance, as a result of the new governance rules contained in the Sarbanes-Oxley Act of 2002, many companies curbed the stock and other equity-based pay given to top managers while increasing their regular salary (Chhaochharia and Grinstein, 2009). This pay redistribution could partially account for the increase in importance of parachutes we uncover after Changes in other components of the merger pay package The relative importance of parachutes might vary due to changes in these payments that we examine during the merger negotiation period. However, such importance could also change, even if parachutes are not revised, when other components of the merger pay package change. To tests this possibility we reestimate the models in Table 11 related to parachute augmentations and reductions, respectively. To each test, we add a (0,1) indicator that is 1 if the target CEO s compensation contract is amended to include a special merger bonus or if the chief executive receives stock, restricted stock, or stock options during merger negotiations. The new estimates indicate that parachute augmentations are about 2.38 percentage points less likely when CEOs receive any of the above payments. Conversely, we also find that altering the non-parachute components of the merger pay package is unrelated to the reduction of parachutes Family firms and founding CEOs 28

30 Villalonga and Amit (2006) show that founders serving as CEOs can enhance firm value. In a related context, Anderson and Reeb (2003) find a similar result for family firms headed by family CEOs. In our sample, we identify that in 32 out 80 family firms, the CEO is a founding member. Our data also include 109 cases in which the CEO is the firm s founder. With this information we estimate whether founders and/or family firms affect the relative importance of parachutes. We alter the tests in Table 4 to include a (0,1) indicator that is 1 if the firm is a family firm. We also add a separate dummy variable for founder CEOs. 28 The results reveal that, relative to their merger pay package, parachutes are 9 percentage points less important for CEOs in family firms and about 7.9 percentage points less important for founding CEOs. Since we show that higher parachutes are associated with lower takeover premia, these results might provide additional evidence on the benefits of founders and families to firm value. However, the results might also reflect the fact that families and founder CEOs earn more equity based pay CEOs near retirement In preceding sections, we argue that parachutes might be particularly important for CEOs expected to remain in office for several years. This argument is based on the idea that these executives will lose the compensation and benefits of being CEO when their firms are sold. As a result, parachutes make them partially whole for such loss. A plausible inference related to this argument is that the relative importance of parachutes might decrease as the chief executive approaches retirement. However, if parachute agreements are not reduced prior to a CEO s retirement, parachute importance relative to expected lost compensation may experience a mechanical increase. Such increase might be driven by a reduction in the expected lost compensation for CEOs approaching retirement. To test these issues, we first evaluate whether parachutes are likely to be either augmented or reduced as target CEOs approach retirement. To the tests reported in Table 11 we add a variable, labeled years to retirement, that subtracts the target CEO s age from 65. This variable, which uses the retirement age threshold in Hartzell, Ofek, and Yermack (2004), is set to 0 for negative values (situations in which the CEO is still in office past the retirement age). Coefficients estimates reveal that parachutes are more likely to be augmented as CEOs approach 28 To avoid possible confounding effects, we include the founder (0,1) and the family firm (0,1) separately. 29

31 retirement, though the effect appears quite small. Every year the target CEO gets closer to retirement increases the probability of parachute augmentation by about 0.17 percentage points. In contrast, approaching retirement is unrelated to parachute reductions. We also re-estimate the parachute importance (model 4 of Table 4) for CEOs approaching retirement. The coefficient estimate on the years to retirement variable ( , p-value<0.001) indicates that every year the target CEO approaches retirement increases the importance of the parachute relative to the expected lost compensation by about 3.15 percentage points. This result probably reflects the fact that parachutes are not reduced as CEOs approach retirement while, at the same time, the expected lost compensation decreases. 8. Conclusions In recent years, the controversy surrounding golden parachutes has intensified. Opponents of golden parachutes claim that it is unfair to provide managers with a financial safety net regardless of the fortunes of shareholders. Advocates of parachutes view them as an indispensable part of a competitive compensation package required to attract and retain highly qualified human capital. From an academic perspective, this controversy can be summarized with the following research question: Do golden parachutes align the incentives of the executives receiving them and the shareholders in the companies that grant them? We frame this research question in the context of three well-known hypotheses in corporate finance: incentive alignment, rent extraction, and ex post settling up. We investigate these hypotheses in a sample of 851 acquisition offers during The hypotheses we examine continue to receive considerable attention because they are well suited in situations in which agency problems arise. In our case, the problem is a form of moral hazard. If a parachute insulates target CEOs from a severe personal wealth loss related to the acquisition of their firm, they may behave differently from the way they would if they were fully exposed to such loss. Such behavior may be counterproductive for shareholders if it generates either a rush to sale or unyielding resistance despite the acquisition price offered. 30

