Ontheimportanceofgoldenparachutes.pdf

JOURNAL OF FINANCIAL AND QUANTITATIVE ANALYSIS Vol. 48, No. 6, Dec. 2013, pp. 1717–1753 COPYRIGHT 2013, MICHAEL G. FOSTER SCHOOL OF BUSINESS, UNIVERSITY OF WASHINGTON, SEATTLE, WA 98195 doi:10.1017/S002210901300063X

On the Importance of Golden Parachutes

Eliezer M. Fich, Anh L. Tran, and Ralph A. Walkling∗

Abstract

In acquisitions, target chief executive officers (CEOs) face a moral hazard: Any personal gain from the deal could be offset by the loss of the future compensation stream associated with their jobs. Larger, more important parachutes provide greater relief for these losses. To explicitly measure the moral hazard target CEOs face, we standardize the parachute payment by the expected value of their acquisition-induced lost compensation. We examine 851 acquisitions from 1999–2007, finding that more important parachutes benefit target shareholders through higher completion probabilities. Conversely, as parachute importance increases, target shareholders receive lower takeover premia, while acquirer shareholders capture additional rents from target shareholders.

I. Introduction

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 ex- penses 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 (Feb. 4, 2009)1

∗Fich, [email protected], Walkling, [email protected], LeBow College of Business, 3141 Chestnut St, Philadelphia, PA 19104; and Tran, [email protected], Cass Business School, City University London, 106 Bunhill Row, London EC1Y 8TZ, United Kingdom. For very helpful comments, we thank Alberto Banal, Leonce Bargeron, Roland Battie, Jie Cai, Naveen Daniel, David Denis, Diane Denis, Bill Greene, Raj Gupta, Shane Heitzman, Richard Jaffe, Kathy Kahle, Paul Malatesta (the editor), Harold Mulherin, Lalitha Naveen, Micah Officer, Matthew Rhodes-Kropf, Javier Suarez, and David Yermack; members of the Advisory Board of Drexel’s Center for Corporate Governance; sem- inar attendees at the Cass Business School, Drexel University, Erasmus University, Fordham Univer- sity, IESEG School of Management, Syracuse University, State University of New York Binghamton, University of South Florida, and Vlerick Leuven Gent Management School; and session participants at the 2009 Finance Forum held at Instituto de Estudios Superiores de la Empresa (IESE), the 2009 University of Southern California Law School Conference on Empirical Legal Studies, the University of Oregon 2010 Research Conference honoring the scholarly contributions of Larry Dann, and the 2011 meetings of the Klynveld Peat Marwick Goerdeler (KPMG) PhD Project. We also appreciate the constructive comments and insightful reviews provided by an anonymous referee. Fich gratefully acknowledges financial support from the Center for Research Excellence at the LeBow College of Business. All errors are our responsibility.

1The full speech by President Obama can be viewed at http://www.whitehouse.gov/blog post/ new rules/

1717

1718 Journal of Financial and Quantitative Analysis

Golden parachutes are more controversial today than when they first ap- peared over 20 years ago. Advocates argue that parachutes are a necessary part of a competitive pay package required to attract and retain talented executives. It is also argued that 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 com- pany, not to its continuing or past performance. Detractors portray parachutes as guaranteeing managers “pay for failure,” regardless of shareholder returns. Head- lines from the popular press regularly 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. On Jan. 25, 2011, by a 3-2 vote, the Securities and Exchange Commission (SEC) approved an amendment that adds Section 14A to the Securities Exchange Act of 1934, bowing to pressure from institutional investors and other corporate gov- ernance activist groups. Under this amendment, companies soliciting votes to approve a merger, acquisition, or similar business combination need to disclose golden parachute compensation arrangements. The new law also requires these firms to conduct a separate shareholder advisory vote to approve golden parachute compensation.2

The preceding discussion suggests that the controversy surrounding golden parachutes is alive and well. At the heart of the controversy over parachutes is a moral hazard problem: Target chief executive officers (CEOs) have direct influ- ence over actions that provide personal benefit or loss at the possible expense of their shareholders. To address the moral hazard issue in a modern sample of firms, we study 851 acquisition offers during 1999–2007 to learn whether 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 classic themes in the literature: incentive alignment and managerial interest.3

Academic research has greatly enhanced our knowledge of parachutes, but to date, empirical analyses have not explicitly modeled the financial trade-off meet- ing target CEOs. The moral hazard problem is best captured by recognizing the relative takeover-related gains and losses experienced by the person (arguably) most responsible for the completion and terms of a merger: the target CEO. Con- sequently, we re-examine existing hypotheses on a recent sample of acquisitions using a measure of parachute importance that mirrors the moral hazard the target CEO encounters. It scales the parachute payment by the expected pay loss this CEO incurs if the merger is completed.

Our tests reveal that a 1-standard-deviation increase in parachute impor- tance is associated with an increase of 6.9 percentage points in deal completion. Our tests also indicate that parachute provisions affect the wealth of target CEOs

2The new rules affect Section 14d-10(a)(2) of the 1934 Securities Act, which provides a safe harbor enabling the compensation committee of a target’s board of directors to grant golden parachutes or other benefits to its executives during a tender offer negotiation. The SEC approved the safe harbor provision on Oct. 18, 2006.

3Incentive alignment and managerial interest are hypotheses often studied in settings prone to agency problems (see, e.g., Jensen and Meckling (1976)).

Fich, Tran, and Walkling 1719

and target shareholders in a nontrivial manner. On average, target CEOs cash in about $4.9 million from parachutes when their firms are sold. Conversely, a 1-standard-deviation increase in parachute importance is associated with a drop in premia of about 2.6 percentage points. This shortfall implies a reduction of $127 million in deal value for the average transaction in our sample.

Given the effect of parachutes on both merger completion probabilities and takeover premia, we examine whether it makes sense for target CEOs to accept a lower premium (even with a larger parachute) because the value of their target- equity-based portfolio (which depends on the takeover premium) will decline. Similarly, is it logical for shareholders to provide a parachute to their CEO if this benefit might make them worse off in case of a merger?

To address rationality concerns related to target shareholders, we follow the method in Comment and Schwert (1995) and estimate an unconditional premium regression. We find that the unconditional premium is a positive function of the presence of a golden parachute. This result indicates that including a parachute provision in the CEO’s compensation contract is associated with a net gain to shareholders. This finding is significant not only because it shows that it is indeed rational for shareholders to provide a parachute to their CEOs but also because it suggests that what really matters (during mergers) is the parachute’s relative importance, not its mere presence.

In our sample, the unconditional probability of deal completion is 87.8% and the mean takeover premium offered is around 35.9%. As noted previously, a 1-standard-deviation increase in parachute importance raises the probability of merger completion by 6.9 percentage points but lowers the takeover premium by 2.6 percentage points. These estimates imply that the expected appreciation accruing to the target CEO’s equity-based portfolio is the same (at 31.5%) with or without such an increase in parachute importance. Given this evidence, the ac- tions of target CEOs who get more important parachutes appear perfectly rational. Interestingly, these results also imply that the expected premium to target share- holders is essentially the same even with an increase in parachute importance. This raises the possibility that target shareholders are not really hurt by more important parachutes. In fact, risk-averse shareholders might prefer the same ex- pected payoff with less risk (higher probability of deal completion). In a similar fashion, a certainty equivalent argument can explain the actions of target CEOs in settling for certain lower premia (and the consequent triggering of their merger pay package) rather than bargaining for higher premia at a possible risk to deal completion. That is, the negotiated premium represents the target CEO’s own reservation premium, which provides this executive with a certainty equivalent of his or her lost compensation.

We also analyze the investor reactions to the acquisition announcement of the publicly traded bidders in our sample. These tests reveal that as the impor- tance of the parachute to target CEOs increases, bidding firms earn higher merger announcement returns. This finding indicates that deals in which the target CEO gets a relatively more important parachute exhibit a wealth transfer from share- holders of the target to shareholders of the buyer.

We identify a number of empirical issues that could raise concerns related to endogeneity or to other econometric biases. First, parachutes are endogenously

1720 Journal of Financial and Quantitative Analysis

chosen, which introduces the potential of self-selection bias. Second, since firms do not randomly become takeover targets, our analyses might be vulnerable to sample selection bias. Third, because industry and/or time trends could affect the incidence of mergers and the way executive pay is structured, our tests might be prone to an omitted variables bias. Fourth, since parachutes are common pro- visions in many compensation contracts, their effects might be anticipated and impounded in a target’s price. Accordingly, our analysis could be susceptible to anticipation bias. Fifth, foreknowledge of the premium a potential target could command in the event of a takeover might dictate how that firm structures a merger-related parachute for its CEO. Under this scenario, the direction of causal- ity would be reversed.

To address the issues described above, we use different empirical specifica- tions and econometric methods. Our multivariate tests control for self-selection endogeneity with the Heckman (1979) approach. We use the same procedure to address sample selection issues by controlling for the probability that a firm be- comes a takeover target. Also, to account for anticipation bias, we employ the multistage process in Comment and Schwert (1995) and divide parachutes into predictable and surprise components. To control for an omitted variables bias, our multivariate tests include year and industry fixed effects. To consider reverse causality concerns, we estimate several two-stage instrumental variable systems. The inverse association between parachute importance and premia remains under the different empirical specifications and econometric techniques we employ. In addition, our results are robust to alternative parachute proxies, including a mea- sure of parachute importance that scales its value by the value of the merger pay package received by the target CEO.

Aside from the econometrics issues noted above, it is possible that the results herein obtain because the bargaining power of targets offering more important parachutes is low and not because their CEOs give away rents. To assuage such concern, we add controls that potentially capture the target’s bargaining power. Rhodes-Kropf and Kadyrzhanova (2011) argue that characteristics (such as the level of industry concentration) that allow managers to delay takeovers have a significant bargaining effect. Consequently, our Heckman (1979) selection equa- tion of the probability of becoming a target controls for the Herfindahl-Hirschman index to proxy for the firm’s power in its own industry. Additionally, our mul- tivariate tests control for target-initiated deals because the results in Aktas, de Bodt, and Roll (2010) suggest that this variable is a reasonable proxy for the tar- get’s bargaining power. Our regressions also include input-output/sales-purchases (independent) variables between the target and the acquirer industries similar to those in Ahern (2012). He argues that these customer-supplier variables capture the market power of the parties to an acquisition and, therefore, help account for the role of product markets on bargaining outcomes in mergers. Our results are robust to these different controls for bargaining power.

Our work provides a better understanding of the wealth effects and incentives of merger-related exit pay to target CEOs. This evidence is relevant in the ongo- ing policy debate regarding best practices in corporate governance. Our results are consistent with the following interpretation: As the importance of the parachute to target CEOs increases, they negotiate an offer up to their own reservation

Fich, Tran, and Walkling 1721

premium, which provides them with a certainty equivalent that is proportional to their expected lost compensation. At the same time, acquirers experience higher returns, which might be a manifestation of the lower premium. Therefore, condi- tional on receiving a bid, i) target CEOs are partially made whole for their per- sonal losses, ii) target shareholders are worse off, and iii) bidder shareholders are better off. This evidence appears consistent with the managerial interest hypoth- esis of golden parachutes. Nonetheless, this interpretation of our findings ignores the fact that parachutes also increase the probability of receiving and complet- ing a bid and, thus, increase the welfare of target shareholders. Once this factor is considered, it is possible that target shareholders are better off (they obtain a completed bid they would not have otherwise received), and bidder sharehold- ers are also better off because they get a good deal conditional on making a bid. Under this interpretation of our findings, more important parachutes align the incentives of target shareholders and target CEOs: These executives achieve their own interests while still completing advantageous deals for their shareholders.

This paper contributes to the literature as follows: First, we provide an up- dated analysis on an unresolved topic. To our knowledge, even recent published papers on parachutes (Hartzell, Ofek, and Yermack (2004), Bange and Mazzeo (2004)) use samples ending in 1997 and 1990, respectively. Because the last 10–15 years have arguably witnessed the most dramatic changes in corporate gov- ernance in history, analyzing parachutes in the current decade is important.4 As the president’s recent comments indicate, golden parachutes remain a controver- sial tool of corporate governance.

Second, we develop a new measure of the importance of parachutes. It re- flects the moral hazard issue faced by target CEOs, generally the single most im- portant executive in merger negotiations. Our measure, which scales the parachute payment by the expected pay loss this CEO incurs if the merger is completed, is unlike those in the extant literature. Indeed, existing studies in this literature either control for the presence of a parachute or assess the increased acquisition costs related to the parachute. However, none measures the importance of the parachute to the target CEO.5 We show that the certainty equivalent of the lost compen- sation to target CEOs is proportional to the expected value of that compensa- tion. This result indicates that our measure of parachute importance is unique in that it captures the incentives CEOs face when their firms become acquisition targets.

Third, existing studies focus on the impact parachutes have on the perfor- mance of the firms granting these benefits. We advance this literature by also examining the potential effect of the parachute given to the target CEO on the return to shareholders in the acquiring firm.

4Cheffins (2009) chronicles numerous governance regulatory changes occurring in the United States during 1990–2007.

5Among published papers in the literature studying parachutes, Knoeber (1986), Denis and Serrano (1996), Cotter, Shivdasani, and Zenner (1997), Evans, Noe, and Thornton (1997), Agrawal and Knoeber (1998), Hartzell et al. (2004), and Bange and Mazzeo (2004) use dummy variables to capture the presence of a parachute. Others studies like Lambert and Larcker (1985), Machlin, Choe, and Miles (1993), and Lefanowicz, Robinson, and Smith (2000) divide the value of the parachute by the target’s market value of equity.

1722 Journal of Financial and Quantitative Analysis

Fourth, our empirical evidence supports the theoretical prediction in Ross (2004) that the overall structure of a pay schedule (even one markedly convex) could result in more (instead of less) risk aversion. Ross argues that attitudes toward risk depend not only on the convexity of an agent’s compensation sched- ule, but also on how the overall schedule maps into more (or less) risk-averse regions of the agent’s utility function to the extent it can undo the impact of con- vex (or concave) pay schedules. Our findings suggest that the relative importance of the parachute curtails the convexity that equity-based pay imposes upon the tar- get CEO’s utility function. Importantly, under this interpretation, our results offer a plausible answer to a paradox in the literature showing that target CEOs often accept lower premia in exchange for benefits (like unscheduled option grants (Fich, Cai, and Tran (2011)), augmented parachutes or bonuses (Hartzell et al. (2004)), or jobs in the merged firm (Wulf (2004))) that are unlikely to fully cover their merger-related personal losses.

