ARTICLE IN PRESS. Journal of Financial Economics



Similar documents
Do Shareholders Listen? M&A Advisor Opinions and Shareholder Voting

Morgan Stanley Online (ClientServ) Trading Agreement

Going Private: What Companies Need to Know. Ellenoff Grossman & Schole LLP

FREQUENTLY ASKED QUESTIONS ABOUT RULE 10b - 18 AND STOCK REPURCHASE PROGRAMS

The Role of an Independent Financial Adviser in ESOP Installation Transactions

IPS RIA, LLC CRD No

Private Placements In Public Equity (PIPEs): Best Practices for FINRA Members 1

EXPERT GUIDE Mergers & Acquisitions May 2014

Role of the Independent Fiduciary. Samuel W. Halpern Area Executive Vice President (Ret.) Institutional Investment & Fiduciary Services

The Determinants and the Value of Cash Holdings: Evidence. from French firms

Antonio Gastelum, Inc. INVESTMENT MANAGEMENT AND ADVISORY AGREEMENT ( Agreement )

More M&A activity over the next 18 months is expected

The Georgia Brokerage Relationships in Real Estate Transactions Act

L2: Alternative Asset Management and Performance Evaluation

How To Set Up A Committee To Check On Cit

Incentives for Improving Cybersecurity in the Private Sector: A Cost-Benefit Perspective

Spark Advisors Advisory Agreement

SCHEDULE TO. NATIONSTAR MORTGAGE HOLDINGS INC. (Name of Subject Company (Issuer) and Filing Person (Offeror))

Law Firm Expertise and Merger and Acquisition Outcomes

Our Value Proposition

United States of America Takeover Guide

Institutional Trading, Brokerage Commissions, and Information Production around Stock Splits

Delaware Supreme Court s Rulings Regarding Fiduciary Duties in Alternative Entities

Code of Conduct for Persons Providing Credit Rating Services

Research Objectivity Standards

LONDON STOCK EXCHANGE HIGH GROWTH SEGMENT RULEBOOK 27 March 2013

JPMORGAN CHASE & CO FORM FWP. (Free Writing Prospectus - Filing under Securities Act Rules 163/433) Filed 07/16/14

WISLAR WEALTH MANAGEMENT, LLC 10 East Broad Street Hopewell, NJ 08525

PHOENIX NEW MEDIA LIMITED STATEMENT OF POLICIES GOVERNING MATERIAL, NON-PUBLIC INFORMATION AND THE PREVENTION OF INSIDER TRADING

RETIREMENT INSIGHTS. Understanding your fiduciary role. A plan sponsor fiduciary guide

Lawyer Experience and IPO Pricing *

F R E Q U E N T L Y A S K E D Q U E S T I O N S A B O U T R E G U L A T I O N F D

OFFICE OF THE REGISTRAR OF INSURANCE AND RETIREMENT FUNDS CODE OF CONDUCT FOR INSURANCE BROKERS

Unison Advisors LLC. The date of this brochure is March 29, 2012.

The Changing Landscape of Executive Compensation after Dodd-Frank

Investment Advisory Agreement

BAILEY CAVALIERI LLC ATTORNEYS AT LAW

Compilation of Financial Statements: Accounting and Review Services Interpretations of Section 80

THE AMERICAN LAW INSTITUTE Continuing Legal Education

January 29, Role of the Board of Directors ( The Board ) and Director Responsibilities 2. Selection of Chairman 3.

The More Things Change? Delaware Supreme Court Applies Business Judgment Standard of Review in Going-Private Transaction

Discretionary Investment Management Agreement. Premier SEP IRA. Ameritas Investment Corp "O" Street Lincoln, NE

INDEX METHODOLOGY MSCI REIT PREFERRED. Index Construction and Maintenance Methodology for the MSCI REIT Preferred Index.

Insights Spring ESOP Transaction Insights. Michael McGinley

Meridian Client Update

Stock Buybacks. SEC Hot Topics Institute November 28, Robert L. Kimball, Vinson & Elkins L.L.P. Janet Wright, Dell Inc.