32 Our work also indicates the need to measure the relative importance of parachutes. In this paper, we propose two different proxies to assess the degree by which golden parachutes protect target CEOs from acquisition-related personal losses. The first proxy scales the size of the parachute by the total mergerrelated pay target CEOs receive; the second scales it by the expected lost compensation target CEOs incur when their firms are sold. These proxies are unlike those in the extant literature that either track the existence of a parachute or appraise the increased acquisition costs related to the parachute. Because our proxies measure the relative importance of the parachute to the target CEO, they provide a unique prism to examine our hypotheses. Tests related to deal negotiations reveal that the extent by which parachutes protect target CEOs from personal wealth losses causes these executives to behave differently in the face of an acquisition offer. Specifically, we find that as the relative importance of the parachute increases, deals are less likely to be hostile, more likely to involve an auction, subject to lower acquirer termination fees, and more likely to be completed. These results suggest that the motivation to sell a target and complete a deal increases with the relative importance of golden parachutes. On the surface, these findings appear consistent with the incentive alignment hypothesis. Nonetheless, because parachute payments are contingent on deal completion it is possible that target CEOs sacrifice premia for personal gain. Consequently, we test the rent extraction hypothesis by examining the relation between parachute importance and shareholder wealth. Our findings indicate that a parachute is more likely to be augmented (or put in place for the first time) for target CEOs (i) who enjoy higher excess total compensation during the year before the merger, or (ii) who expect larger compensation losses after their firms are sold. We also find that target CEOs are more likely to experience reductions in the size of their parachutes when they collect higher excess pay in the year preceding the deal. Together, these findings indicate that the revision of the importance of golden parachutes appears to help target boards in ex post settling up wages for their CEOs. The empirical analyses related to the benefits of parachutes to target CEOs and target shareholders indicate that these payments have a material and diverging effect on the wealth of these parties. The wealth effects we estimate indicate that golden parachutes cost target firms several percentage points in potential 31

33 acquisition premium. For the average target CEO in our sample, parachutes represent 30.5% or $4.9 million of the pay package the executive receives when the target is sold. In contrast, a 10% increase in parachute importance relative to the merger pay package is associated with a shortfall in premia of about 3.97 percentage points. This shortfall implies a reduction of $196.7 million in deal value for the average acquisition in our sample. Consequently, the financial costs to target shareholders in firms offering parachutes considerably surpass the benefits received by their CEOs. These results are consistent with the rent extraction hypothesis. This paper illuminates several issues related to the debate and polemic surrounding golden parachutes. Our results suggest that inappropriately designed golden parachutes to target CEOs create a conflict of interests between these executives and their shareholders at the onset of a merger. In this context, our research indicates that as target CEOs become more insulated from personal losses due to relatively larger parachutes, the moral hazard problem is exacerbated. We show that as parachute importance increases, targets earn premia that fall short from expected values. Moreover, we estimate that about 60% of targets in our sample offer too generous a parachute. Consequently, our results also indicate that the moral hazard problem associated with golden parachutes appears to be a widespread issue in acquisitions. 32

34 References Agrawal, A., and Knoeber, C., 1998, Managerial compensation and the threat of takeover, Journal of Financial Economics 47, Almazan, A., and Suarez, J., 2003, Entrenchment and severance pay in optimal governance structures, Journal of Finance 58, Anderson, R., and Reeb, D., 2003, Founding family ownership and firm performance: evidence from the S&P 500, Journal of Finance 58, Baker, J., 1939, Executive compensation payments by large and small industrial companies, Quarterly Journal of Economics 53, Bargeron, L.L., Schlingemann, F.P., Stulz, R.M., and Zutter, C.J., 2008, Why do private acquirers pay so little compared to public acquirers? Journal of Financial Economics 89, Bates, T.W., and Lemmon, M.L., 2003, Breaking up is hard to do? An analysis of termination fee provisions and merger outcomes, Journal of Financial Economics 69, Bebchuk, L., and Ferrell, A., 2002, On takeover law and regulatory competition, Business Lawyer 57, Bebchuk, L.A., and Grinstein, Y., 2005, The growth of executive pay, Oxford Review of Economic Policy 21, Berkovitch, E., and Khanna, N., 1991, A theory of acquisition markets: mergers versus tender offers, and golden parachutes, Review of Financial Studies 4, Boone, A., and Mulherin, J.H., 2007, How are firms sold? Journal of Finance 62, Boone, A., and Mulherin, J.H., 2008, Do auctions induce a winner's curse? New evidence from the corporate takeover market, Journal of Financial Economics 89, Chhaochharia, V., and Grinstein, Y., 2009, CEO compensation and board oversight, Journal of Finance, forthcoming. Comment, R., and Schwert, G.W., 1995, Poison or placebo? Evidence on the deterrence and wealth effect of modern antitakeover measures, Journal of Financial Economics 39, Cotter, J.F., Shivdasani, A., and Zenner, M., 1997, Do independent directors enhance target shareholder wealth during tender offers? Journal of Financial Economics 43, Cotter, J.F., and Zenner, M., 1994, How managerial wealth affects the tender offer process, Journal of Financial Economics 35, Dvorak, F., 2008a, Theory and practice: companies cut holes in CEOs' golden parachutes; New disclosure rules prompt more criticism of guaranteed payouts, Wall Street Journal, New York, September 15, Dvorak, F., 2008b, Proxy firm targets practice of paying executives' tax bills, Wall Street Journal, New York, November 24,