The paper proceeds as follows: Section II describes our data. Section III contains our empirical analyses. Section IV addresses a number of robustness issues. Section V concludes.

II. Data and Sample Characteristics

We begin with a base sample of 4,381 mergers and acquisitions (M&A) an- nounced during 1999–2007 and tracked in the Securities Data Company’s (SDC) M&A database. We require the target to be a publicly traded U.S. firm and exclude spinoffs, recapitalizations, exchange offers, repurchases, self-tenders, pri- vatizations, acquisitions of remaining interest, partial interests or assets, and trans- actions in which deal value is less than $1 million. From this group, we keep 3,521 deals in which targets have stock return and accounting data available from the Center for Research in Security Prices (CRSP) and Compustat, respectively. We lose 278 deals because premium data are missing from SDC and from other sources such as CRSP, LexisNexis, or Factiva. After filtering out deals in which governance data for targets are not available from RiskMetrics, our final sample consists of 851 offers.

A. Target and Deal Characteristics

We read the S-4, DEFM14A, SC-TO, and DEF14A proxies filed with the SEC by the target and/or acquiring firm. From these proxies, we obtain informa- tion on the sale procedure, the party that initiates the deal, and the date merger negotiations begin. Panel A of Table 1 reports the offer characteristics in our sam- ple. Among the 851 transactions, about 18% are tender offers and 7% are hostile takeovers. These statistics compare favorably to those in Officer (2003). His sam- ple of acquisitions during 1988–2000 consists of about 20% tender offers and 8% hostile deals. Similar to Moeller, Schlingemann, and Stulz (2005), almost 55% of the transactions in our sample are paid in cash. The deals we study have a com- pletion rate close to 88%, which is comparable to that of Officer, who reports a completion rate of 83%. We follow the procedure in Boone and Mulherin (2007) to identify the start of merger negotiations and the party responsible for initiating

Fich, Tran, and Walkling 1723

the deal. We find that in over 39% of all deals the target firm initiates the sale. Aktas et al. (2010) find that in about 42% of the cases they study, target firms initiate the merger. Grinstein and Hribar (2004) report a mean deal value of $4.7 billion for the transactions they examine, which is similar to the $4.76 billion mean value in our sample.

Panel B of Table 1 contains key financial characteristics for the target firms in our sample. The average (median) market value of equity is $3.302 billion ($0.991 billion), and leverage accounts for 26% (25%) of total assets. These

TABLE 1

Sample Description

Table 1 describes our sample, which consists of 851 mergers and acquisitions announced during 1999–2007 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 RiskMetrics (formerly the Investor Responsibility Research Center) database, respectively. In Panel A, deal status, mode of acquisition, method of payment, and deal attitude are obtained from SDC. As in Officer (2003), we classify a deal as a hostile takeover if the bid is recorded by SDC as “hostile” or “unsolicited.” 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 debt divided by market value of assets. Deal value is obtained from SDC. In Panel C, 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. In Panel D, compensation data are as of the end of the fiscal year before the announcement date. Estimated lost compensation is the estimated present value of the CEO’s lost compensation when his/her firm is sold as in Fich et al. (2011). We obtain information on the golden parachute payment from the last proxy filed by the targets prior to the merger announcement, the S-4 proxy filed by the acquirers, and/or the DEFM14A proxy filed by the targets following the merger announcement.

Mean Median

Panel A. Deal Characteristics

Completion (0, 1) 0.878 Tender offer (0, 1) 0.182 Stock payment (0, 1) 0.162 Cash payment (0, 1) 0.549 Hostile takeover (0, 1) 0.069 Auction (0, 1) 0.337 Target-initiated deal (0, 1) 0.393 Same industry (0, 1) 0.561 Deal value ($ billion) 4.758 1.544

Panel B. Target Characteristics

Market value ($ billion) 3.302 0.991 Market-to-book 1.734 1.422 Leverage 0.260 0.248

Panel C. Target CEO Characteristics

Chairman (0, 1) 0.570 Founder (0, 1) 0.128 Compensation committee member (0, 1) 0.013 Age (years) 54.390 55.000 Tenure (years) 7.165 4.786 Ownership (%) 4.632 1.836 Market value of ownership ($ million) 96.079 22.728

First Third Mean Quartile Median Quartile

Panel D. Target CEO Compensation and Golden Parachute Characteristics

Salary and bonus ($ million) 1.662 0.636 0.940 1.525 Total compensation ($ million) 5.366 1.170 2.615 5.022 Parachute (0, 1) 0.864 Parachute multiple 2.225 2.000 2.999 3.000 Parachute value ($ million) 4.873 1.482 2.553 4.573 Lost compensation ($ million) 39.896 7.501 16.387 36.524

1724 Journal of Financial and Quantitative Analysis

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%. Targets in our sample have a median market- to-book ratio of 1.42, which is close to the median ratio of 1.69 reported by Officer (2003) for the same variable.

B. Target CEO Characteristics

In Panel C of Table 1, we report the target CEO’s characteristics. On average, 57% of all CEOs also chair their boards and almost 13% are their firm’s 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 concur with those in Hartzell et al. (2004), who report the following CEO statistics: mean age of 54, average equity ownership of 3.6%, and median tenure of 5 years.

We collect compensation data from proxy statements filed by each target with the SEC. In some instances, we supplement these data with information in the ExecuComp database. Key compensation characteristics for target CEOs in our sample appear in Panel D of Table 1. Bebchuk and Grinstein (2005) report an average of $5.01 million in total CEO compensation.6 During the last year in office prior to the deal, the average CEO in our sample earns about $5.4 million in annual total pay.

C. Lost Compensation

CEOs who sell their firms forfeit the compensation they would earn if they were to remain in office. We follow the methodology and assumptions in Yermack (2004) and in Fich et al. (2011) to calculate the expected lost compensation for the target CEOs in our sample. First, we use information on their current com- pensation, their restricted stock, and their option holdings as reported in proxy statements before the merger announcement. Second, we assume that all CEOs retire by age 65 and that CEOs who are at least 65 years old expect to stay in office 1 more year before retiring. Third, we assume that the probability of depar- ture increases by 4% each year due to acquisitions, delistings, or other turnover reasons. Fourth, we assume that salary and bonus increase by 2% from that re- ceived during the year prior to the acquisition when firm performance is above the Fama and French (1997) median industry return on assets. Fifth, we assume that the probability of departure increases by an additional 2% if the company performs below the industry median. Finally, we use a real rate of 3% to discount cash flows. Fich and Shivdasani (2007) estimate that the present value of lost in- come for CEOs expected to remain in office for another 7 years is $45.5 million. On average, the present value of the expected lost compensation for target CEOs in our sample is close to $40 million. Given our estimates, it appears that

6Specifically, they report an average total compensation of $9.41 million for CEOs of Standard & Poor’s (S&P) 500 firms, $3.94 million for CEOs of MidCap 400 firms, and $2.05 million for CEOs of SmallCap 600 firms during 1993–2003.

Fich, Tran, and Walkling 1725

employment termination due to a takeover triggers nontrivial wealth losses for target CEOs.

D. Parachute Provisions for Target CEOs

Many boards of directors provide parachutes to their CEOs. We obtain in- formation on these provisions from the last proxy filed by the targets prior to the merger announcement, the S-4 proxy filed by the acquirers, and/or the DEFM14A proxy filed by the targets following the merger announcement. Among the 851 targets, 735 (or about 86%) have a golden parachute in place for their CEOs be- fore merger negotiations begin. From the target CEO’s employment agreement, we are able to estimate the size of the parachute. Specifically, when a parachute is provided, the employment agreement often stipulates that the parachute pay- ment is based on a multiple of the executive’s regular cash compensation. Panel D of Table 1 shows that the mean (median) parachute payment is $4.87 million ($2.55 million).7

Section 280G of the Internal Revenue Code states, “If the present value of a change-in-control payment (golden parachute) exceeds the safe harbor (three times the average taxable compensation over the 5 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.” Given this tax rule, it would be reasonable to assume that most firms would set the multiple used to value a golden parachute to 3. Con- sistent with this assumption, the information in Panel D of Table 1 indicates that at least 75% of our target firms use a multiple of 3 or lower to value a parachute. Nonetheless, in our sample, the highest parachute valuation multiple equals 5.25.

E. Temporal and Industrial Distribution of Parachute Importance

As noted earlier, we measure the relative importance of golden parachutes to target CEOs by dividing the value of the parachute by the compensation these executives expect to forego when their firms are acquired. In Panels A and B of Table 2, we show the distribution of parachute importance in our sample over time and across industries, respectively. Our parachute importance measure appears generally stable over time, albeit slightly larger in 2002.

The information in Panel A of Table 2 also shows that the annual number of mergers is higher at the beginning and at the end of our sample period, which coincides with periods of economic expansion when the stock market valuation is higher. Shleifer and Vishny (2003) and Rhodes-Kropf and Viswanathan (2004) theorize that stock market health drives merger activity. The temporal distribution of our sample appears in line with their predictions.

7It is important to emphasize that parachute payments might be subject to either a “single trigger” or a “double trigger” provision. Under a single trigger, the CEO obtains the parachute payment because a change in control occurs or because he or she is terminated without cause. Under a “double trigger,” the CEO receives payment if he or she is terminated without cause or quits for good reason after the change in control. Our results continue to hold when we control for whether a single or double trigger is necessary to obtain the parachute payment.

1726 Journal of Financial and Quantitative Analysis

TABLE 2

Parachute Importance

The sample consists of 851 acquisitions announced during 1999–2007 described in Table 1. In Panel A of Table 2, we provide the temporal distribution of our sample. In Panel B, we report the industrial classification of the deals we study using the Fama French (1997) 12-industry classification. Both panels provide information about our proxy for parachute importance. We measure the importance of the parachute for the target CEO as Parachute/Lost Compensation.

Panel A. Temporal Distribution

Parachute/Lost Compensation

Year N % Mean Median

1999 160 18.80 0.283 0.122 2000 132 15.51 0.255 0.119 2001 69 8.11 0.214 0.145 2002 29 3.41 0.310 0.113 2003 46 5.41 0.254 0.124 2004 77 9.05 0.201 0.112 2005 97 11.40 0.226 0.117 2006 121 14.22 0.226 0.135 2007 120 14.10 0.291 0.125

Panel B. Industrial Classification

Parachute/Lost Compensation

Industry N % Mean Median

Nondurable consumer goods 44 5.17 0.262 0.104 Durable consumer goods 23 2.70 0.201 0.160 Manufacturing 94 11.05 0.311 0.158 Energy 43 5.05 0.279 0.146 Chemical 18 2.12 0.556 0.137 Business equipment 171 20.09 0.171 0.071 Telecommunication 34 4.00 0.327 0.119 Utilities 49 5.76 0.294 0.213 Shops 85 9.99 0.235 0.151 Health 76 8.93 0.166 0.112 Finance 112 13.16 0.359 0.156 Other 102 11.99 0.189 0.120

The industrial distribution of our sample (reported in Panel B of Table 2) is also similar to that reported in the existing M&A literature and to the actual distribution in the base sample from SDC. For example, Officer (2003) reports that 2% of his sample are firms in durable consumer goods, 17.4% in business equipment, 7.8% in shops, and 4.6% in chemicals. The percentage of targets in our sample that belong to those same industries is quite similar: 2.7%, 20.1%, 10%, and 2.1%, respectively. In addition, the base acquisition sample from SDC has 22.6% of targets in business equipment, 3.8% in telecommunications, and 8.9% in the healthcare industry. Analogously, the incidence in our final sample is 20.1%, 4%, and 8.9% for those same industries, respectively.

III. Empirical Analyses

A. Determinants of Parachute Importance

In Table 3, we run three Tobit models to study the importance of parachutes for target CEOs. We run Tobit models because the dependent variable (the ratio of the parachute’s size to the present value of lost pay to the CEO) is left-hand censored. The regressions control for target firm, target CEO, and target firm

Fich, Tran, and Walkling 1727

governance characteristics that could affect the relative importance of parachutes; these are defined in the legend accompanying Table 3. All models include year and industry fixed effects.

Our results indicate that the relative importance of parachutes for target CEOs decreases in larger firms. In addition, the marginal effect implied by our

TABLE 3

Determinants of Parachute Importance

The sample consists of 851 acquisitions announced during 1999–2007 described in Table 1. The dependent variable in both Tobit models is Parachute/Lost Compensation. 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. High R&D (0, 1) equals 1 if the target’s industry is in the top quartile of all industries sorted annually by industry-median R&D scaled by assets (similar to the method used by Masulis et al. (2007)). G index is constructed by adding 24 antitakeover provisions tracked by RiskMetrics as in Gompers et al. (2003). As in Hartzell et al. (2004), a CEO is near retirement age when s/he is at least 62 years old at the time of the acquisition. Tenure is the number of years the CEO has been in the chief executive position until the merger announcement date. Insider ownership and institutional ownership are the percentages of common stock owned by each group, respectively. Percent of independent directors is the fraction of independent directors on board. All ownership variables are measured as a percentage of common equity. Other variables are self-explanatory or defined elsewhere. We report White (1980) heteroskedasticity-consistent p-values in parentheses. *, **, and *** indicate statistical significance at the 10%, 5%, and 1% levels, respectively.