Directed Trusts: Delaware v. Florida Fiduciary Forum September 16, 2014

VALUING BANKING STOCKS

2012 Contingent Consideration Study

THE FAIR CREDIT REPORTING ACT RECENT (AND PENDING) DEVELOPMENTS. By Kevin G. Fitzgerald 1 FAIR AND ACCURATE CREDIT TRANSACTIONS ACT OF 2003

Earnouts in Mergers & Acquisitions Transactions. 23 December

Part 2A of Form ADV: Firm Brochure. Accredited Investors Inc W. 73rd Street Edina, MN 55439

GOING PUBLIC IN CANADA

Harmonic Investment Advisors

The MC Academy The Employee Benefits and Executive Compensation Series EXECUTIVE COMPENSATION

DISCLOSEABLE TRANSACTION ACQUISITION OF LEAPFROG ENTERPRISES, INC. BY WAY OF MERGER

Private Equity Newsletter

How to Run Offshore Management Companies and Funds with Substance and Corporate Governance

How To Manage The Risks Of An Erisa Fiduciary

Investor Sub Advisory Group GOING CONCERN CONSIDERATIONS AND RECOMMENDATIONS. March 28, 2012

VC - Sample Term Sheet

CHARTER OF THE AUDIT COMMITTEE OF THE BOARD OF DIRECTORS OF ARMSTRONG FLOORING, INC. ADOPTED AS OF MARCH 30, 2016

In an ever changing business and social environment it has become increasingly

Form ADV Part 2A Brochure March 30, 2015

Investment Advisory Agreement

A Fiduciary by Any Other Name Thoughts on Properly Delegating Fiduciary Duties. James P. Baker and David M. Abbey

CHIEF JUSTICE STEELE S POSITION CONCERNING THE APPLICATION OF DELAWARE COMMON LAW FIDUCIARY RULES TO DELAWARE LIMITED PARTNERSHIPS AND

for Analysing Listed Private Equity Companies

Reverse Mergers Barry I. Grossman Ellenoff Grossman & Schole LLP

Keystone Financial Planning, Inc.

Advisory Agreement: Asset Management Services

Chapter Five: Respect for Human Rights in Joint Ventures Relationships

Item 2 Material Changes

Restrictions on Buying and Selling Stock and Securities (Insider Trading Policy)

Restrictions on Research and Investment Banking Personnel; Information Barrier Procedures

FINRA-Broker Dealer Investment Banking Due Diligence

Shareholder Litigation Involving Mergers and Acquisitions

UNITED STATES SECURITIES AND EXCHANGE COMMISSION WASHINGTON, DC FORM 8-K

JACKSON NATIONAL LIFE INSURANCE COMPANY JACKSON NATIONAL LIFE INSURANCE COMPANY OF NEW YORK. 1 Corporate Way Lansing, Michigan 48951

Code of Conduct for Credit Rating Agencies

Mergers and Acquisitions in China

Mutual Fund Governance Independent Directors Rule

COMMUNICATING THE VALUATION REPORT

SUPERIOR COURT OF THE STATE OF CALIFORNIA FOR THE COUNTY OF LOS ANGELES IN RE: STRUCTURED SETTLEMENT LITIGATION

LIONS GATE ENTERTAINMENT CORP. DISCLOSURE POLICY

Spinning: FINRA Rule 5131

Transcription:

Journal of Financial Economics 91 (2009) 179 207 Contents lists available at ScienceDirect Journal of Financial Economics journal homepage: www.elsevier.com/locate/jfec Are fairness opinions fair? The case of mergers and acquisitions $ Darren J. Kisgen a, Jun QJ Qian a,, Weihong Song b a Boston College, Chestnut Hill, MA 02467, USA b University of Cincinnati, Cincinnati, OH 45221, USA article info Article history: Received 26 January 2007 Received in revised form 31 January 2008 Accepted 10 March 2008 Available online 14 November 2008 JEL classification: G34 G24 J33 abstract Over the period 1994 2003, 80% of targets and 37% of acquirers obtain a third-party assessment of the fairness of a merger or acquisition. These fairness opinions do not affect deal outcomes when used by targets, but they affect deal outcomes when used by acquirers. The deal premium is lower in transactions if the acquirer obtains a fairness opinion, and is further reduced if multiple advisors provide an opinion. However, the acquirer s announcement-period return is 2.3% lower if the acquirer has a fairness opinion, especially if the acquirer pays a high premium, indicating that investors are skeptical of these transactions. & 2008 Elsevier B.V. All rights reserved. Keywords: Fairness opinion Merger Conflict of interest Deal premium Announcement return 1. Introduction Fairness opinions are prevalent in mergers and acquisitions. Over the period 1994 2003, approximately 80% of $ We appreciate helpful comments from an anonymous referee, Anup Agrawal, Jennifer Bethel, Steven Davidoff, Alex Edmans, Michael Ferguson, Brian Hatch, Knut Heen, Edie Hotchkiss, Yong Kim, Laurie Krigman, Matthew Rhodes-Kropf, Erik Sirri, Steve Wyatt, and seminar/ session participants at Babson College, Boston College, Chinese University of Hong Kong, University of California, Davis, University of Cincinnati, Washington University in St. Louis, the American Finance Association meetings (Chicago), the European Finance Association meetings (Zurich), and the Financial Intermediation Research Society meetings (Shanghai). The legal advice of Andrew Solomon, Esq., the legal research assistance of Andrew Sodl, Esq., the research assistance of Andrew Donahue, Amanda LiDonni, Victoria Mayer, and Dexin Zhou on data collection, and financial support from Boston College and University of Cincinnati are gratefully acknowledged. The authors are responsible for all remaining errors. Corresponding author. E-mail address: qianju@bc.edu (J. QJ Qian). target firms and 37% of acquiring firms obtained a fairness opinion. Fairness opinions (FOs hereafter) are third-party assessments, typically rendered by an investment bank, of the fairness of an acquisition or merger, particularly with regard to price. An FO could provide value to managers and boards as an indication of proper due diligence, and to shareholders as a mechanism to ensure transaction quality. 1 FOs could be particularly beneficial for acquiring firms, since many acquiring firms appear to experience significant losses in mergers (e.g., Moeller, Schlingemann, and Stulz, 2005). However, since the same investment 1 FOs became popular among merging firms after the 1985 case of Smith vs. Van Gorkom (488 A.2d. 858, 875 878 (1985)). The Delaware Supreme Court ruled that the target s Board of Directors was liable for breaching its duty of care owed to the corporation and its shareholders. The court focused on the hasty process by which the Board approved the deal, despite the fact that the merger offer was made at a substantial premium over the target price. By obtaining FOs boards take a significant step in effectively discharging their duty of care regarding the transaction. 0304-405X/$ - see front matter & 2008 Elsevier B.V. All rights reserved. doi:10.1016/j.jfineco.2008.03.001