35 Fama, E., 1980, Agency problems and the theory of the firm, Journal of Political Economy 88, Fama, E., and French, K., 1997, Industry costs of equity, Journal of Financial Economics 43, Farrell, G., 2009, US bank chiefs face $500,000 limit, Financial Times, London, February 5, Fich, E.M., and Shivdasani, A., 2007, Financial fraud, director reputation, and shareholder wealth, Journal of Financial Economics 86, Frydman, C., and Saks, R.E., 2007, Executive compensation: a new view from long-term perspective, SSRN Working Paper, AFA Gompers, P., Ishii, J., and Metrick, A., 2003, Corporate governance and equity prices, Quarterly Journal of Economics 118, Grinstein, Y., and Hribar, P., 2004, CEO compensation and incentives: evidence from M&A bonuses, Journal of Financial Economics 73, Hansen, R., 1987, A theory of choice of exchange medium in mergers and acquisitions, Journal of Business 60, Harris, E., 1990, Antitakeover measures, golden parachutes, and target firm shareholder welfare, RAND Journal of Economics 21, Hartzell, J.C., Ofek, E., and Yermack, D., 2004, What s in it for me? CEOs whose firms are acquired, Review of Financial Studies 17, Hausman, J.A., and Taylor, W.E., 1981, Panel data and unobservable individual effects, Econometrica 49, Heckman, J., 1979, Sample selection bias as a specification error, Econometrica 47, Herbert, B., 2008, When madmen reign, New York Times, New York, September 29, Jensen, M.C., and Meckling, W.H., 1976, Theory of the firm: managerial behavior, agency costs, and ownership structure, Journal of Financial Economics 3, Jensen, M.C., 1988, Takeovers: their causes and consequences, Journal of Economic Perspectives 2, Klemperer, P., 2002, What really matters in auction design? Journal of Economic Perspectives 16, Knoeber, C., 1986, Golden parachutes, shark repellents, and hostile takeovers, American Economic Review 76, Lambert, R.A., and Larcker, D.F., 1985, Golden parachutes, executive decision-making, and shareholder wealth, Journal of Accounting and Economics 7, Lefanowicz, C.E., Robinson, J.R., and Smith, R., 2000, Golden parachutes and managerial incentives in corporate acquisitions: evidence from the 1980s and 1990s, Journal of Corporate Finance 6, Liebowitz, M., 2003, Winning responses to tough tax questions, May

36 Machlin, J., Choe, H., and Miles, J.A., 1993, The effects of golden parachutes on takeover activity, Journal of Law and Economics 36, Malatesta, P.H., and Walkling, R.A., 1988, Poison pill securities: stockholder wealth, profitability, and ownership structure, Journal of Financial Economics 20, Moeller, S.B., Schlingemann, F.B., and Stulz, R.M., 2005, Wealth destruction on a massive scale? A study of acquiring-firm returns in the recent merger wave, Journal of Finance 60, Murphy, K.J., 1998, Executive compensation, Handbook of Labor Economics, Ashenfelter, O. and Card, D. (Editors), Vol. 3, North Holland. Officer, M.S., 2003, Termination fees in mergers and acquisitions, Journal of Financial Economics 69, Raith, M., 2008, Specific knowledge and performance measurement, RAND Journal of Economics 39, Rhodes-Kropf, M., and Viswanathan, S., 2003, Market valuation and merger waves, Journal of Finance 59, Schwert, G.W., 2000, Hostility in takeovers: in the eyes of the beholder? Journal of Finance 55, Shleifer, A., and Vishny, R.W., 2003, Stock market driven acquisitions, Journal of Financial Economics 70, Song, M.H., and Walkling, R.A., 2007, Anticipation, acquisitions, and bidder returns, Working paper, Drexel University. Villalonga, B., and Amit, R., 2006, How do family ownership, control, and management affect firm value? Journal of Financial Economics 80, Walkling, R.A., 1985, Predicting tender offer success: a logistic analysis, Journal of Financial and Quantitative Analysis 20, Walkling, R.A., and Edmister, R.O., 1985, Determinants of tender offer premiums, Financial Analysts Journal 41, Walkling, R.A., and Long, M., 1984, Agency theory, managerial welfare, and takeover bid resistance, RAND Journal of Economics 15, White, H., 1980, A heteroskedasticity-consistent matrix and direct test for heteroskedasticity, Econometrica 48, White House Blog, Feb 4, 2009, Remarks by the President on executive compensation, Yermack, D., 2004, Remuneration, retention, and reputation incentives for outside directors, Journal of Finance 59,