Dependent Variable = Parachute/Lost Compensation

Model 1 Model 2 Model 3

Intercept −5.489*** −5.382*** −5.382*** (0.001) (0.001) (0.001)

Target Characteristics log(Assets) −0.023* −0.028** −0.036**

(0.066) (0.031) (0.017)

Q −0.014 −0.013 −0.003 (0.351) (0.414) (0.831)

Leverage 0.087 0.107 0.139* (0.218) (0.143) (0.093)

Free cash flow −0.084 −0.078 −0.022 (0.657) (0.677) (0.899)

log(Firm age) −0.023 −0.030 −0.005 (0.285) (0.174) (0.790)

Prior year excess return −0.175 −0.193 −0.149 (0.380) (0.369) (0.530)

High R&D (0, 1) 0.244 0.189 0.097 (0.560) (0.650) (0.802)

CEO Characteristics Founder (0, 1) −0.168*** −0.153*** −0.143***

(0.001) (0.004) (0.003)

Compensation committee member (0, 1) 0.172 0.181 0.194 (0.177) (0.158) (0.101)

Number of outside directorships −0.023 −0.025 −0.038 (0.463) (0.431) (0.193)

Chairman (0, 1) 0.000 −0.013 −0.009 (0.991) (0.687) (0.756)

log(Age) 1.491*** 1.455*** (0.001) (0.001)

Near retirement age (0, 1) 0.653*** (0.001)

Tenure 0.009*** 0.010*** 0.010*** (0.001) (0.001) (0.001)

Ownership 0.001 0.000 0.001 (0.385) (0.947) (0.389)

Option value/Total compensation −0.279*** −0.281*** −0.269*** (0.001) (0.001) (0.001)

(continued on next page)

1728 Journal of Financial and Quantitative Analysis

TABLE 3 (continued)

Determinants of Parachute Importance

Dependent Variable = Parachute/Lost Compensation

Model 1 Model 2 Model 3

Governance Characteristics G index (minus parachute) 0.011* 0.010*

(0.090) (0.092)

Pct of independent directors 0.061 0.093 (0.506) (0.273)

Insider ownership (excluding CEO) −0.002 −0.002 (0.214) (0.198)

Institutional ownership 0.001 0.002 (0.210) (0.168)

Year and industry fixed effects Yes Yes Yes N 851 851 851 Adj. R 2 0.258 0.264 0.283 Pr > χ2 0.001 0.001 0.001

estimates indicates that the importance of parachutes decreases by 14.3 percent- age points when the target CEO is also the firm’s founder. Other estimates imply that parachute importance increases by about 0.9 percentage points with a 1-standard-deviation increase in the Gompers, Ishii, and Metrick (2003) G index: Firms with greater takeover defenses are more likely to give their CEOs greater parachutes in case of a merger. In addition, according to model 3 of Table 3, parachute importance increases for target CEOs aged 62 or older.

B. Parachute Importance and Merger Completion

Golden parachutes might be a symptom of managerial entrenchment. In fact, the presence of a parachute is one of the 24 antitakeover provisions tracked by RiskMetrics and indexed by Gompers et al. (2003). Given this, golden parachutes may increase a firm’s ability to defeat a takeover attempt (Malatesta and Walkling (1988)). Nonetheless, the empirical evidence related to the parachutes’ effect on takeover probability is mixed.8

In Table 4, we examine the relation between parachute importance and deal completion. One presumes that completed deals are beneficial to target share- holders, 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. Hence, in Table 4, we report the estimation of two variants of a fixed effects logit model in which the dependent variable equals “1” for completed deals and “0” for with- drawn deals. Officer (2003) and Bates and Lemmon (2003) estimate similar mod- els. Therefore, the control variables in our regressions are similar to theirs. The exception, of course, is our proxy of parachute importance.

The tests in Table 4 also include control variables to proxy for the poten- tial bargaining power of the parties to the deal. We add a dummy variable for

8For instance, whereas Cotter and Zenner (1994) do not find an association between parachutes and the likelihood of a successful takeover, Machlin et al. (1993) and Bebchuk, Cohen, and Wang (2010) do.

Fich, Tran, and Walkling 1729

TABLE 4

Parachute Importance and Deal Completion

The sample consists of 851 acquisitions announced during 1999–2007 described in Table 1. The dependent variable in the logit models equals 1 if the proposed merger is ultimately consummated. The key independent variable in both models is (Parachute/Lost compensation). Target termination fee (0, 1) equals 1 if the target has a termination fee provision in the merger contract. Cash payment (0, 1) equals 1 if the deal is paid entirely in cash. Regulated industry (0, 1) equals 1 if the target’s industry belongs to railroads, trucking, airlines, telecommunications, or gas and electric utilities. Target input/Total acquirer output is the industry-level percentage of dollars of target industry input for each acquirer industry output dollar. Target purchases/Total acquirer sales is the percentage of all acquirer industry sales purchased by the target industry. As in Ahern (2012), we calculate these two measures of customer-supplier relationship between the target and the acquirer using data from the U.S. Bureau of Economic Analysis Input-Output “Use” and “Make” tables. The Parachute Heckman (1979) lambda and the Target Heckman lambda involve a first-stage estimation of the probability of having a golden parachute and the probability of becoming a target as in models 3 and 4 of Tables A1 and A2, respectively. In the second stage, the inverse Mills ratio from the first-stage model is included in the estimation as a variable to control for endogenous self- selection. Other variables are self-explanatory or defined elsewhere. We report White (1980) heteroskedasticity-consistent p-values in parentheses. *, **, and *** indicate statistical significance at the 10%, 5%, and 1% levels, respectively.

Dependent Variable = 1 If the Deal Is Completed

Model 1 Model 2

Intercept −1.194 −1.251 (0.632) (0.607)

Parachute/Lost compensation 1.574** 1.571** (0.028) (0.030)

Target termination fee (0, 1) 1.442*** 1.456*** (0.001) (0.001)

Target lockup (0, 1) −0.517 −0.552 (0.645) (0.621)

Prior bidding (0, 1) −2.502*** −2.524*** (0.001) (0.001)

Cash payment (0, 1) 0.162 0.169 (0.704) (0.691)

Tender offer (0, 1) 1.345*** 1.352*** (0.009) (0.008)

Hostile deal (0, 1) −3.010*** −2.991*** (0.001) (0.001)

Regulated industry (0, 1) −0.383 −0.406 (0.699) (0.680)

Same industry (0, 1) 1.060*** 1.046*** (0.003) (0.004)

Target-initiated deal (0, 1) 0.146 0.156 (0.666) (0.645)

Target input/Total acquirer output −0.458 −0.449 (0.931) (0.933)

Target purchases/Total acquirer sales 0.090 0.065 (0.984) (0.988)

CEO near retirement (0, 1) −0.393 −0.342 (0.465) (0.522)

CEO-chairman (0, 1) 0.170 0.146 (0.631) (0.676)

CEO equity ownership −0.013 −0.013 (0.506) (0.517)

log(Target’s assets) −0.262** −0.256** (0.032) (0.039)

Parachute Heckman lambda −0.146 (0.576)

Target Heckman lambda −0.088 (0.731)

Year and industry fixed effects Yes Yes N 851 851 Adj. R 2 0.451 0.451 Pr > χ2 0.001 0.001

1730 Journal of Financial and Quantitative Analysis

target-initiated deals. Following Ahern (2012), we also include input-output/ sales-purchases variables (for the target and acquirer industries). Ahern notes that these variables control for the effect of product markets on bargaining outcomes in mergers.

Because golden parachutes are endogenously determined, in model 1 of Table 4 we control for endogenous self-selection by using the Heckman (1979) inverse Mills ratio (λ1). Moreover, since firms do not randomly become takeover targets, in model 2 we control for sample selection by using a different inverse Mills ratio (λ2) based on a regression of the probability of becoming an acquisi- tion target.9

Results for the control variables in Table 4 are consistent with those in the ex- isting M&A literature. Transactions are about 9.5 percentage points more likely to materialize if there is a target termination fee. This marginal effect is comparable to that of 11 percentage points in Officer (2003). Tender offers are 4.2 percentage points more likely to go through, as are mergers in which the parties to the trans- action are in the same industry. As in Bates and Lemmon (2003), deals are less likely to be completed if there is prior bidding or if the deal is hostile.

Of primary interest is the result that deal completion increases with the im- portance of the parachute. The marginal effect implied by the estimates in Table 4 indicates that a 1-standard-deviation increase in parachute importance raises the probability of deal completion by 6.9 percentage points. This finding could be consistent with the incentive alignment hypothesis in that larger parachutes moti- vate target CEOs to complete the deal. Target CEOs care about deal completion because they can cash in their parachutes and their equity-based portfolio in full, since all restrictions and vesting periods disappear when the target firm ceases to exist as a stand-alone entity.

C. Parachute Importance and Acquisition Premia

Payments under parachute provisions can strengthen a target’s bargaining position with the bidding firm (Comment and Schwert (1995)). However, stud- ies examining the association between parachutes and the premia paid for target firms provide mixed evidence. For example, Cotter and Zenner (1994), Bange and Mazzeo (2004), and Lefanowicz et al. (2000) find no association, while Bebchuk et al. (2010) report an inverse association. Hartzell et al. (2004) examine situa- tions in which the value of the parachute to the target CEO is augmented prior to deal completion. They show that such augmentations are not associated with the premium paid for the target company. None of these studies, however, defines the parachute in terms specifically related to the moral hazard dilemma the target CEO confronts: the gain from the parachute relative to the expected loss of future pay to this executive if the deal is completed.

9The parachute Heckman (1979) self-selection and the target Heckman self-selection involve a first-stage estimation of the probability of having a golden parachute and the probability of being a target, respectively. We report these first-stage models, both of which are estimated in a sample of 14,157 firm-years, in Tables A1 and A2 of the Appendix. In the second stage, the inverse Mills ratio derived from the first-stage model is included in the estimation as a variable to control for endogenous self-selection.

Fich, Tran, and Walkling 1731

We use the 4-week premium reported by SDC as the dependent variable in a set of eight regressions similar to those in Bargeron, Schlingemann, Stulz, and Zutter (2008).10 These premium regressions are reported in Table 5. The indepen- dent variables of interest are four different proxies based on the golden parachute payment to the target CEO. These variables are: in model 1, the value of the parachute divided by the present value of the expected lost compensation to the target CEO; in model 2, a dummy variable set to “1” if the CEO’s compensation

TABLE 5

Golden Parachutes and Acquisition Premia

The sample consists of 851 acquisitions announced during 1999–2007 described in Table 1. The dependent variable in the ordinary least squares (OLS) models is the acquisition premium as reported by SDC, which is calculated as the offer price divided by the target’s stock price 4 weeks before the merger announcement date. Model 1 uses the parachute importance relative to the expected lost compensation to the target CEO as the main independent variable. Model 2 uses the parachute (0, 1) as the key independent variable. The independent variable of interest in model 3 is the natural log of the parachute payment to the target CEO. The main independent variable in model 4 is the parachute multiple. Prior year excess return is the cumulative abnormal return during the 1-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 1 year prior to the beginning of the above window. Overconfident CEO (0, 1) is defined as Malmendier and Tate’s (2005) long-holder measure and follows Hall and Liebman’s (1998) option classification procedure. It equals 1 if the target firm’s CEO owns options at the beginning of the last year of the options’ life that are at least 40% in the money. Target CEO post-deal employment (0, 1) equals 1 if the target CEO already holds or obtains either a directorship or an executive appointment such as CEO of the acquirer or a subsidiary, chief financial officer, chief operating officer, chairman, vice-chairman, president, or vice-president in the bidder firm after deal completion. In case of withdrawn deals, it equals 1 if the target CEO already holds any of the positions just described or if the merger proxy states that the target CEO will be employed by the bidder upon deal completion. Rumor (0, 1) equals 1 if the deal is rumored as reported in SDC. Litigation (0, 1) equals 1 if the deal has associated litigation reported in SDC. Time to completion measures the number of days to close the transaction from the time it is announced. The 1-year change in IP index is the difference in the industrial production index over a 1-year period before the merger. Other variables are self-explanatory or defined elsewhere. All regressions include year and industry fixed effects. We report White (1980) heteroskedasticity-consistent p-values in parentheses. *, **, and *** indicate statistical significance at the 10%, 5%, and 1% levels, respectively.

Model 1 Model 2 Model 3 Model 4

A B A B A B A B

Intercept 0.305 0.292 0.519*** 0.531*** 0.495*** 0.508*** 0.483** 0.494*** (0.324) (0.345) (0.007) (0.006) (0.010) (0.008) (0.012) (0.010)

Golden Parachute (GP) Measures GP/Lost compensation −0.058** −0.060**

(0.023) (0.017)

Parachute (0, 1) −0.063** −0.062** (0.032) (0.026)

log(Parachute value) −0.007* −0.007** (0.050) (0.040)

Parachute multiple −0.018* −0.018** (0.052) (0.045)

Target Characteristics log(Assets) −0.020** −0.016** −0.016* −0.013 −0.014* −0.011 −0.013* −0.011

(0.010) (0.048) (0.051) (0.116) (0.090) (0.183) (0.095) (0.185)

Q −0.011 −0.013 −0.010 −0.011 −0.010 −0.011 −0.011 −0.012 (0.283) (0.219) (0.327) (0.282) (0.316) (0.271) (0.280) (0.240)

Leverage 0.095** 0.096** 0.066 0.066 0.067 0.067 0.069 0.069 (0.049) (0.047) (0.184) (0.185) (0.177) (0.178) (0.167) (0.166)

Free cash flow −0.080 −0.075 −0.097 −0.094 −0.087 −0.084 −0.089 −0.086 (0.512) (0.538) (0.451) (0.463) (0.498) (0.511) (0.489) (0.502)

Liquidity 0.169*** 0.174*** 0.116* 0.120* 0.112* 0.117* 0.114* 0.118* (0.006) (0.005) (0.061) (0.052) (0.070) (0.060) (0.065) (0.056)

Prior year excess return 0.088*** 0.088*** 0.099*** 0.098*** 0.098*** 0.098*** 0.098*** 0.097*** (0.001) (0.001) (0.001) (0.001) (0.001) (0.001) (0.001) (0.001)

(continued on next page)

10Following Officer (2003), we restrict this premium measure to 2 (or 200%) to avoid extreme outliers.