180 D.J. Kisgen et al. / Journal of Financial Economics 91 (2009) 179 207 bank often provides both advisory services and an FO, a potential conflict of interest can arise since the bank has an incentive to see the transaction completed. The average fee paid for an FO is small relative to the overall fees paid to investment banks in M&As, which include significant fees contingent on deal completion. 2 Firm executives might also have an incentive to see a transaction completed, and therefore desire a favorable FO for the transaction. Using manually compiled data on a large sample of deals announced over the period 1994 2003, this paper is the first to examine the use and effect of FOs in M&A transactions. Specifically, we examine which types of transactions and firms are most likely to obtain an FO, and whether the use of an FO for either the target firm or the acquiring firm affects the probability of deal completion, the transaction deal premium, and the acquirer announcement return. We also examine whether the reputation of the advisor providing the FO and the independence of the advisor (or lack thereof) affects these transaction outcomes. In particular, we examine whether the integrity of the FO is compromised due to a conflict of interest if the advisor that provides the FO also receives contingency fees. In evaluating these outcomes, we consider two hypotheses regarding the use of an FO. We conjecture that FOs are used at a minimum for legal protection, with managers and board members of merging firms obtaining FOs to protect themselves against potential lawsuits triggered by the outcome or performance of the M&A transaction. The legal protection only hypothesis stipulates that FOs serve only to provide legal protection for managers and board members. One reason that an FO might have no other impact on a transaction is that conflicts of interest could inhibit FOs from containing any substance beyond a rubber stamp for the transaction. 3 We also consider the transaction improvement hypothesis, whereby FOs could significantly improve transaction outcomes if they lead to more rigorous deal negotiation (e.g., one side of the transaction uses its FO as a means to negotiate more favorable terms, or if an FO limits the potential for subsequent renegotiation by capping the price that the acquirer pays). FOs could also be significant in a transaction if, prior to the transaction being announced, firms try to obtain an FO for both high-quality and low-quality deals, but investment banks only provide FOs for high-quality deals because they are unwilling to risk their reputation on lower-quality deals. 4 In this 2 During the period 1994 2003, the median fee for an acquirer or target FO was $300,000, whereas the median total advisory fees for an acquiring (target) firm in a completed merger were $2.39 ($2.37) million. 3 For example, in its acquisition of Bank One Corp., J.P. Morgan s inhouse bankers not only provided the advisory service but also the FO. This alleged conflict of interest was cited as one of the reasons for a (J.P. Morgan) shareholder lawsuit. Another reason for the lawsuit was that Bank One s CEO offered to sell the company at a lower price as long as he became the CEO of the joined firm, a proposal rejected by J.P. Morgan (Wall Street Journal, 12/29/04). 4 Throughout the paper, a higher-quality transaction is defined as one with a higher probability of wealth creation for the firm s shareholders. situation, the FO does not improve the deal terms directly, but the presence of an FO indicates that the transaction is of higher quality due to this screening process. In either case, the implication of the transaction improvement hypothesis is that FOs provide some benefit to shareholders. We interpret the tests of the paper in the context of these two hypotheses regarding the use of FOs. We find that the use of FOs by a target firm does not significantly affect the deal premium, the acquirer s announcement-period return, or the likelihood of deal completion. These results on the target side are consistent with the view that FOs provide little value to shareholders and serve only to provide legal protection for managements and boards. The conflict of interest of investment banks might prevent the FO from providing value to shareholders. The use of an FO by the acquiring firm, however, does affect deal outcomes. Transactions with an acquirer FO have a 4.3% lower average premium paid for the target, consistent with the transaction improvement hypothesis. We also find that deal premiums are further reduced in transactions in which acquirers have two or more financial advisors and one or more FOs. These results indicate that this multi-advisor structure is particularly beneficial to acquirer shareholders. With two or more advisors, the likelihood that conflicts of interest will result in a biased FO is lower because multiple investment banks would have to produce a similarly biased opinion. Obtaining an FO does not occur exogenously, so we conduct these tests using both ordinary least squares (OLS) regressions and a two-equation treatment procedure; the same results are found using either approach. We also find that the use of an acquirer FO increases the likelihood of deal completion and is associated with lower announcement-period returns compared to those without an acquirer FO, consistent with the legal protection only hypothesis. Both of these effects are incrementally more significant in transactions with a high premium, indicating that the market is particularly skeptical about the quality of the transaction and the value of an acquiring firm FO when the firm pays a high premium for the target and uses an FO. By contrast, the use of multiple advisors in providing an FO for the acquirer does not lead to a higher probability of deal completion as compared to deals without an FO, and the market does not react negatively to the announcement of this type of deal. These results indicate mixed effects of FOs for acquiring firms shareholders. The use of FOs is associated with lower deal premiums, suggesting that investment banks have fulfilled their obligation of improving the pricing of the transactions. However, the market reacts negatively to transactions with an acquirer FO the announcement return is 2.3% lower than in transactions without an acquirer FO. The announcement return is the market s net assessment of all aspects of the transaction, including what the existence of an FO conveys about the quality of the deal overall. A lower announcement return for transactions with an FO indicates that any benefit an FO provides in reduced premiums is dominated by the