37 Table 1 Golden parachutes: Selected academic studies Paper Sample Hypotheses Parachute Measure(s) Findings Lambert and Larcker (JAE, 1985) N= 57 to 90 deals ( ) Incentive alignment Wealth transfer PV GP payout/mve target PV GP payout/# managers with GPs Incentive alignment Knoeber (AER, 1986) N= 246 (1982) Incentive alignment Rent extraction GP dummy Incentive alignment Machlin, Choe, and Miles (JLE, 1993) N= ( ) Managerial resistance Incentive alignment GP size/ MVE target Incentive alignment Cotter and Zenner (JFE 1994) N= 141 tender offers ( ) Managerial resistance Incentive alignment PV GP payout Managerial resistance Cotter, Shivdasani and Zenner (JFE 1997) N= 169 tender offers ( ) Managerial resistance Incentive alignment GP dummy Managerial resistance Agrawal and Knoeber (JFE 1998) N= 446 deals (1987) Rent extraction Incentive alignment GP dummy Rent extraction Incentive alignment Hartzell, Ofek, and Yermack (RFS, 2004) N= 311 offers ( ) Ex-post settling up Agency conflicts GP augmentation dummy Ex-post settling up Agency conflicts 36

38 Table 2 Sample description This table describes our sample which consists of 851 mergers and acquisitions announced during and tracked in the Securities Data Company s (SDC) merger and acquisition database in which the target is a publicly traded U.S. company and the deal value is at least $1 million. For selecting the sample, we require that target firms have stock return, accounting, and governance data available from the Center for Research in Security Prices (CRSP), Compustat, and Investor Responsibility Research Center (IRRC) database, respectively. In Panel A, deal status, mode of acquisition, method of payment, and deal attitude are obtained from SDC. Information on sale procedure and initiator is obtained from reading the merger background filed with the SEC. As in Boone and Mulherin (2007), auction refers to cases in which the selling firm contacts multiple potential buyers while negotiation focuses on a single buyer. Initiator is the party that first contacts the other party in the sale process. A deal is in the same industry if both the target and the acquirer belong to the same Fama and French (1997) 48-industry classification. In Panel B, all financial variables are measured at the end of the fiscal year before the merger announcement date. Market-to-book is market value of equity divided by book value of equity. Leverage equals the book value of debts divided by market value of assets. Deal value is obtained from SDC. In Panel C, compensation data are as of the end of the fiscal year before the announcement date. Ownership is the percentage of stock and options owned by the CEO. Market value of ownership is measured as of 20 trading days before the announcement date. Estimated lost compensation is the estimated present value of the CEO s lost compensation when his/her firm is sold. Estimated parachute payment includes cash payment and bonus to the CEO as a result of the merger. Panel A: Deal characteristics Deal status Complete Withdrawn Total 747 (87.78%) 104 (12.22%) 851 (100%) Mode of acquisition Merger Tender offer 696 (81.79%) 155 (18.21%) Method of payment Stock Cash Mixed Unknown 138 (16.22%) 467 (54.88%) 226 (26.55%) 20 (2.35%) Deal attitude Friendly Hostile Neutral 783 (92.01%) 59 (6.93%) 9 (1.06%) Sale procedure Auction Negotiation 287 (33.73%) 564 (66.27%) Initiator Target Acquirer Unknown 334 (39.25%) 465 (54.64%) 52 (6.11%) Same industry Yes No 477 (56.05%) 374 (43.95%) 37

39 Panel B: Target characteristics Mean Median Market value ($ billion) Market-to-book Leverage ROA Age (years) Deal value ($ billion) Panel C: CEO characteristics and compensation Mean Median Age (years) Tenure (years) Chairman Founder Compensation committee member Ownership (%) Market value of ownership ($ million) Salary and bonus ($ million) Total compensation ($ million) Estimated lost compensation ($ million) % of option value in total compensation Panel D: Parachute characteristics Estimate dollar amount of parachute Mean First quartile Median Third quartile upon completion ($ million) Dollar change ($ million) Percentage change (%) Parachute revision N % of sample Mean Median Mean Median Augmentation Diminution