1732 Journal of Financial and Quantitative Analysis

TABLE 5 (continued)

Golden Parachutes and Acquisition Premia

Model 1 Model 2 Model 3 Model 4

A B A B A B A B

Target CEO & Board Characteristics CEO near retirement (0, 1) −0.013 −0.010 −0.047* −0.047* −0.047* −0.046* −0.047* −0.046*

(0.675) (0.760) (0.076) (0.081) (0.080) (0.086) (0.081) (0.087)

Overconfident CEO (0, 1) −0.015 −0.016 −0.007 −0.007 −0.007 −0.007 −0.008 −0.007 (0.456) (0.451) (0.716) (0.731) (0.730) (0.744) (0.707) (0.717)

CEO-chairman (0, 1) −0.025 −0.026 −0.021 −0.021 −0.021 −0.020 −0.022 −0.022 (0.209) (0.194) (0.307) (0.310) (0.315) (0.318) (0.286) (0.284)

CEO-founder (0, 1) 0.033 0.036 0.019 0.019 0.019 0.020 0.020 0.021 (0.275) (0.242) (0.538) (0.526) (0.524) (0.511) (0.515) (0.498)

CEO tenure 0.002 0.002 0.001 0.001 0.002 0.002 0.002 0.002 (0.299) (0.285) (0.359) (0.351) (0.326) (0.319) (0.283) (0.278)

CEO equity ownership 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 (0.476) (0.432) (0.585) (0.552) (0.594) (0.559) (0.611) (0.577)

CEO post-deal employment 0.019 0.016 0.020 0.019 0.020 0.019 0.020 0.018 (0, 1) (0.314) (0.397) (0.289) (0.326) (0.287) (0.325) (0.303) (0.343)

G index (minus parachute) −0.003 −0.002 −0.002 −0.002 −0.002 −0.002 −0.002 −0.002 (0.528) (0.606) (0.562) (0.615) (0.556) (0.611) (0.575) (0.626)

Board ownership 0.001 0.001 0.001 0.001 0.001 0.001 0.001 0.001 (0.407) (0.392) (0.160) (0.164) (0.157) (0.161) (0.169) (0.170)

Pct of independent directors 0.017 0.005 0.026 0.023 0.027 0.024 0.024 0.021 (0.772) (0.924) (0.649) (0.689) (0.639) (0.680) (0.675) (0.723)

Deal Characteristics Private acquirer (0, 1) −0.048 −0.049* −0.062** −0.062** −0.061** −0.062** −0.061** −0.061**

(0.107) (0.100) (0.042) (0.039) (0.042) (0.040) (0.045) (0.043)

Cash payment (0, 1) 0.073*** 0.071*** 0.066*** 0.065*** 0.066*** 0.065*** 0.065*** 0.065*** (0.002) (0.002) (0.005) (0.005) (0.005) (0.005) (0.005) (0.006)

Tender offer (0, 1) 0.090*** 0.089*** 0.104*** 0.103*** 0.104*** 0.103*** 0.103*** 0.102*** (0.001) (0.001) (0.001) (0.001) (0.001) (0.001) (0.001) (0.001)

Hostile (0, 1) 0.064* 0.066* 0.052 0.053 0.053 0.054 0.053 0.054 (0.092) (0.085) (0.171) (0.163) (0.164) (0.157) (0.165) (0.158)

Same industry (0, 1) −0.004 −0.008 0.001 −0.001 0.001 −0.001 0.001 −0.001 (0.867) (0.720) (0.951) (0.978) (0.951) (0.977) (0.979) (0.950)

Rumor (0, 1) 0.087** 0.083** 0.085** 0.084** 0.085** 0.084** 0.084** 0.082** (0.024) (0.031) (0.027) (0.030) (0.026) (0.029) (0.030) (0.033)

Litigation (0, 1) −0.096 −0.092 −0.105 −0.103 −0.105 −0.103 −0.100 −0.098 (0.312) (0.333) (0.270) (0.277) (0.270) (0.277) (0.294) (0.304)

Prior bidding (0, 1) 0.075*** 0.074** 0.062** 0.062** 0.064** 0.064** 0.063** 0.063** (0.009) (0.010) (0.030) (0.029) (0.026) (0.026) (0.029) (0.029)

Toehold (0, 1) −0.002 0.004 0.003 0.008 0.004 0.008 0.003 0.008 (0.962) (0.927) (0.941) (0.861) (0.933) (0.851) (0.938) (0.861)

Target termination fee (0, 1) 0.045* 0.047* 0.045* 0.047* 0.045* 0.046* 0.044* 0.046* (0.072) (0.058) (0.070) (0.062) (0.073) (0.064) (0.076) (0.067)

Time to completion 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 (0.405) (0.388) (0.443) (0.430) (0.452) (0.438) (0.467) (0.454)

Target-initiated deal (0, 1) −0.054*** −0.055*** −0.053*** −0.053*** −0.053*** −0.053*** −0.053*** −0.054*** (0.004) (0.004) (0.006) (0.005) (0.006) (0.005) (0.005) (0.005)

Target input/ 0.174 0.174 0.110 0.101 0.102 0.092 0.120 0.112 Total acquirer output (0.505) (0.505) (0.668) (0.695) (0.692) (0.720) (0.638) (0.662)

Target purchases/ −0.298 −0.298 −0.223 −0.217 −0.217 −0.211 −0.230 −0.224 Total acquirer sales (0.195) (0.194) (0.321) (0.336) (0.335) (0.350) (0.307) (0.320)

1-year change in IP index −0.004 −0.004 −0.002 −0.002 −0.002 −0.001 −0.002 −0.001 (0.745) (0.756) (0.879) (0.904) (0.895) (0.920) (0.896) (0.918)

Parachute Heckman −0.014 −0.002 −0.003 −0.004 lambda (0.341) (0.884) (0.865) (0.805)

Target Heckman lambda −0.024 −0.016 −0.016 −0.015 (0.113) (0.316) (0.308) (0.344)

Adj. R 2 0.226 0.228 0.236 0.237 0.235 0.236 0.235 0.236 p-value of F-test 0.001 0.001 0.001 0.001 0.001 0.001 0.001 0.001

Fich, Tran, and Walkling 1733

contract includes a parachute provision; in model 3, the natural logarithm of the payments we identify as parachute compensation; and in model 4, the multiple used to calculate the value of the parachute. Although the first proxy is designed to measure the relative impact of the parachute on the target CEO’s wealth in particular, the remaining proxies also measure the importance of the parachute in general. For each proxy, we estimate the premium regression model twice: once controlling for self-selection and then controlling for sample selection. All other independent variables are defined in the legend accompanying Table 5.

The estimates in models 1A and 1B of Table 5 indicate that a 1-standard- deviation increase in parachute importance is associated with a decrease in premia of 2.6 percentage points. This drop in premia translates to an average decline of $127 million in terms of deal value.

The coefficients related to the other parachute proxies in models 2–4 are also negative and significant. The estimates in model 2 indicate that when the parachute has zero importance to the target CEO, takeover premia increase by 6.2 percentage points. The estimates in model 3 imply a drop in premia of 4.8 per- centage points for a $1 million dollar increase in the value of the parachute. According to model 4, targets experience a 1.8-percentage-point decline in pre- mia for a 1-unit increase in the parachute multiple. Consequently, the estimates related to the proxies in models 2–4 also document an inverse association be- tween parachute importance and takeover premia. However, the interpretation that arises from these proxies is not as economically informative as that arising from model 1. This occurs because it is possible that parachutes of the same value (or those calculated with the same multiple) deliver very different incentives. There- fore, by standardizing the value of the parachute by the pay target CEOs expect to give up, we are able to more accurately assess the incentives of parachutes during acquisitions.

The estimates for other independent variables in Table 5 are consistent with the existing M&A literature. We also find that acquisition premia increase with recent excess returns, liquidity, and deals structured as tender offers. Bid premia also increase with rumors, prior bidding, and the existence of a target termination fee. As in Hartzell et al. (2004) and Bargeron, Schlingemann, Stulz, and Zutter (2010), we do not find an association between the premia paid and whether the target CEO gets a job in the merged firm.

Notably, bargaining power affects the gains to target shareholders in mergers: Bid premia are around 5% lower in deals initiated by the target firm. This result agrees with those in Fich et al. (2011), who also document an inverse association between premia and target-initiated deals. In addition, bid premia decrease in the size of the target company, in deals by private acquirers, and in situations when the CEOs near retirement age. This last result is consistent with the findings in Jenter and Lewellen (2011). They show that premia are 8–10 percentage points lower when the target firm has a retirement-age CEO.

D. Rationality Considerations

Several papers in the M&A literature document (but do not explicitly recog- nize) a potential paradox. Specifically, studies find that target CEOs often accept

1734 Journal of Financial and Quantitative Analysis

lower premia when they obtain a benefit that is unlikely to totally make up for their takeover-related personal losses. For example, Hartzell et al. (2004) show that when target CEOs get extraordinary personal treatment (such as a parachute augmentation or a merger bonus), target shareholders receive a lower premium. Likewise, Wulf (2004) reports that target CEOs trade takeover premia for a pow- erful job in the merged firm. Fich et al. (2011) document a similar result when the benefit involves unscheduled stock option grants approved during nonpublic merger negotiations.

Our results on golden parachutes suggest a similar paradox for both the target CEOs and their shareholders. From the target CEOs’ perspective, a decrease in the takeover premium will create a private loss for these CEOs (due to a drop in the value of their equity-based portfolio), which their parachutes might not fully cover.11 This raises a question: Why would target CEOs with parachutes consent to a lower premium if doing so possibly makes them worse off? From the target shareholders’ perspective, a similar question emerges: Why would they include a golden parachute provision in their CEO’s pay contract if this benefit makes them worse off at the time of an acquisition bid?

1. Target CEOs

For the transactions we study, the unconditional probability that an acqui- sition is completed is approximately 87.8%, and the average takeover premium offered is about 35.9%. Ceteris paribus, in this base case, the appreciation of the target CEO’s equity-based portfolio is a function of an expected takeover pre- mium of approximately 31.51%. Nonetheless, our results indicate that a single standard-deviation increase in parachute importance raises the probability of deal completion by about 6.9 percentage points but lowers the premium by around 2.6 percentage points. Under these circumstances, we estimate the expected takeover premium to be approximately 31.52%. Consequently, it appears that the expected appreciation of the target CEO’s equity-based portfolio is about the same in the base case and in the presence of a more important parachute. A certainty equiv- alent argument can also explain the actions of target CEOs. Rather than bargain hard for higher premia and risk not completing the deal, target CEOs settle for smaller but certain premia that ensure deal success and, at the same time, partially make these CEOs whole for their expected personal losses.12 Put differently, by increasing the target CEO’s total merger payout relative to the expected value of his or her lost compensation, the parachute gets the CEO up to or past the certainty equivalent of that expected lost compensation.

Hence, the behavior we document for these target CEOs appears rational because their actions reveal that they are i) utility-maximizing agents (who get more satisfaction from current consumption than from deferred consumption),

11The equity-based portfolio contains the target CEOs’ (and their immediate family’s) share- ownership in the firm, stock options, and restricted stock.

12In fact, target CEOs are probably better off in the latter case. They get a relatively larger parachute, and (if the deal is completed) they can cash in their equity-based portfolio in full, since all stock restrictions and option vesting periods are lifted when the target ceases to exist as a stand- alone firm.

Fich, Tran, and Walkling 1735

and ii) risk-averse agents (who always prefer the same expected payoff with less risk).13

Under a similar logic, target shareholders might not be at a disadvantage when their CEO gets a more important parachute because they get essentially the same expected premium with or without an increase in the importance of the parachute. In fact, target shareholders are probably better off when their CEO gets a more important parachute, because under these circumstances they get a completed bid they would not otherwise have received.

We note that the finding that target CEOs settle for a lower premium (even CEOs with an equity-based portfolio that includes option grants and other stock pay) conforms to the theories in Ross (2004). He argues that attitudes toward risk depend not only on the convexity of an agent’s compensation schedule, but also on how the overall schedule maps into more (or less) risk-averse areas of the agent’s utility function. Ross states that the mapping of the overall compensation schedule can undo the impact of convex (or concave) pay schedules. Our findings suggest that raising the importance of the parachute mitigates the convexity that equity- based pay imposes in the target CEO’s utility function, which, in turn, makes the target CEO more risk averse.

2. Target Shareholders

Using the method in Comment and Schwert (1995), we estimate an uncon- ditional premium regression in a sample of 14,157 firm-years with data available from CRSP, Compustat, and RiskMetrics during 1999–2007. In this regression, which is reported as model 1 of Table 6, the key independent variable is a parachute (0,1) indicator. As in Comment and Schwert, we set the premium to 0 in non- takeover firm-years. The estimates in model 1 of Table 6 show that the presence of a parachute is a positive and statistically significant function of the uncondi- tional premium. This result indicates that a parachute is associated with an un- conditional net gain to shareholders.14 Consequently, this evidence suggests that providing their CEOs with a parachute appears to be a perfectly rational choice for target shareholders.15 Moreover, when the parachute provision is first put in place, it is unlikely that shareholders know i) whether (or when) their firm will become an acquisition target or ii) how relatively important (to the CEO) the parachute will turn out to be in case of a merger. This lack of foreknowledge also

13Our results indicate that by acquiescing to a lower premium, target CEOs potentially give up substantial value related to their stock and option holdings. This finding is consistent with those in Huddart and Lang (1996). They show that top managers tend to exercise options long before expiration (often around vesting dates and following price run-ups), sacrificing, on average, 50% of their Black- Scholes (1973) value.

14Our interpretation is based on the assertion that “the estimated effect of antitakeover measures on the unconditional premium is of interest because it is a net effect of a decrease in the premium if antitakeover devices deter offers and an increase if they increase premiums in successful takeovers” (Comment and Schwert (1995), p. 30).

15In addition, existing academic work provides additional rationales for the shareholders’ approval of a parachute for their CEO, particularly when the firm is not a takeover target. For example, it is argued that parachutes i) encourage managerial human capital investment in the firm (i.e., Knoeber (1986), Berkovitch and Khanna (1991)), ii) prompt top managers to eliminate redundant or ineffi- cient operations (i.e., Lambrecht and Myers (2007)), and iii) promote innovation (Francis, Hasan, and Sharma (2009), Manso (2011)).

1736 Journal of Financial and Quantitative Analysis

TABLE 6

Unconditional Premia and Predicted and Surprise Parachute Analyses

Table 6 presents ordinary least squares (OLS) regressions of the relation of acquisition premia with golden parachutes, antitakeover provisions, financial data, and characteristics of the takeover as in Comment and Schwert (1995). The de- pendent variable is the acquisition premium as reported by SDC. In model 1, the sample consists of pooled time-series cross-sectional data of 14,157 firm-years with data available from CRSP, Compustat, and RiskMetrics during 1999–2007. In this unconditional premium regression, the takeover premium is set to 0 in nontakeover firm-years. In models 2 and 3, the sample consists of 851 deals described in Table 1. All financial characteristics are averaged over 3 fiscal years. Predicted parachute is the fitted parachute and surprise parachute is the error term from model 3 of Table A1. These two variables enter models 2 and 3 in this table separately from the Parachute (0, 1). Other variables are self-explanatory or defined elsewhere. The p-values are White (1980) heteroskedasticity-consistent and adjusted for clustering by firms. *, **, and *** indicate statistical significance at the 10%, 5%, and 1% levels, respectively.