D.J. Kisgen et al. / Journal of Financial Economics 91 (2009) 179 207 181 perception that an FO indicates a transaction of lower quality overall. 5 We find additional evidence that conflicts of interest and advisor quality affect the objectivity and quality of FOs. We find that acquirers with FO providers that also receive a fee contingent on deal completion have significantly lower announcement-period returns. Less than 10% of FOs are provided by an advisor that is otherwise unaffiliated with the transaction, i.e., one that does not receive an additional fee for advisory services, and the market does not react negatively to the use of these advisors. We also find evidence that the reputation of acquirers FO advisors positively affects deal quality. The use of top-tier FO advisors on the acquirer side reduces the deal premium, while the use of lower-tier FO advisors is associated with a higher probability of completing the deal, higher premiums paid, and significantly lower announcement returns. Thus, the incentives of lower-tier advisors might not be aligned with those of the acquirer shareholders to complete quality transactions at reasonable prices. Regarding the decision to obtain an FO, we find that FOs are more likely to be included in transactions in which the complexity and uncertainty of the deal is higher and when legal risk due to conflicts of interest is higher. The use of FOs on either the target or acquirer side becomes significantly more likely in tender offers that are friendly, and the use of FOs on the acquirer side becomes more likely for larger deals and when the method of payment is the acquirer s stock. A friendly acquisition is more likely to appear to be a non-arm s-length transaction. Transactions financed by the acquirer s stock are typically more complex with more uncertain outcomes and the boards must be more cautious in approving the deal. Larger deals are more likely to come under public scrutiny, in particular from more sophisticated institutional investors. The decision by managements and boards to include an FO as part of due diligence is also related to measures of corporate governance. For example, we find that the use of an FO on either the target or acquirer side becomes more likely when the board is smaller and has a higher fraction of outside members, both of which are proxies for more effective monitoring of management. Finally, transactions with FOs are not necessarily of higher or lower quality, as measured by the target firm s return on equity, sales growth, or leverage prior to deal announcement. Our data collection process begins with Securities Data Company s (SDC) M&A database, which specifies the identities of and fees paid to the financial advisors involved in a given deal. While information from SDC on advisors that perform the advisory task is adequate, information on the advisors providing the FOs is sporadic and inaccurate. To ensure that our sample is both 5 The average size of an acquirer is about six times that of its target in our sample, so the lower premium (4.3%) associated with the use of an FO implies that the positive pricing impact of the FO to shareholders is around 0.7% of acquirer value. Therefore, any pricing benefits could reasonably be dominated by other factors, since the gain to acquirers that use an FO from lower premiums is small relative to the negative announcement returns. comprehensive and accurate on FOs, for each deal from SDC we manually search and extract information from the merging firms online SEC filings (available for deals announced in 1994 or later). In more than one-third of our initial sample of 2,072 deals, we find at least one FO for the target and/or acquirer where the SDC database indicated no FO, or more FOs and advisor than indicated in the SDC database. We believe that our manually collected dataset provides a significantly more accurate representation of deal structure during this period. Our paper relates to existing research on financial advisors in M&A transactions. Previous literature has investigated the advisory role played by investment banks in M&A (e.g., McLaughlin, 1990, 1992; Servaes and Zenner, 1996; Rau, 2000; Kale, Kini, and Ryan, 2003; Allen, Jagtiani, Peristiani, and Saunders, 2004). We extend this literature by studying the role of FOs and their issuing advisors. A few recent papers also examine FOs in M&A (e.g., Bowers, 2002; Bowers and Latham, 2004). These papers, however, rely on SDC data alone and thus miss a significant number of deals in which we identified FOs from SEC filings but not reported by the SDC. Bowers (2002) only examines the determinants of FOs on the target side, while our paper examines the use, structure, and effectiveness of both target and acquirer FOs. Our paper is also the first to examine the impact of FOs on deal premiums and performance, and the first to find any relation between FOs on the acquirer side and deal pricing and performance. The rest of the paper is organized as follows. We review the legal and institutional background of FOs in Section 2, and develop our hypotheses in Section 3. In Section 4 we provide empirical tests using a large M&A sample. Section 5 concludes. Appendix A provides a summary of the predictions regarding acquirer FOs based on the two hypotheses, and Appendix B contains descriptions of all the variables in our empirical tests. 2. Institutional background on the use of fairness opinions FOs are obtained in mergers, acquisitions, goingprivate transactions, negotiated share repurchases, and other significant corporate transactions. We focus on M&A in this paper, since this is the most prevalent, and arguably most significant, arena in which FOs are used. The management and board of a firm must consider the business judgment rule in a merger or acquisition which establishes a gross negligence standard. While this standard largely remains vague in practice, courts advanced a standard that requires some lower level of dereliction than ordinary negligence (e.g., Gevurtz, 2000). The Smith v. Van Gorkom (1985) case helped define the gross negligence standard and the importance of an outside opinion for M&A. In this case, the court concluded that the target board had been grossly negligent in recommending the merger because it had not made an informed decision with regard to the offer price. The determination of damages was based on the difference between the fair market value of the shares and