40 Table 3 Parachute importance characteristics The sample consists of 851 acquisitions announced during described in Table 2. We use two measures of the importance of parachute for the CEO in acquisitions: (Parachute/Merger pay package) and (Parachute/Lost compensation). Estimated parachute payment includes cash payment and bonus to the CEO as a result of the merger. Merger pay package includes gains from parachute, common equity, and options. Lost compensation is the estimated present value of the CEO s lost compensation when his/her firm is sold. In Panel C, we use Fama French (1997) 12-industry classification to show the industrial distribution of our sample. Panel A: Summary statistics Parachute importance Mean Std. dev. First quartile Median Third quartile Parachute / Merger pay package Parachute / Lost compensation Merger pay package / Lost compensation Panel B: Temporal characteristics Parachute. Parachute. Merger pay package Lost compensation Year N % Mean Median Mean Median Panel C: Industrial characteristics Parachute. Parachute. Merger pay package Lost compensation Industry N % Mean Median Mean Median Nondur. consumer goods Durable consumer goods Manufacturing Energy Chemical Business equipment Telecommunication Utilities Shops Health Finance Other

41 Table 4 Determinants of parachute importance The sample consists of 851 acquisitions announced during described in Table 2. The dependent variable is (Parachute/Merger pay package) in tobit models (1) and (2), and (Parachute/Lost compensation) in tobit models (3) and (4). All financial variables are measured at the end of the fiscal year before the merger announcement date. Q is defined as the book value of assets minus the book value of equity plus the market value of equity, divided by the book value of assets. Free cash flow is operating income before depreciation minus interest expenses, income taxes, and capital expenditures, scaled by book value of total assets. Firm age is the number of years from incorporation until the merger announcement date. Family firm (0,1) indicator equals one if a family, group of families, or firm founder controls more than 20% of the outstanding equity. G index is constructed by adding 24 antitakeover provisions tracked by IRRC as in Gompers, Ishii, and Metrick (2003). Tenure is the number of years the CEO has been in the chief executive position until the merger announcement date. Board ownership, insider ownership, and institutional ownership are the percentage of common stock owned by each group, respectively. Percent of independent directors is the fraction of independent directors on board. Other variables are self-explanatory or defined elsewhere. We report White (1980) heteroskedasticity consistent p-values in parentheses. The symbols *, **, and *** denote statistical significance at the 10%, 5%, and 1% levels, respectively. 40

42 (Table 4 continued) Parachute. Parachute. Merger pay package Lost compensation (1) (2) (3) (4) Intercept *** *** (0.824) (0.800) (<0.001) (<0.001) Firm characteristics Log (Assets) *** *** * * (<0.001) (<0.001) (0.090) (0.062) Q *** *** (0.002) (0.003) (0.485) (0.608) Leverage (0.240) (0.111) (0.238) (0.190) Free cash flow (0.640) (0.669) (0.696) (0.684) Log (Firm age) (0.124) (0.331) (0.312) (0.155) Family firm (0,1) *** ** * (<0.001) (0.036) (0.058) (0.235) CEO characteristics Founder (0,1) * *** *** (0.092) (0.169) (0.006) (0.010) Compensation committee member (0,1) (0.431) (0.328) (0.190) (0.148) Number of outside directorships (0.705) (0.818) (0.452) (0.413) Chairman (0,1) 0.053** 0.040* (0.020) (0.083) (0.886) (0.631) Log (Age) *** 1.457*** (0.125) (0.194) (<0.001) (<0.001) Tenure *** *** 0.009*** 0.010*** (<0.001) (<0.001) (<0.001) (<0.001) Ownership (%) ** (0.038) (0.111) (0.216) (0.139) Option value / Total compensation *** *** *** *** (0.001) (0.001) (<0.001) (<0.001) Governance characteristics G index 0.012*** 0.014** (0.006) (0.019) Board ownership (%) (0.525) (0.994) Percent of independent directors (0.354) (0.679) Insider ownership (%) (0.163) (0.469) Institutional ownership (%) (0.282) (0.251) Year and industry fixed effects Yes Yes Yes Yes N Adjusted R Pr>χ 2 <0.001 <0.001 <0.001 <