Model 1 Model 2 Model 3

Coefficient p-Value Coefficient p-Value Coefficient p-Value

Intercept 0.025 0.119 0.966*** 0.001 0.848*** 0.004 Parachute (0, 1) 0.006*** 0.001 −0.076** 0.012 −0.069** 0.012

Predicted parachute −0.202 0.412 −0.163 0.405 Surprise parachute −0.069** 0.031 −0.066** 0.030

Poison pill (0, 1) 0.003** 0.039 0.016 0.400 0.010 0.590 Classified board (0, 1) −0.005 0.895 −0.004 0.860 −0.017 0.374 Supermajority to approve merger (0, 1) −0.004 0.135 −0.026 0.467 −0.019 0.533 Delaware incorporation (0, 1) 0.002 0.333 −0.008 0.708 −0.010 0.607 log(Assets) −0.003*** 0.001 −0.015* 0.072 −0.013 0.101 Q −0.002** 0.018 −0.018* 0.061 −0.016 0.104 Leverage 0.007 0.194 0.094* 0.062 0.087* 0.074 Liquidity 0.006 0.331 0.097 0.242 0.105* 0.089 Free cash flow −0.006 0.203 −0.120 0.387 −0.109 0.392 Prior year excess return 0.043 0.305 0.095*** 0.001 0.089*** 0.001 CEO near retirement (0, 1) −0.041* 0.098 Private acquirer (0, 1) −0.051* 0.089 Cash payment (0, 1) 0.063*** 0.006 Tender offer (0, 1) 0.107*** 0.001 Hostile (0, 1) 0.047 0.217 Same industry (0, 1) 0.000 0.995 Prior bidding (0, 1) 0.060** 0.031 Target termination fee (0, 1) 0.043* 0.084 Target-initiated deal (0, 1) −0.048** 0.011 Target input/Total acquirer output 0.169 0.509 Target purchases/Total acquirer sales −0.258 0.249 Year and industry fixed effects Yes Yes Yes N 14,157 851 851 Adj. R 2 0.028 0.179 0.235 p-value of F-test 0.001 0.001 0.001

rationalizes the shareholders’ choice with regard to including a parachute provi- sion in their CEO’s compensation contract.

E. Anticipation Bias

It is no surprise to the market that many firms offer parachutes to their CEOs. Moreover, Jensen and Zimmerman (1985) argue that stock prices might reflect the anticipation of a takeover premium if a golden parachute reveals that a takeover is more likely. To recognize this, we follow the methodology of Com- ment and Schwert (1995) and replace the (0,1) indicator for the presence of a parachute with variables related to the anticipated and surprise components of the parachute. These components are estimates from the parachute prediction re- gression reported as model 2 of Table A1. We estimate this prediction regres- sion in a sample of 14,157 firm-years with data available from CRSP, Compustat, and RiskMetrics during 1999–2007. The predictable component is an estimate of the probability that the target CEO’s compensation contract includes a parachute

Fich, Tran, and Walkling 1737

provision. The surprise component is computed as the parachute indicator minus the estimated probability that the target CEO has a parachute.

In models 2 and 3 of Table 6, we present two regressions of the conditional takeover premium in which the parachute components are the independent vari- ables of interest. For reference and in the spirit of Comment and Schwert (1995), in both tests we include the estimate for the parachute dummy variable from separate similarly structured premium regressions that do not include the golden parachute components.

The coefficient on the surprise parachute variable in models 2 and 3 of Table 6 is negative and significant, indicating that the unanticipated effect of a golden parachute is associated with lower bid premia. In contrast, the predictable parachute component does not attain statistical significance. Therefore, the most we can conclude is that the known existence of golden parachutes is already im- pounded in a target’s value. Nonetheless, this conclusion is important because it validates the view that it is not the mere presence of a parachute, but its rela- tive importance to the target CEO, that matters. Consequently, it is plausible that the unanticipated negative effect captured by the surprise parachute variable in the conditional premium tests reflects the amount by which parachutes wind up insulating target CEOs from personal losses.

F. Parachute Importance and Acquirer Returns

The foregoing results indicate that as the level of parachute importance increases, the premia paid to target firms decrease and the probability of deal completion increases. Because of this trade-off, our tests show that target share- holders get the same expected premium despite an increase in the importance of the parachute to their CEO. Since the premium paid for targets and the probability of deal success may also affect the bidder, we now evaluate whether the impor- tance of parachutes to target CEOs affects the acquirer shareholders’ wealth.

To test whether (and how) the importance of parachutes to target CEOs affects the returns to acquirers, in Table 7 we run two ordinary least squares (OLS) regressions of the 3-day merger announcement cumulative abnormal re- turn (CAR) meeting the 459 publicly traded bidders in our sample. We follow the M&A literature in order to properly specify our acquirer return regressions. For instance, both models in Table 7 control for deal, market, and bidder characteris- tics similar to those in Moeller et al. (2005) and Masulis, Wang, and Xie (2007). Model 2 of Table 7 includes target characteristics similar to those in Moeller (2005) and Wang and Xie (2009) as additional controls.16 The independent vari- able of interest in Table 7 is our proxy of parachute importance: the value of the golden parachute scaled by the expected lost compensation to the target CEO.

The estimates in Table 7 indicate that acquirer returns increase as the level of parachute importance to the target CEO increases. A 1-standard-deviation

16The control variables in Table 7 yield results similar to those in other papers. As in Moeller et al. (2005), bidder size is negatively related to the acquirer return. Like Masulis et al. (2007), the competitive industry indicator yields positive coefficients, while the bidder G index and the relative size variable yield negative coefficients.

1738 Journal of Financial and Quantitative Analysis

TABLE 7

Parachute Importance and Acquirer Returns

From the original 851 acquisitions announced during 1999–2007 described in Table 1, we examine 459 offers made by U.S. public bidders in which data for these firms are available from CRSP, Compustat, and RiskMetrics. We run ordinary least squares (OLS) regressions of acquirer returns similar to those in Moeller et al. (2005) and Masulis et al. (2007). The dependent variable is the acquirer’s cumulative abnormal return over the 3 days around the merger announcement date, calculated as the residual from the market model estimated during the (−272,−21) interval. The main independent variable is the parachute importance relative to the expected lost compensation to the target CEO. The competitive industry (0, 1) equals 1 if the bidder’s industry is in the bottom quartile of all industries sorted annually by the Herfindahl-Hirschman index. An industry’s Herfindahl-Hirschman index is computed as the sum of squared market shares of all firms in the industry using data on sales (as in Masulis et al.). The unique industry (0, 1) equals 1 if the bidder’s industry is in the top quartile of all industries sorted annually by industry-median product uniqueness. Product uniqueness is defined as selling expenses scaled by sales (as in Masulis et al.). As in Schlingemann, Stulz, and Walkling (2002), liquidity index is the liquidity of the market for corporate control for the target firm’s industry. This variable is defined as the value of all corporate control transactions for US$1 million or more reported by SDC for each year and industry divided by the total book value of assets of all Compustat firms in the same industry and year. Other variables are self-explanatory or defined elsewhere. The reported p-values are White (1980) heteroskedasticity-consistent. *, **, and *** indicate statistical significance at the 10%, 5%, and 1% levels, respectively.

Model 1 Model 2

Coefficient p-Value Coefficient p-Value

Intercept 0.142 0.137 0.209** 0.046 Parachute importance 0.015* 0.056 0.017** 0.031

Deal Characteristics Relative size −0.038*** 0.001 −0.031*** 0.001 Stock payment (0, 1) −0.026*** 0.005 −0.028*** 0.003 Tender offer (0, 1) 0.013 0.266 0.011 0.342 Friendly deal (0, 1) 0.032** 0.023 0.033** 0.021 Prior bidding (0, 1) 0.003 0.776 0.002 0.868 Toehold (0, 1) 0.040 0.157 0.035 0.229 Same industry (0, 1) 0.002 0.819 0.005 0.571 Target-initiated deal (0, 1) 0.008 0.314 0.005 0.472 Target input/Total acquirer output −0.237 0.241 −0.201 0.330 Target purchases/Total acquirer sales 0.003 0.985 0.052 0.720

Market Characteristics Competitive industry (0, 1) 0.015* 0.072 0.014* 0.083 Unique industry (0, 1) −0.034*** 0.001 −0.038*** 0.001 Liquidity index −0.012 0.583 −0.032 0.202 1-year macroeconomic change −0.009** 0.029 −0.008* 0.055 Bidder Characteristics log(Assets) −0.001** 0.013 −0.001** 0.012 Q 0.002** 0.025 0.002** 0.031 Leverage −0.019 0.467 −0.014 0.581 Free cash flow 0.019 0.585 0.025 0.475 Prior year stock returns 0.008 0.224 0.006 0.357 G index −0.003** 0.049 −0.003* 0.062 Board size −0.004 0.767 −0.001 0.992 Delaware firm (0, 1) −0.004 0.590 −0.001 0.420 Target Characteristics Q 0.000 0.963 Leverage −0.023 0.520 Free cash flow 0.087 0.501 Prior year stock returns −0.001 0.978 G index −0.003* 0.075 Board size −0.031* 0.066 Delaware firm (0, 1) −0.001 0.928 CEO near retirement (0, 1) −0.011 0.370 CEO-chairman (0, 1) −0.005 0.537 CEO ownership 0.000* 0.069 Independent board (0, 1) 0.006 0.631 Board ownership 0.000 0.312

Year and industry fixed effects Yes Yes N 459 459 Adj. R 2 0.215 0.216 p-value of F-test 0.001 0.001

Fich, Tran, and Walkling 1739

increase in parachute importance is associated with an increase in bidder returns of 0.7%. This increase translates into a gain of $205 million in terms of market capitalization for the average bidder in our sample. This finding conflicts with the notion that target shareholders are not worse off when their CEO gets a more important parachute because they get essentially the same expected premium. In- stead, our bidder return tests document a transfer of rents from target shareholders to acquirer shareholders when target CEOs have more important parachutes. Our previous results indicate that it is rational for target shareholders to implement a golden parachute. But, as we have also noted, it is impossible for future targets to precisely anticipate the timing of the bid or the relative importance of golden parachutes to the CEO. An unintended consequence is that parachutes that turn out to be overly generous can cause target CEOs to become more risk averse and (perhaps needlessly) to surrender rents to the bidding firm. At the same time, the fact that acquirer returns are higher is another manifestation of the lower bid premium (the other is the higher probability of completing the deal). As a re- sult, the evidence that acquirer shareholders are better off when more important parachutes are provided to the target CEOs is not inconsistent with the finding that target shareholders also benefit because they get a completed bid they would not have otherwise received.

IV. Additional Tests

In this section, we describe alternative tests we conduct in order to assess the robustness of the preceding results.

A. Reverse Causality

A key test of the incentive alignment vs. managerial interest hypotheses is the relation of the parachute to the premium paid in the acquisition. The analyses in Table 5 document an inverse association between parachute importance and takeover premia. However, companies expecting a low premium if they become takeover targets might provide a more generous (and important) parachute to their CEOs. Under this possibility, the direction of causality would be reversed.

To address whether the endogenous choice between parachute importance and deal premia affects the results presented in Table 5, we estimate four different systems of simultaneous equations following the methodology in Maddala (1983). Each system uses a different golden parachute proxy. In all systems, bid premia and the individual parachute proxy are provided as the two endogenous variables. For every system, the parachute variable and bid premia instruments are sepa- rately estimated from first-stage regressions. The second-stage tests consist of an OLS regression of bid premia on the parachute instrument and a regression of the parachute proxy on the instrument for the bid premia, respectively. The standard errors in these tests are adjusted for the fact that the instrumental variables for the parachute and bid premia are estimated.

To identify the simultaneous system, we must exclude one exogenous variable from each of the two second-stage regression equations. For the parachute equation, we must satisfy the relevancy condition with a variable that is

1740 Journal of Financial and Quantitative Analysis

correlated with the parachute after controlling for all other exogenous variables. The same variable will satisfy the exclusion restriction if it is uncorrelated with the error term of the second-stage premium regression. For this variable we use the CEO founder (0,1) dummy. Table 3 indicates that this variable is significantly re- lated to our parachute proxy. Prior research by Moeller (2005) and the estimates in Table 5 show that the founder (0,1) dummy variable is unrelated to premia. For the premium equation, we use the target’s excess stock return during the year prior to the acquisition. This variable appears to satisfy the relevancy condition and the exclusion restriction. A recent study by Aktas et al. (2010), as well as the results in Table 5, shows that a target’s prior excess return is related to the bid premium. The evidence in both Table 3 and Table A1 indicates that the excess return variable is not related to the parachute.

Table 8 presents our simultaneous equations analyses. In Panel A we use the importance of the parachute relative to the lost compensation

TABLE 8

Simultaneous Equations Analyses

Table 8 reports simultaneous equations regressions in which we treat golden parachute and acquisition premium as en- dogenous variables. We analyze 851 acquisitions announced during 1999–2007 described in Table 1. We report simulta- neous equations results using the relative importance of the golden parachute to the lost compensation (GP/LC) in Panel A and those using the parachute dummy variable, the parachute value, and the parachute multiple in Panel B. The instruments in the second-stage regressions equal the fitted value from the first-stage regression. We use probit regressions when the dependent variable is the parachute dummy and ordinary least squares (OLS) regressions otherwise. Other variables are self-explanatory or defined elsewhere. We report White (1980) heteroskedasticity-consistent p-values in parentheses. *, **, and *** indicate statistical significance at the 10%, 5%, and 1% levels, respectively.

Panel A. Simultaneous Equations Using the Relative Importance of GP/LC

Model 1A Model 1B

1st Stage 2nd Stage 1st Stage 2nd Stage

Dependent Variable

GP/LC Premium Premium GP/LC

Intercept 0.224*** 0.464*** 0.370*** 0.294*** (0.007) (0.001) (0.001) (0.002)

GP/LC (instrument) −0.419** (0.037)

Premium (instrument) −0.019 (0.538)

log(Assets) −0.023*** −0.026*** −0.017*** −0.026*** (0.010) (0.002) (0.005) (0.005)

Q 0.000 0.010 0.010 0.002 (0.980) (0.346) (0.277) (0.910)

Leverage 0.050 0.154** 0.133** 0.075 (0.577) (0.027) (0.026) (0.406)

Free cash flow 0.064 −0.063 −0.089 0.047 (0.688) (0.606) (0.400) (0.767)

Liquidity −0.040 0.056 0.073 −0.027 (0.570) (0.313) (0.128) (0.711)

Prior year excess return −0.143 0.702*** 0.762*** (0.538) (0.001) (0.001)

CEO near retirement (0, 1) 0.674*** −0.242 −0.041* 0.667*** (0.001) (0.142) (0.095) (0.001)

CEO-chairman (0, 1) −0.009 −0.001 0.002 −0.009 (0.737) (0.949) (0.891) (0.750)

CEO-founder (0, 1) −0.132*** 0.055 −0.122*** (0.001) (0.344) (0.003)

(continued on next page)

Fich, Tran, and Walkling 1741

TABLE 8 (continued)

Simultaneous Equations Analyses

Panel A. Simultaneous Equations Using the Relative Importance of GP/LC (continued)

Model 1A Model 1B

1st Stage 2nd Stage 1st Stage 2nd Stage

Dependent Variable

GP/LC Premium Premium GP/LC

CEO tenure 0.010*** 0.003 −0.001 0.010*** (0.001) (0.181) (0.486) (0.001)

Overconfident CEO (0, 1) −0.050 −0.041* −0.020 −0.054 (0.176) (0.093) (0.295) (0.156)

CEO equity ownership 0.001 0.000 0.000 0.001 (0.481) (0.857) (0.813) (0.499)

Pct. of independent directors 0.104 −0.045 −0.088* 0.088 (0.151) (0.480) (0.070) (0.229)

Adj. R 2 0.131 0.139 0.288 0.129 Regression’s p-value 0.001 0.001 0.001 0.001

Panel B. Simultaneous Equations Using Alternative Parachute Proxies

GP Proxy = GP(0, 1) GP Proxy = log(GP value) GP Proxy = GP Multiple

Model 2A Model 2B Model 3A Model 3B Model 4A Model 4B

2nd Stage 2nd Stage 2nd Stage 2nd Stage 2nd Stage 2nd Stage

Dependent Variable

Premium GP Proxy Premium GP Proxy Premium GP Proxy

Intercept 0.389*** 0.380 0.538*** 3.381*** 0.460*** 0.721*** (0.001) (0.372) (0.001) (0.001) (0.001) (0.010)

GP proxy (instrument) −0.116** −0.054* −0.136* (0.044) (0.075) (0.071)

Premium (instrument) −0.584 −0.757 −0.165 (0.143) (0.258) (0.532)

log(Assets) −0.013** 0.025 −0.001* 0.281*** 0.001* 0.124*** (0.048) (0.560) (0.063) (0.001) (0.062) (0.001)

Q 0.004 −0.050 0.004 −0.103 0.001 −0.067* (0.704) (0.432) (0.701) (0.321) (0.946) (0.099)

Leverage 0.121** −0.029 0.141** 0.246 0.197*** 0.487* (0.044) (0.944) (0.038) (0.713) (0.010) (0.065)

Free cash flow −0.209* −1.090 −0.110 −0.445 −0.163 −0.557 (0.087) (0.180) (0.362) (0.704) (0.193) (0.229)

Liquidity 0.055 −0.114 0.024 −0.841 0.023 −0.352* (0.262) (0.724) (0.691) (0.115) (0.702) (0.095)

Prior year excess return 0.710*** 0.730*** 0.744*** (0.001) (0.001) (0.001)

CEO near retirement (0, 1) −0.079*** −0.351 −0.063** −0.438 −0.058** −0.136 (0.008) (0.227) (0.032) (0.307) (0.038) (0.207)

CEO-chairman (0, 1) 0.040 0.328*** 0.037 0.640*** 0.026 0.175** (0.115) (0.009) (0.179) (0.001) (0.266) (0.027)

CEO-founder (0, 1) −0.446*** −0.981*** −0.397*** (0.007) (0.001) (0.001)

CEO tenure −0.002 −0.010 −0.001 −0.011 0.000 0.004 (0.205) (0.215) (0.375) (0.488) (0.777) (0.524)

Overconfident CEO (0, 1) −0.021 −0.026 −0.016 0.048 −0.020 −0.007 (0.255) (0.843) (0.445) (0.819) (0.336) (0.936)

CEO equity ownership −0.001 −0.003 0.000 −0.006 0.000 −0.002 (0.465) (0.393) (0.576) (0.466) (0.600) (0.528)

Pct of independent directors 0.090 1.486*** 0.063 2.732*** 0.053 1.019*** (0.396) (0.001) (0.552) (0.001) (0.597) (0.001)

Adj. R 2 0.188 0.105 0.136 0.130 0.129 0.127 Regression’s p-value 0.001 0.001 0.001 0.001 0.001 0.001

1742 Journal of Financial and Quantitative Analysis

as our proxy for the parachute. After accounting for endogeneity, the parachute instrument in the second-stage premium regression is negative and statistically significant. In contrast, the premium instrument in the second-stage parachute regression is not different from 0. This last result indicates that bid premia are unrelated to parachute importance and provides no evidence of causation run- ning in the reverse direction. The interpretation that arises from our second-stage premium regression is similar to those in Table 5. According to our estimates, a 10-percentage-point increase in parachute importance with respect to the expected lost compensation is associated with a reduction in premia of about 3 percentage points.

In Panel B of Table 8 we use the other three parachute proxies described earlier. For each of these proxies, we also estimate a simultaneous system con- sisting of two first-stage and two second-stage regressions. To conserve space, we only report the two second-stage regressions for each system. The tests in Panel B also document an inverse and significant association between the parachute instru- ment and bid premia. However, the same tests reveal no association between the premium instrument and the golden parachute dependent variables. Overall, the results of our simultaneous equations analyses lend no support to the conjecture that causality runs in the reverse direction.17

B. Alternative Scaling of the Parachute Payment

In the preceding tests, we assess the relative importance of parachutes by scaling their value by the expected lost compensation to target CEOs if the deal is completed. There are at least two issues of potential concern with this measure of parachute importance. First, as noted in Section II.C, we use current pay as the main input to estimate the lost compensation to the target CEO upon an ac- quisition. While our tests control for immediate past performance, the literature shows that current pay is typically a function of multiple years of performance.18

Consequently, if pay reacts with a lag to cumulated poor performance, the cur- rent pay used in the lost compensation calculation could be high. This issue could be relevant because we use excess returns for the identification in the simultane- ous systems tests. While excess returns may not be related to the existence of the parachute, they may be related to its importance through the compensation effect described above. Second, we note that to fully understand the incentives of the target CEOs, we would need to consider how the change in value in their existing firm-specific equity-based portfolios (plus accelerated vesting, etc.) affects their expected wealth loss due to a merger. However, neither our measure of parachute importance nor the other proxies we use in our earlier tests consider the magnitude of the target CEOs’ equity-based payoff.

To address the issues just noted, we scale the parachute payment by a to- tally different measure: the value of the total merger pay package to the target

17We note that the simultaneous system in which we use the parachute indicator (0,1) only reveals that there is no reverse causality. The economic effect of the parachute is not readily interpretable from the second-stage premium regression that uses this indicator as the parachute proxy (Maddala (1983)).

18See, for example, Boschen and Smith (1995).

Fich, Tran, and Walkling 1743

CEO.19 This proxy, which considers the target CEO’s equity-based payoff, tracks the importance of the parachute relative to the entire acquisition-related compen- sation the executive gets.

In Table 9, we replicate earlier tests using the alternative proxy of parachute importance. Specifically, Panel A of Table 9 presents premium regressions similar to those reported in Table 5. These tests document an inverse and statistically significant association between parachute importance (relative to the merger pay

TABLE 9

Analyses of GPs Relative to the MPP

Table 9 reports robustness tests of the importance of golden parachutes relative to the merger pay package on acqui- sition premium. Panel A presents the premium regressions and Panel B presents the simultaneous equations results. The key independent variable is the importance of the golden parachute relative to the merger pay package (GP/MPP). Other variables are self-explanatory or defined elsewhere. We report White (1980) heteroskedasticity-consistent p-values in parentheses. *, **, and *** indicate statistical significance at the 10%, 5%, and 1% levels, respectively.

Panel A. Premium Regressions Using the Relative Importance of Parachute to the MPP

Model 1 Model 2

Coefficient p-Value Coefficient p-Value

Intercept 0.642*** 0.001 0.656*** 0.001 GP/MPP −0.444*** 0.001 −0.442*** 0.001 Target Characteristics log(Assets) −0.036*** 0.001 −0.033*** 0.001 Q −0.011 0.274 −0.012 0.233 Leverage 0.055 0.247 0.053 0.258 Free cash flow −0.036 0.769 −0.034 0.782 Liquidity 0.080 0.176 0.085 0.150 Prior year excess return 0.117*** 0.001 0.116*** 0.001

Target CEO & Board Characteristics CEO near retirement (0, 1) −0.048* 0.058 −0.048* 0.060 Overconfident CEO (0, 1) −0.016 0.419 −0.015 0.442 CEO-chairman (0, 1) −0.022 0.248 −0.022 0.264 CEO-founder (0, 1) −0.000 0.988 −0.001 0.985 CEO tenure −0.001 0.535 −0.001 0.554 CEO equity ownership 0.001 0.228 0.001 0.226 CEO post-deal employment (0, 1) 0.017 0.364 0.015 0.396 G index (minus parachute) 0.000 0.957 0.000 0.901 Board ownership 0.001 0.221 0.001 0.237 Pct. of independent directors 0.052 0.345 0.051 0.351

Deal Characteristics Private acquirer (0, 1) −0.062** 0.033 −0.063** 0.030 Cash payment (0, 1) 0.044** 0.050 0.043* 0.052 Tender offer (0, 1) 0.098*** 0.001 0.097*** 0.001 Hostile (0, 1) 0.035 0.331 0.036 0.319 Same industry (0, 1) 0.010 0.639 0.008 0.698 Rumor (0, 1) 0.087** 0.018 0.085** 0.020 Litigation (0, 1) −0.101 0.266 −0.101 0.266 Prior bidding (0, 1) 0.073*** 0.007 0.073*** 0.007 Toehold (0, 1) −0.024 0.560 −0.020 0.639 Target termination fee (0, 1) 0.036* 0.068 0.037* 0.062 Time to completion 0.000 0.201 0.000 0.193 Target-initiated deal (0, 1) −0.039** 0.034 −0.039** 0.031 Target input/Total acquirer output 0.134 0.584 0.122 0.617 Target purchases/Total acquirer sales −0.174 0.417 −0.166 0.388 1-year change in IP index −0.001 0.916 −0.001 0.947 Parachute Heckman lambda −0.003 0.819 Target Heckman lambda −0.017 0.263 Adj. R 2 0.230 0.231 p-value of F-test 0.001 0.001

(continued on next page)

19As in Hartzell et al. (2004), this package includes common equity and stock option appreciation, the parachute, and, in some cases, a special merger bonus. The average merger pay package drawn by target CEOs in our sample is worth almost $36 million.

1744 Journal of Financial and Quantitative Analysis

TABLE 9 (continued)

Analyses of GPs Relative to the MPP

Panel B. Simultaneous Equations Using the Relative Importance of Parachute to the MPP

Model 1 Model 2

1st Stage 2nd Stage 1st Stage 2nd Stage

Dependent Variable

GP/MPP Premium Premium GP/MPP

Intercept 0.645*** 0.715*** 0.370*** 0.845*** (0.001) (0.001) (0.001) (0.001)

GP/MPP (instrument) −0.534** (0.025)

Premium (instrument) −0.054 (0.219)

log(Assets) −0.049*** −0.043*** −0.017*** −0.058*** (0.001) (0.001) (0.005) (0.001)

Q −0.017* 0.001 0.010 −0.011 (0.080) (0.902) (0.277) (0.186)

Leverage 0.048 0.159*** 0.133** 0.120** (0.437) (0.004) (0.026) (0.032)

Free cash flow 0.075 −0.050 −0.089 0.026 (0.495) (0.608) (0.400) (0.790)

Liquidity −0.129*** 0.004 0.073 −0.090** (0.009) (0.947) (0.128) (0.044)

Prior year excess return −0.213 0.541*** 0.762*** (0.401) (0.001) (0.001)

CEO near retirement (0, 1) 0.016 −0.032 −0.041* −0.007 (0.536) (0.150) (0.095) (0.775)

CEO-chairman (0, 1) 0.032* 0.020 0.002 0.033** (0.086) (0.267) (0.891) (0.046)

CEO-founder (0, 1) −0.104*** 0.055 −0.074*** (0.001) (0.344) (0.004)

CEO tenure −0.004*** −0.003* −0.001 −0.005*** (0.003) (0.080) (0.486) (0.001)

Overconfident CEO (0, 1) −0.020 −0.030* −0.020 −0.031* (0.301) (0.081) (0.295) (0.077)

CEO equity ownership −0.003*** −0.002 0.000 −0.003*** (0.001) (0.107) (0.813) (0.001)

Pct. of independent directors 0.194*** 0.015 0.088 0.146*** (0.001) (0.820) (0.170) (0.001)

Adj. R 2 0.135 0.143 0.298 0.137 Regression’s p-value 0.001 0.001 0.001 0.001

package) and premia. Based on the coefficient estimates, a 5-percentage-point increase in parachute importance is associated with a decline in premia of 2.2 per- centage points. This decline is economically meaningful: For the average sample target, the lower premia imply a shortfall of about $108 million in terms of deal value.20

20We are sensitive that the results that obtain when we scale the parachute by the merger pay package are driven by the fact that the offer price is used to value the equity components of the merger pay package. To address this issue and purge the offer price from the merger pay package, we record each target’s stock price 6 weeks prior to the start of merger negotiations. We use this price and the Black-Scholes (1973) methodology to value all the stock options held by the target CEO. Similarly, we use this price to value all stock and restricted stock owned by the target CEO. With these new values, we re-estimate the dollar amount of the merger pay package 6 weeks prior to the start of merger negotiations. Finally, we standardize the parachute by this alternative estimate of the

Fich, Tran, and Walkling 1745

Panel B of Table 9 presents simultaneous regression analyses similar to those in Panel B of Table 8. Consistent with our earlier tests, whereas the premium in- strument is not statistically significant in the parachute regression, the parachute instrument is negative and significant in the premium regression. In addition, in untabulated tests we find that a 1-standard-deviation increase in the alternative measure of parachute importance increases i) the probability of merger comple- tion by 8.5 percentage points and ii) the return to the acquirer firms by 0.34%. Overall, the analyses on the importance of the parachute relative to the merger pay package generate inferences similar to those obtained using our other proxy of parachute importance.