43 Table 5 Parachute importance and deal attitude The sample consists of 851 acquisitions announced during described in Table 2. The dependent variable in the logit models equals one if the deal attitude is coded by SDC as hostile. All financial variables are measured at the end of the fiscal year before the merger announcement date. Sale growth is measured as the proportional change in sales over the prior fiscal year. Liquidity is the ratio of net liquid assets to total assets. Price/earnings is the ratio of the year end stock price to earnings per share for the prior fiscal year. Toehold (0,1) variable equals one if the bidder owns at least 5% of the target common stock. Stock payment (0,1) variable equals one if the form of payment involves stocks. Heckman self-selectivity correction involves using a firststage estimation of the probability of having golden parachute for the CEO using a firm-year fixed effect probit model as described in the Appendix. In the second stage, the inverse Mill's ratio from the probit model is included in the estimation as a variable to control for self-selection. Other variables are self-explanatory or defined elsewhere. We report White (1980) heteroskedasticity consistent p-values in parentheses. The symbols *, **, and *** denote statistical significance at the 10%, 5%, and 1% levels, respectively. Dependent variable = 1 if deal attitude is hostile (1) (2) (3) (4) Intercept (0.888) (0.888) (0.883) (0.883) Parachute / Merger pay package *** *** (0.006) (0.006) Parachute / Lost compensation *** *** (0.007) (0.007) Target characteristics ROE (0.741) (0.739) (0.749) (0.753) Sale growth (0.823) (0.816) (0.896) (0.891) Leverage (0.880) (0.868) (0.961) (0.940) Liquidity (0.511) (0.528) (0.711) (0.739) Market-to-book (0.663) (0.660) (0.794) (0.796) Price/earnings ( 10 4 ) (0.917) (0.936) (0.956) (0.989) Log (Assets) 0.234** 0.232** 0.259** 0.260** (0.037) (0.039) (0.020) (0.019) Deal characteristics Stock payment (0,1) (0.206) (0.202) (0.156) (0.156) Toehold (0,1) (0.589) (0.592) (0.420) (0.419) Same industry (0,1) (0.291) (0.317) (0.220) (0.244) Heckman self-selectivity (0.623) (0.553) Year and industry fixed effects Yes Yes Yes Yes N Adjusted R Pr>χ

44 Table 6 Parachute importance and sale procedure The sample consists of 851 acquisitions announced during described in Table 2. The dependent variable in the logit models equals one if the sale procedure is auction, and zero in case of negotiation. All financial variables are measured at the end of the fiscal year before the merger announcement date. The intangible assets is defined as one minus the ratio of the target s property, plant, and equipment divided by its assets. Industry count is the natural log of the number of firms in the same Fama and French industry as the target that have market value of equity of greater than that of the target in the year prior to the merger as in Boone and Mulherin (2008). Strong anti-takeover state (0,1) variable equals one for targets incorporated in states determined by Bebchuk and Ferrel (2002) to have strong takeover impediments (Idaho, Indiana, Maryland, Nevada, Ohio, Pennsylvania, South Dakota, Tennessee, and Wisconsin). Rumor (0,1) variable equals one if the deal is rumored as reported in SDC. Tender offer (0,1) variable equals one if the form of the deal is tender offer. Other variables are self-explanatory or defined elsewhere. We report White (1980) heteroskedasticity consistent p-values in parentheses. The symbols *, **, and *** denote statistical significance at the 10%, 5%, and 1% levels, respectively. Dependent variable = 1 if sale procedure is auction (1) (2) (3) (4) Intercept (0.705) (0.307) (0.638) (0.258) Parachute / Merger pay package 1.033*** 0.658** (<0.001) (0.039) Parachute / Lost compensation 0.352** 0.387** (0.039) (0.048) Target characteristics Log (Assets) (0.889) (0.827) Intangible assets (0.396) (0.308) Industry count 0.259** 0.298** (0.045) (0.025) Strong anti-takeover state (0,1) (0.670) (0.503) Deal characteristics Target initiated deal (0,1) 2.333*** 2.372*** (<0.001) (<0.001) Toehold (0,1) * * (0.093) (0.064) Stock payment (0,1) ** ** (0.015) (0.021) Same industry (0,1) (0.437) (0.459) Heckman self-selectivity (0.489) (0.494) Year and industry fixed effects Yes Yes Yes Yes N Adjusted R Pr>χ < <

45 Table 7 Parachute importance and acquirer termination fee The sample consists of 851 acquisitions announced during described in Table 2. The dependent variable in the tobit models is the natural logarithm of the dollar amount of termination fee. Target s size variable is also log-transformed. Target termination fee (0,1) variable equals one if the target has a termination fee provision in the merger contract. Target lockup (0,1) variable equals one if the deal includes a lockup of target shares. Prior bidding (0,1) variable equals one if the deal follows a prior bid within one year. Tender offer (0,1) variable equals one if the form of the deal is tender offer. Other variables are self-explanatory or defined elsewhere. We report White (1980) heteroskedasticity consistent p-values in parentheses. The symbols *, **, and *** denote statistical significance at the 10%, 5%, and 1% levels, respectively. (1) (2) (3) (4) Intercept (0.923) (0.900) (0.911) (0.850) Parachute / Merger pay package *** *** (<0.001) (<0.001) Parachute / Lost compensation ** *** (0.012) (0.003) Target s size 0.646*** 0.648*** 0.675*** 0.678*** (<0.001) (<0.001) (<0.001) (<0.001) Target s market-to-book 0.044*** 0.044*** 0.047*** 0.047*** (<0.001) (<0.001) (<0.001) (<0.001) Target termination fee (0,1) (0.455) (0.461) (0.727) (0.733) Target lockup (0,1) (0.283) (0.280) (0.415) (0.391) Prior bidding (0,1) * (0.120) (0.106) (0.117) (0.089) Toehold (0,1) (0.933) (0.919) (0.816) (0.810) Stock payment (0,1) 0.324** 0.337** 0.277* 0.306** (0.025) (0.021) (0.062) (0.039) Tender offer (0,1) (0.872) (0.847) (0.656) (0.604) Hostile deal (0,1) (0.132) (0.124) (0.186) (0.155) Same industry (0,1) (0.523) (0.563) (0.169) (0.226) Heckman self-selectivity * (0.392) (0.076) Year and industry fixed effects Yes Yes Yes Yes N Adjusted R Pr>χ 2 <0.001 <0.001 <0.001 <