C. Takeover Premia Alternatives

The estimates presented in Tables 5, 6, 8, and 9 are based on the 4-week pre- mium reported by SDC. We re-estimate all target premia using i) the combined method in Officer (2003); ii) the CAR running from 20 days before the deal an- nouncement (AD−20) until the day after (AD + 1), following Jarrell and Poulsen (1989); and iii) the cumulated returns over the (−42,+126) window as in Schwert (1996). Our results continue to hold when we use these alternative premium mea- sures. For example, using the combined premium (Officer (2003)), we estimate that a 1-standard-deviation increase in parachute importance (relative to the tar- get CEO’s expected lost compensation) is related to a target premium that is 3.1 percentage points lower. This shortfall triggers a decline of about $152 million in deal value for the average sample target. This result is similar to those tabulated.21

D. Clustering

The industrial distribution of our sample, which we report in Panel B of Table 2, exhibits some clustering in Business Equipment. We re-estimate all of the premium regressions in Tables 5 and 9, clustering the standard errors by the target’s 2-digit Standard Industrial Classification (SIC) code and also by their Fama and French (1997) 48 industry groups. The target premium results related to all of our parachute proxies are robust to clustering the standard errors by the targets’ industrial classification.

E. Selection Issues

The deal completion tests in Table 4 and the target premia regressions in Tables 5 and 9 use the methodology in Heckman (1979) to control for

merger pay package. This ratio becomes the key independent variable in regressions similar to those in Table 9. The results from these tests produce inferences similar to those in the main text: Higher parachute importance is related to lower premia.

21Following Officer (2003), 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 announce- ment. 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).

1746 Journal of Financial and Quantitative Analysis

self-selection (because firms can choose to offer a parachute) and for sample selection (since companies do not randomly become takeover targets). The self- selection and sample selection inverse Mills ratios we use in our tests are respec- tively based on the first-stage regressions we report as model 3 of Table A1 and model 4 in Table A2. In employing Heckman’s model, it is important to correctly specify the first-stage regressions. Hence, our first-stage tests are constructed following Agrawal and Knoeber (1998), Palepu (1986), Comment and Schwert (1995), and Bates, Becher, and Lemmon (2008). The results in these first-stage regressions conform to the evidence in prior work. For example, in Table A1, the coefficient for target size is consistent with the findings in Agrawal and Knoeber (1998). In Table A2, the estimates for target size, poison pill, and classified board are in line with those in Palepu (1986), Comment and Schwert (1995), and Bates et al. (2008), respectively.

When we employ the Heckman methodology, we find that none of the se- lection controls in Tables 4, 5, and 9 attain statistical significance at conventional levels. Therefore, we interpret these findings to indicate that our analyses are im- mune to selection issues. Nonetheless, this interpretation assumes that our first- stage Heckman (1979) regressions are properly specified. The correspondence of our first-stage estimates to the extant literature suggests that this assumption is reasonable.22

F. Bargaining Power

We recognize that with a fixed amount of synergies, a lower premium will directly increase the bidder return. This possibility raises the issue of whether the bargaining power of the target is low or whether the target CEO gives away rents due to more important parachutes. Moreover, if the selection model fails to capture the relations between firm-specific bargaining power, the parachute terms, and being selected as a target, this issue is not mitigated with the two- stage approach we perform. As a result, it is imperative to account for the role of bargaining power, because it directly impacts the inferences we can draw from our tests.

We implement a number of empirical controls to alleviate the bargaining power-related concerns just described. First, the selection equation of the prob- ability of becoming a takeover target (Table A2) controls for the Herfindahl- Hirschman index to capture a firm’s power in it its own industry. This choice is motivated by the argument in Rhodes-Kropf and Kadyrzhanova (2011) that char- acteristics (such as the level of industry concentration) that allow managers to de- lay takeovers have a significant bargaining effect. Second, when appropriate, our

22We recalculate the inverse Mills ratio from a regression that augments the parachute determinants model in Agrawal and Knoeber (1998). This alternative first-stage regression is reported as model 1 of Table A1. We then use the alternate self-selection inverse Mills ratio in the tests in Tables 4 and 5. The findings are similar to those reported, and the self-selection control remains statistically insignificant. We also produce two other sample selection inverse Mills ratios using regressions 1 and 2 in Table A2 as the first-stage models. These regressions expand the specifications in Palepu (1986) and in Com- ment and Schwert (1995), respectively. We again use the alternative inverse Mills ratios in Tables 4 and 5: All results hold, and the sample selection controls remain insignificant.

Fich, Tran, and Walkling 1747

tests control for “target-initiated deals,” a variable shown to affect the distribution of gains during mergers (Aktas et al. (2010)) and, therefore, a reasonable proxy for a target’s bargaining power. We also control for input-output/sales-purchases variables between the target and the bidder industries. Ahern (2012) argues that these customer-supplier relations proxy for the market power of the merger par- ticipants and help account for the role of product markets on bargaining outcomes in mergers. Our results prove robust to controls for bargaining power.

G. Tax Regulations and the Sarbanes-Oxley Act

On Feb. 19, 2002, the Internal Revenue Service proposed new regulations to Section 280G of the Internal Revenue Code.23 The new regulations provide amendments and clarifications to those issued on May 5, 1989, and apply to parachute payments occurring on or after Jan. 1, 2004. The amendments clar- ify that the safe harbor related to change-in-control payments is three times the average taxable compensation over the five most recent calendar years prior to the change in control. It also states that a company that exceeds the safe harbor will lose tax deductions for the excess amounts and that the executive will be liable for a 20% excise tax on the excess payment.

A 2008 study by RiskMetrics finds that the new tax regulations have done lit- tle to reduce parachute payments.24 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 their top executives. The excise tax gross-ups essentially free 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 $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 is possible that the new Section 280G rules may have affected the size of parachutes and, in turn, the relative importance of these benefits.

To investigate the potential effect of the new tax rules on parachutes, we revisit the regressions reported in Table 3 related to the relative importance of parachutes. In an untabulated test similar to that in model 1 of Table 3, we in- clude a dummy variable for deals initiated after Feb. 19, 2002. This variable also controls for the effect of other potentially important events that occur during 2002. For example, due to the new rules contained in the Sarbanes-Oxley Act of 2002 (SOX), many firms curbed the equity-based pay given to top managers while increasing their base salary (Chhaochharia and Grinstein (2009)). This pay redis- tribution could partially account for an increase in parachute importance. Nonethe- less, the estimate for the 2002 indicator is not statistically significant.

We also examine whether the shareholder wealth effects related to the impor- tance of parachutes change following the passing of SOX. In unreported analyses,

23See REG-209114-90 at http://www.irs.gov/pub/irs-regs/20911490.pdf 24See “Gilding Golden Parachutes: The Impact of Excise Tax Gross-Ups” by Kosmas Papadopou-

los at http://www.riskmetrics.com/docs/2008ExciseTax

1748 Journal of Financial and Quantitative Analysis

we estimate premium regressions similar to those in model 1 of Table 5. In these tests we add a control variable that interacts our parachute importance proxy with a dummy variable for merger deals occurring after the enactment of SOX. These tests reveal that after SOX, a 1-standard-deviation increase in parachute impor- tance is associated with a statistically significant decline in deal value of about $148 million. While this marginal effect is higher than that of the $127 million we estimate for the entire sample period, the pre- and post-SOX point estimates are not statistically different. We also add a similar interaction term to an ac- quirer return regression similar to those reported in Table 7. This additional test shows that after SOX, increasing parachute importance by a single standard devi- ation raises the return to the acquirer by almost 1%. In general, according to the robustness tests in this section, the reported parachute wealth effects related to both target and acquirer shareholders continue to obtain after SOX.

H. Changes of Parachute Importance during Merger Negotiations

The vast majority of parachutes we examine are in place in the CEOs’ com- pensation contracts before their firms become takeover targets. We note that 116 out of 851 target CEOs in our sample (about 14%) do not have a parachute prior to the start of merger negotiations. However, 23 of the 116 firms that do not of- fer a parachute put one in place once merger talks begin. In addition, 30 of the 735 firms that do have parachutes for their CEOs raise their value during merger negotiations. Removing these 53 observations from our sample does not alter our results. We also run premium regressions similar to those in Table 5 in which the key independent variable is a dummy that is “1” if targets either augment the size of an existing golden parachute or put one in place during negotiations (the 53 cases described above). Similar to Hartzell et al. (2004), the estimates for this variable are negative but not statistically significant.

V. Summary and Conclusions

The debate surrounding golden parachutes (particularly during acquisitions) has recently intensified. Advocates argue that during mergers, parachutes induce target CEOs to act in the best interests of their shareholders. Opponents claim that it is unfair to provide managers in the acquired firm with a financial safety net despite the fortunes of their shareholders. Regulatory actions have intensi- fied this controversy: New securities laws mandate the disclosure of parachute compensation during mergers. We summarize this controversy with the follow- ing research question: When a firm becomes a takeover target, do parachutes align the incentives of the target managers receiving them and the shareholders in the targeted firms that grant them? We frame this question in the context of well-known hypotheses in corporate finance: incentive alignment and managerial interest.

We test these hypotheses in a recent sample of 851 acquisition bids dur- ing 1999–2007 using a novel measure of parachute importance that captures the moral hazard meeting the target CEO. Our measure scales the parachute payment by the expected pay loss this CEO incurs if the merger is completed.

Fich, Tran, and Walkling 1749

In our M&A sample, the unconditional probability of deal completion is 87.8% and the average premium offered is 35.9%. We find that a 1-standard- deviation increase in parachute importance raises this probability to 94.7% but lowers the premium to 33.3%. These estimates imply that the expected premium is very similar (at 31.5%) with or without an increase in parachute importance. This result suggests that it is rational for target CEOs to accept a lower premium when they have a more important parachute because the expected value of their equity-based portfolio is unaltered. We also find a positive association between parachutes and the unconditional takeover premium, indicating that their presence yields an unconditional net gain to shareholders. This result indicates that it is also rational for shareholders to give their CEO a parachute.

The fact that raising the importance of parachutes does not alter the expected premium suggests that these provisions are not really harmful to the target share- holders. Our results indicate that with a more important parachute, target CEOs negotiate a takeover price that reflects their own reservation premium. This pre- mium leads to a certainty equivalent: CEOs accept a lower bid premium to both ensure bid success and trigger the receipt of their merger pay package. We also examine the returns accruing to our sample bidders upon merger announcement. These tests reveal that bidder returns increase in parachute importance. This result documents a transfer of wealth from shareholders of the target to shareholders of the acquirer. At first glance, this finding appears consistent with the managerial interest hypothesis. Nonetheless, the fact that acquirers are able to execute a good deal when more important parachutes are provided to target CEOs does not under- mine the conclusion that target shareholders also benefit from golden parachutes because they get a completed bid they might not have received otherwise. In sum, our results show that conditional on a merger, target shareholders are worse off than they would have been in a deal without a parachute for the target CEO, but they are unconditionally better off, because with a parachute a merger is more likely to occur.

From a public policy perspective, our paper informs the ongoing debate about the effectiveness of golden parachutes and provides timely evidence re- lated to recently passed securities laws regulating parachute provisions during acquisitions.

From an academic perspective, our paper has broad implications for the empirical literature studying whether (and how) the structure of managerial com- pensation affects firm value. In this context, our measure of parachute impor- tance (which reflects the moral hazard issue target CEOs confront) provides a new, unique, and economically informative prism to examine the incentives of parachutes. Indeed, a distinct attribute of our measure of parachute importance is that it leads to a certainty equivalent that is proportional to the expected lost compensation for target CEOs. In addition, our empirical evidence conforms to the theoretical prediction in Ross (2004). He argues that attitudes toward risk de- pend not only on the convexity of an agent’s overall pay schedule, but also on how the schedule maps into more (or less) risk-averse areas of the agent’s utility func- tion to the extent that the mapping can remove the effect of convex (or concave) compensation schedules. Our findings suggest that a relatively more important parachute alters how (and where) the target CEO’s entire pay schedule maps into

1750 Journal of Financial and Quantitative Analysis

his utility function, thereby making the executive more risk averse. Under this possibility, our results offer a solution to the paradox in the literature showing that target CEOs negotiate lower takeover offers when they get personal benefits, even though it is improbable that the value of those benefits always makes these CEOs whole from their acquisition-related personal losses.

Appendix. Heckman (1979) First-Stage Regressions

In this section, we perform the first-stage regressions according to Heckman (1979) to address the sample selection problem on the probability of having a golden parachute and on the probability of becoming a takeover target. All regressions are logit models that use 14,157 firm-years with data available from CRSP, Compustat, and RiskMetrics during 1999–2007. We present the regressions of golden parachute determinants in Table A1 and those of takeover target probability in Table A2. In the second stage of our main tests, the

TABLE A1

Probability of Having a Golden Parachute

Table A1 presents regressions of the first-stage Heckman (1979) selectivity correction on the probability of having a golden parachute. Both regressions are logit models that use 14,157 firm-years with data available from CRSP, Compustat, and RiskMetrics during 1999–2007. Model 1 includes the takeover threat variable from Agrawal and Knoeber (1998), which is defined as the relative frequency of acquisitions in the 2-digit SIC industry of a firm among all Compustat firms over the next 3 years. Model 2 includes poison pill and classified board variables among others as in Comment and Schwert (1995) and Bates et al. (2008). Model 3 controls for industry concentration with the Herfindahl-Hirschman index using sales. All firm financial characteristics are averaged over 3 fiscal years. The p-values in parentheses are White (1980) heteroskedasticity- consistent and adjusted for clustering by firms. *, **, and *** indicate statistical significance at the 10%, 5%, and 1% levels, respectively.