46 Table 8 Parachute importance and deal completion The sample consists of 851 acquisitions announced during described in Table 2. The dependent variable in the logit models equals one if the proposed merger is ultimately consummated. Target termination fee (0,1) variable equals one if the target has a termination fee provision in the merger contract. Cash payment (0,1) variable equals one if the deal is paid entirely in cash. Regulated industry (0,1) variable equals one if the target s industry belongs to railroads, trucking, airlines, telecommunications, or gas and electric utilities. Other variables are self-explanatory or defined elsewhere. We report White (1980) heteroskedasticity consistent p-values in parentheses. The symbols *, **, and *** denote statistical significance at the 10%, 5%, and 1% levels, respectively. Dependent variable = 1 if the deal is completed (1) (2) (3) (4) Intercept (0.428) (0.439) (0.400) (0.409) Parachute / Merger pay package 1.419** 1.417** (0.025) (0.026) Parachute / Lost compensation 1.363** 1.319** (0.036) (0.043) Target termination fee (0,1) 1.298*** 1.301*** 1.293*** 1.295*** (<0.001) (<0.001) (<0.001) (<0.001) Target lockup (0,1) (0.273) (0.301) (0.348) (0.349) Prior bidding (0,1) *** *** *** *** (<0.001) (<0.001) (<0.001) (<0.001) Cash payment (0,1) (0.743) (0.781) (0.885) (0.895) Tender offer (0,1) 1.301*** 1.322*** 1.363*** 1.369*** (0.009) (0.008) (0.007) (0.007) Hostile deal (0,1) *** *** *** *** (<0.001) (<0.001) (<0.001) (<0.001) Regulated industry (0,1) (0.588) (0.540) (0.814) (0.784) Same industry (0,1) 1.045*** 1.071*** 1.073*** 1.087*** (0.002) (0.002) (0.002) (0.001) Heckman self-selectivity (0.359) (0.616) Year and industry fixed effects Yes Yes Yes Yes N Adjusted R Pr>χ 2 <0.001 <0.001 <0.001 <

47 Table 9 Univariate analyses of acquisition premia The sample consists of 851 acquisitions announced during described in Table 2. Acquisition premium is obtained from SDC and calculated as the offer price divided by the target s stock price four weeks before the merger announcement date. In panel A, we sort mean and median premiums by whether the target firm offers its CEO a golden parachute. In panel B, we extend the analysis by sorting reporting mean and median premiums for quintiles of parachute importance. The symbols *, **, and *** denote statistical significance at the 10%, 5%, and 1% levels, respectively. Panel A: Acquisition premia Firms without parachutes (N=116) Firms with parachutes (N=735) (t- statistic) [Wilcoxon Z] for differences All firms (N=851) Acquisition premium (%) (3.10)*** [31.48] [34.86] [30.47] [2.70]*** Panel B: Parachute importance and acquisition premia Quintile Parachute. Parachute. Actual premium Merger pay package Lost compensation Actual premium [43.00] [33.63] [40.09] [33.19] [34.46] [30.05] [27.57] [31.99] [12.43] [27.38] 46

48 Table 10 Multivariate analyses of acquisition premia The sample consists of 851 acquisitions announced during described in Table 2. In Panel A, the dependent variable in the OLS model is the acquisition premium as reported by SDC. In Panel B, the dependent variable in the OLS models is the actual premium targets earn less the predicted premium we estimate in Panel A. A CEO is near retirement age when s/he is at least 62 years old at the time of the acquisition. Prior year excess return is the cumulative abnormal return during the one year window ending 20 trading days prior to the merger public announcement, calculated from the market model using the CRSP value-weighted return as the benchmark with an estimation period of one year prior to the beginning of the above window. Rumor (0,1) variable equals one if the deal is rumored as reported in SDC. One year change in macroeconomic indicator is the difference in the industrial production index over one year period before the merger. Post SOX (0,1) equals one if the deal is announced after the Sarbanes-Oxley Act is promulgated (8/29/2002). Other variables are selfexplanatory or defined elsewhere. In Panel B, we report White (1980) heteroskedasticity consistent p-values in parentheses. The symbols *, **, and *** denote statistical significance at the 10%, 5%, and 1% levels, respectively. Panel A: Base premium regression OLS model with premium as dependent variable coefficient p-value Intercept 0.411** Target characteristics Log (Assets) *** Market-to-book Leverage 0.097** Liquidity 0.149** Free cash flow Prior year excess return *** <0.001 Target CEO and board characteristics CEO near retirement age (0,1) ** Overconfident CEO (0,1) CEO s equity ownership G index Board ownership Percent of independent directors Deal characteristics Private acquirer (0,1) ** Cash payment (0,1) 0.058** Tender offer (0,1) 0.092*** Hostile (0,1) 0.071* Same industry (0,1) Rumor (0,1) 0.092** Prior bidding (0,1) 0.072** Toehold (0,1) Target termination fee (0,1) 0.051** Year and industry fixed effects Yes N 851 Adjusted R p-value of regression s F test <