Model 1 Model 2 Model 3

Intercept 1.363*** 0.341 0.227 (0.001) (0.119) (0.388)

log(Assets) −0.028*** −0.021** −0.021** (0.003) (0.033) (0.033)

Q −0.085*** −0.074*** −0.074*** (0.001) (0.001) (0.001)

Leverage 0.416*** 0.392*** 0.393*** (0.001) (0.001) (0.001)

Sales growth 0.001 0.000 0.000 (0.504) (0.708) (0.718)

Liquidity −0.671*** −0.577*** −0.577*** (0.001) (0.001) (0.001)

Free cash flow 0.179** 0.181** 0.181** (0.026) (0.032) (0.032)

Prior year excess return 0.651 0.414 0.413 (0.246) (0.477) (0.478)

Takeover threat 0.222* (0.072)

Poison pill (0, 1) 0.683*** 0.683*** (0.001) (0.001)

Classified board (0, 1) 0.215*** 0.216*** (0.001) (0.001)

Supermajority to approve merger (0, 1) −0.001 −0.001 (0.989) (0.974)

Delaware incorporation (0, 1) −0.032 −0.032 (0.202) (0.203)

Herfindahl-Hirschman index 0.478 (0.434)

Year and industry fixed effects Yes Yes Yes N 14,157 14,157 14,157 Adj. R 2 0.094 0.149 0.149 Pr > χ2 0.001 0.001 0.001

Fich, Tran, and Walkling 1751

inverse Mills ratios derived from these first-stage regressions are included in the estimation as a variable to control for endogenous self-selection.

TABLE A2

Probability of Becoming a Takeover Target

Table A2 presents regressions of the first-stage Heckman (1979) selectivity correction on the probability of becoming a takeover target. All models are logit regressions that use 14,157 firm-years with data available from CRSP, Compustat, and RiskMetrics during 1999-2007. Model 1 uses variables similar to those in Palepu (1986). Model 2 includes a poison pill indi- cator variable as in Comment and Schwert (1995). Model 3 augments the specification Bates et al. (2008). Model 4 controls for industry concentration with the Herfindahl-Hirschman index using sales at the industry level. All firm financial charac- teristics are averaged over 3 fiscal years. The p-values in parentheses are White (1980) heteroskedasticity-consistent and adjusted for clustering by firms. *, **, and *** indicate statistical significance at the 10%, 5%, and 1% levels, respectively.

Model 1 Model 2 Model 3 Model 4

Intercept −2.246*** −2.279*** −2.463*** −2.518*** (0.001) (0.001) (0.001) (0.001)

log(Assets) −0.099*** −0.101*** −0.103*** −0.103*** (0.001) (0.001) (0.001) (0.001)

Q −0.073*** −0.074*** −0.069*** −0.069*** (0.001) (0.001) (0.001) (0.001)

Leverage 0.177 0.158 0.120 0.121 (0.122) (0.136) (0.298) (0.296)

Sales growth 0.000 0.000 0.000 0.000 (0.952) (0.798) (0.942) (0.946)

Liquidity 0.010 0.018 0.038 0.039 (0.942) (0.887) (0.776) (0.774)

Free cash flow −0.216 −0.190 −0.220 −0.220 (0.282) (0.413) (0.277) (0.277)

Prior year excess return 0.093 0.090 0.040 0.041 (0.919) (0.918) (0.965) (0.965)

Poison pill (0, 1) 0.100*** 0.073* 0.073* (0.008) (0.071) (0.071)

Control share law (0, 1) −0.133 (0.300)

Business combination law (0, 1) −0.005 (0.942)

Classified board (0, 1) −0.077** −0.076* (0.050) (0.051)

Golden parachute (0, 1) 0.213*** 0.213*** (0.001) (0.001)

Supermajority to approve merger (0, 1) −0.093 −0.093 (0.101) (0.102)

Delaware incorporation (0, 1) 0.111*** 0.111*** (0.006) (0.006)

Herfindahl-Hirschman index 0.229 (0.813)

Year and industry fixed effects Yes Yes Yes Yes N 14,157 14,157 14,157 14,157 Adj. R 2 0.069 0.072 0.077 0.077 Pr > χ2 0.001 0.001 0.001 0.001

References Agrawal, A., and C. Knoeber. “Managerial Compensation and the Threat of Takeover.” Journal

of Financial Economics, 47 (1998), 219–339. Ahern, K. R. “Bargaining Power and Industry Dependence in Mergers.” Journal of Financial

Economics, 103 (2012), 530–550. Aktas, N.; E. de Bodt; and R. Roll. “Negotiations under the Threat of an Auction.” Journal of Financial

Economics, 98 (2010), 241–255. Bange, M. M., and M. A. Mazzeo. “Board Composition, Board Effectiveness, and the Observed Form

of Takeover Bids.” Review of Financial Studies, 17 (2004), 1185–1215.

1752 Journal of Financial and Quantitative Analysis

Bargeron, L.; F. Schlingemann; R. Stulz; and C. Zutter. “Why Do Private Acquirers Pay So Little Compared to Public Acquirers?” Journal of Financial Economics, 89 (2008), 375–390.

Bargeron, L.; F. Schlingemann; R. Stulz; and C. Zutter. “Are Acquisition Premiums Lower Because of Target CEOs’ Conflicts of Interest?” Working Paper, Ohio State University (2010).

Bates, T. W.; D. A. Becher; and M. L. Lemmon. “Board Classification and Managerial Entrenchment: Evidence from the Market for Corporate Control.” Journal of Financial Economics, 87 (2008), 656–677.

Bates, T. W., and M. L. Lemmon. “Breaking Up Is Hard To Do? An Analysis of Termination Fee Provisions and Merger Outcomes.” Journal of Financial Economics, 69 (2003), 469–504.

Bebchuk, L.; A. Cohen; and C. Wang. “Golden Parachutes and the Wealth of Shareholders.” Harvard Law and Economics Discussion Paper No. 683 (2010).

Bebchuk, L., and Y. Grinstein. “The Growth of Executive Pay.” Oxford Review of Economic Policy, 21 (2005), 283–303.

Berkovitch, E., and N. Khanna. “A Theory of Acquisition Markets: Mergers versus Tender Offers, and Golden Parachutes.” Review of Financial Studies, 4 (1991), 149–174.

Black, F., and M. Scholes. “The Pricing of Options and Corporate Liabilities.” Journal of Political Economy, 3 (1973), 637–654.

Boone, A., and J. H. Mulherin. “How Are Firms Sold?” Journal of Finance, 62 (2007), 847–875. Boschen, J., and K. Smith. “You Can Pay Me Now and You Can Pay Me Later: The Dynamic

Response of Executive Compensation to Firm Performance.” Journal of Business, 68 (1995), 577–608.

Cheffins, B. “Did Corporate Governance ‘Fail’ during the 2008 Stock Market Meltdown? The Case of the S&P 500.” Business Lawyer, 65 (2009), 1–65.

Chhaochharia, V., and Y. Grinstein. “CEO Compensation and Board Structure.” Journal of Finance, 64 (2009), 231–261.

Comment, R., and G. W. Schwert. “Poison or Placebo? Evidence on the Deterrence and Wealth Effect of Modern Antitakeover Measures.” Journal of Financial Economics, 39 (1995), 3–43.

Cotter, J. F.; A. Shivdasani; and M. Zenner. “Do Independent Directors Enhance Target Shareholder Wealth During Tender Offers?” Journal of Financial Economics, 43 (1997), 195–218.

Cotter, J. F., and M. Zenner. “How Managerial Wealth Affects the Tender Offer Process.” Journal of Financial Economics, 35 (1994), 63–97.

Denis, D. J., and J. M. Serrano. “Active Investors and Management Turnover Following Unsuccessful Control Contests.” Journal of Financial Economics, 40 (1996), 239–266.

Evans, J.; T. Noe; and J. Thornton, Jr. “Regulatory Distortion of Management Compensation: The Case of Golden Parachutes for Bank Managers.” Journal of Banking and Finance, 21 (1997), 825–848.

Fama, E., and K. French. “Industry Costs of Equity.” Journal of Financial Economics, 43 (1997), 153–193.

Fich, E. M.; J. Cai; and A. L. Tran. “Stock Option Grants to Target CEOs During Private Merger Negotiations.” Journal of Financial Economics, 101 (2011), 413–430.

Fich, E. M., and A. Shivdasani. “Financial Fraud, Director Reputation, and Shareholder Wealth.” Journal of Financial Economics, 86 (2007), 306–336.

Francis, B.; I. Hasan; and Z. Sharma. “Do Incentives Create Innovation? Evidence from CEO Com- pensation Contracts.” Working Paper, Rensselaer Polytechnic Institute (2009).

Gompers, P.; J. Ishii; and A. Metrick. “Corporate Governance and Equity Prices.” Quarterly Journal of Economics, 118 (2003), 107–155.

Grinstein, Y., and P. Hribar. “CEO Compensation and Incentives: Evidence from M&A Bonuses.” Journal of Financial Economics, 73 (2004), 119–143.

Hall, B.; and J. Liebman. “Are CEOs Really Paid Like Bureaucrats?” Quarterly Journal of Economics, 113 (1998), 653–691.

Hartzell, J. C.; E. Ofek; and D. Yermack. “What’s in It for Me? CEOs Whose Firms Are Acquired.” Review of Financial Studies, 17 (2004), 37–61.

Heckman, J. “Sample Selection Bias as a Specification Error.” Econometrica, 47 (1979), 153–161. Huddart, S., and M. Lang. “Employee Stock Option Exercises: An Empirical Analysis.” Journal

of Accounting and Economics, 21 (1996), 5–43. Jarrell, G., and A. Poulsen. “Stock Trading before the Announcement of Tender Offers: Insider Trad-

ing or Market Anticipation?” Journal of Law, Economics and Organizations, 5 (1989), 223–248. Jensen, M. C., and W. H. Meckling. “Theory of the Firm: Managerial Behavior, Agency Costs, and

Ownership Structure.” Journal of Financial Economics, 3 (1976), 305–360. Jensen, M. C., and J. L. Zimmerman. “Management Compensation and the Managerial Labor Market.”

Journal of Accounting and Economics, 7 (1985), 3–9. Jenter, D., and K. Lewellen. “CEO Preferences and Acquisitions.” Working Paper, Stanford University

(2011).

Fich, Tran, and Walkling 1753

Knoeber, C. “Golden Parachutes, Shark Repellents, and Hostile Takeovers.” American Economic Review, 76 (1986), 155–167.

Lambert, R. A., and D. F. Larcker. “Golden Parachutes, Executive Decision Making, and Shareholder Wealth.” Journal of Accounting and Economics, 7 (1985), 179–203.

Lambrecht, B., and S. Myers. “A Theory of Takeover and Disinvestment.” Journal of Finance, 62 (2007), 809–845.

Lefanowicz, C. E.; J. R. Robinson; and R. Smith. “Golden Parachutes and Managerial Incentives in Corporate Acquisitions: Evidence from the 1980s and 1990s.” Journal of Corporate Finance, 6 (2000), 215–239.

Machlin, J.; H. Choe; and J. A. Miles. “The Effects of Golden Parachutes on Takeover Activity.” Journal of Law and Economics, 36 (1993), 861–876.

Maddala, G. S. Limited Dependent and Qualitative Variables in Econometrics. Cambridge, MA: Cambridge University Press (1983).

Malatesta, P. H., and R. A. Walkling. “Poison Pill Securities: Stockholder Wealth, Profitability, and Ownership Structure.” Journal of Financial Economics, 20 (1988), 347–376.

Malmendier, U., and G. Tate. “CEO Overconfidence and Corporate Investment.” Journal of Finance, 60 (2005), 2661–2700.

Manso, G. “Motivating Innovation.” Journal of Finance, 66 (2011), 1823–1860. Masulis, R.; C. Wang; and F. Xie. “Corporate Governance and Acquirer Returns.” Journal of Finance,

62 (2007), 1851–1889. Moeller, S.; F. Schlingemann; and R. Stulz. “Wealth Destruction on a Massive Scale? A Study of

Acquiring Firm Returns in the Recent Merger Wave.” Journal of Finance, 60 (2005), 757–782. Moeller, T. “Let’s Make a Deal! How Shareholder Control Impacts Merger Payoffs.” Journal of

Financial Economics, 76 (2005), 167–190. Officer, M. S. “Termination Fees in Mergers and Acquisitions.” Journal of Financial Economics, 69

(2003), 431–467. Palepu, K. G. “Predicting Takeover Targets: A Methodological and Empirical Analysis.” Journal

of Accounting and Economics, 8 (1986), 3–35. Rhodes-Kropf, M., and D. Kadyrzhanova. “Concentrating on Governance.” Journal of Finance, 66

(2011), 1649–1685. Rhodes-Kropf, M., and S. Viswanathan. “Market Valuation and Merger Waves.” Journal of Finance,

59 (2004), 2685–2718. Ross, S. A. “Compensation, Incentives, and the Duality of Risk Aversion and Riskiness.” Journal

of Finance, 59 (2004), 207–225. Schlingemann, F.; R. Stulz; and R. Walkling. “Divestitures and the Liquidity of the Market for

Corporate Assets.” Journal of Financial Economics, 64 (2002), 117–144. Schwert, G. W. “Markup Pricing in Mergers and Acquisitions.” Journal of Financial Economics,

41 (1996), 153–192. Shleifer, A., and R. W. Vishny. “Stock Market Driven Acquisitions.” Journal of Financial Economics,

70 (2003), 295–489. Wang, C., and F. Xie. “Corporate Governance Transfer and Synergistic Gains from Mergers and

Acquisitions.” Review of Financial Studies, 22 (2009), 829–858. White, H. “A Heteroskedasticity-Consistent Covariance Matrix Estimator and a Direct Test for

Heteroskedasticity.” Econometrica, 48 (1980), 817–838. Wulf, J. “Do CEOs in Mergers Trade Power for Premium? Evidence from Mergers of Equals.” Journal

of Law, Economics, and Organization, 20 (2004), 60–101. Yermack, D. “Remuneration, Retention, and Reputation Incentives for Outside Directors.” Journal

of Finance, 59 (2004), 2281–2308.

Copyright of Journal of Financial & Quantitative Analysis is the property of Cambridge University Press and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use.