49 Panel B: Premium difference regressions Regression models of premium difference (1) (2) (3) (4) Intercept (0.584) (0.388) (0.285) (0.277) Parachute / Merger pay package *** (<0.001) Quintile 2 (0,1) (0.653) Quintile 3 (0,1) * (0.072) Quintile 4 (0,1) *** (<0.001) Quintile 5 (0,1) *** (<0.001) Parachute / Lost compensation ** (0.049) Quintile 2 (0,1) (0.891) Quintile 3 (0,1) (0.868) Quintile 4 (0,1) (0.475) Quintile 5 (0,1) ** (0.048) Target CEO s employment in the combined firm (0,1) (0.910) (0.700) (0.588) (0.581) Target initiated deal (0,1) * * *** *** (0.091) (0.091) (0.006) (0.006) Days to completion (0.423) (0.422) (0.761) (0.795) One year change in macroeconomic indicator (0.422) (0.374) (0.400) (0.399) Prior year market return (0.276) (0.270) (0.154) (0.165) Post-SOX deal (0,1) *** *** *** *** (0.004) (0.003) (0.003) (0.003) Heckman self-selectivity (0.694) (0.697) (0.733) (0.621) Industry fixed effects Yes Yes Yes Yes N Adjusted R p-value of regression s F test <0.001 <0.001 <0.001 <

50 Table 11 The probability of parachute revision In the logit model (1) of parachute augmentation, we use the sample of 851 acquisitions announced during described in Table 2 and the dependent variable equals one if the board increases the parachute multiple for the CEO at the time of the merger. In the logit model (2) of parachute diminution, we use 735 firms that have parachute prior to the merger and the dependent variable equals one if the board decreases the parachute multiple for the CEO at the time of the merger. The excess compensation variable is constructed following Hartzell, Ofek, and Yermack (2004) as the residual from a basic compensation regression model using a panel data of large universe of firms from our sample and Execucomp. We regress salary and bonus against two-digit SIC industry indicator variables, the log of market capitalization, year indicators, and the prior stock return. Other variables are self-explanatory or defined elsewhere. p-values are White (1980) heteroskedasticity consistent. The symbols *, **, and *** denote statistical significance at the 10%, 5%, and 1% levels, respectively. (1) (2) Parachute augmentation (0,1) Parachute diminution (0,1) coefficient p-value coefficient p-value Intercept Target characteristics Log (Assets) *** Market-to-book Leverage G index % independent directors ** ** Family firm (0,1) * Prior year excess return CEO characteristics Excess compensation * * Founder (0,1) Compensation committee member (0,1) Chairman (0,1) * Log (Age) 2.175* Equity ownership (%) * Log (Lost compensation) 0.366*** Deal characteristics Cash payment (0,1) Hostile deal (0,1) Tender offer (0,1) Prior bidding (0,1) * Target initiated deal (0,1) 0.507* * Regulated industry (0,1) Financial industry (0,1) * Heckman self-selectivity *** Year fixed effects Yes Yes N Pr>χ

51 Chart 1: Proportion of merger pay package upon completion N = 851 Chart 2: Parachute revision for firms with parachute before merger N = 735 Chart 3: Parachute revision for firms without parachute before merger N = 116 Figure 1: Proportion of target CEO s merger pay package and parachute revisions In Chart 1, gain in the value of CEO common equity holdings is computed using four-week premiums as measured by SDC. Option gain is calculated similarly using an assumption that all options are at- or in-themoney prior to the transaction. Parachute is estimated based on the target company's last proxy statement filed prior to the transaction. Parachute revision and special bonus represent special merger-related payments negotiated and awarded to CEOs at the time of the acquisition. These calculations follow Hartzell, Ofek, and Yermack (2004). Chart 2 and Chart 3 show the proportion of parachute revision for firms with and without parachute before merger, respectively. 50