182 D.J. Kisgen et al. / Journal of Financial Economics 91 (2009) 179 207 the price paid. Total damages were $33.5 million, $10 million of which was covered by insurance, with the remainder assessed against the directors. The court did not specifically state that an FO would meet the hurdle of an informed decision: We do not imply that an outside valuation study is essential to support an informed business judgment; nor do we state that fairness opinions by independent investment bankers are required as a matter of law. (Smith, 488 A.2d. at 873) Despite this language, several cases subsequent to the Van Gorkom decision established the use of an FO as meeting the standard set out by this case (e.g., Cottle v. Storer Communication Inc.). As such, firms that obtain an FO effectively remove legal risk resulting from their negligence in their duty of care regarding the merger decision. Thus, we postulate that one value of obtaining an FO is to protect managements and boards from subsequent shareholder lawsuits. Nonetheless, FOs are not required by law in corporate control transactions, despite the legal precedents set by Van Gorkom and other cases. Additionally, no standard has been promulgated regarding who may provide an FO (e.g., a consulting firm, accountancy firm, investment bank) or the methodology to be employed in the FO (e.g., discounted cash flows, comparable ratio analysis, comparable transaction premium, etc.). Furthermore, fairness can be stated in terms of a range of values, which in some cases can be very wide. As a result of the varying techniques allowed in an FO and the lack of standardization, legal actions taken by shareholders against a company or the investment bank that provides the FO have been largely unsuccessful. 6 Investment banks typically exclude specific facts from the opinion to avoid future scrutiny regarding the approach used to determine fairness. Courts for the most part are satisfied as long as there is evidence of the process of an FO having been undertaken; they place less emphasis on the quality of that opinion although at least one exception to this was the case of Pure Resources Inc., Shareholders Litigation (2002), in which the court emphasized the valuation approach in addition to the mere existence of the FO (Davidoff, 2006). Federal securities laws stipulate that an opinion is untrue if it is issued without a genuine belief or reasonable basis or if there is a gross disparity between prediction and fact. In re McKesson HBOC, Inc., the court concluded, In the case of a fairness opinion, then, the plaintiff must plead with particularity why the statement of opinion was objectively 6 The Board of Directors is liable to the company s shareholders as a fiduciary of sorts and thus has a duty of care responsibility to the shareholders. Further, it is by nature of contract that the board employs advisors to draft an FO, and therefore contract law and theories of ordinary negligence control their relationship. On the other hand, the company s shareholders are third party beneficiaries of the contractual relationship formed between the board and the FO advisors, but they are afforded some legal standing to hold FO advisors liable. This standing comes in the form of Securities Exchange Act 17 (C.F.R. 240.14a-9(a)), which makes drafters of false statements provided in proxy solicitations, such as an FO letter submitted to the board, liable. See Gevurtz (2000) for more details. and subjectively false. In the cases of City Partnership Co. v. Lehman Bros., Inc. and Rosser v. New Valley Corp. (2005 WL 1364624 (Del.Ch.)), the FO advisors were accused of issuing incompetent opinions, but in both cases the courts held that while the FOs could have been better they were not outside the realm of reasonableness. In the cases of General Motors (Hughes), Shareholder Litigation (2005 WL 1089021 (Del.Ch.)) and City Partnership Co. vs. Lehman Bros. Inc. (344 F. Supp.2d 1241 (2004)), the FO advisors had a long-term relationship with management, but they were cleared of any wrongdoing because these advisors disclosed all existing and potential conflicts of interest to the board and shareholders in their FO letters. In some rare cases, plaintiffs have been successful in taking action against financial advisors (e.g., Herskowitz v. Nutri/System or Mills Acquisition Co. v. Macmillan), but generally it is difficult to establish liability. FOs are generally obtained prior to the board s decision on a transaction and, consequently, prior to the deal being announced. For an FO to provide the legal protection that most boards seek, the board must demonstrate that it reviewed the opinion before making any final decision to proceed with the transaction. This concern leads to FOs being provided early in the process. For example, in the acquisition of Handspring, Inc. by Palm, Inc. in 2003, the initial press release announcing that a definitive agreement had been reached for the acquisition indicated that Morgan Stanley&Co.Incorporatedservedasfinancialadvisorto Palm and also provided a fairness opinion and Credit Suisse First Boston LLP served as financial advisor to Handspring and also provided a fairness opinion. Ultimately, however, since an FO is not required for any transaction, the decision rests with the board on whether and when to obtain the opinion. For transactions that have material renegotiation after the announcement of the deal, boards can also seek a revised FO based on the new terms of the transaction, but in most cases only one version of an opinion is obtained and disclosed in companies filings. The use of an FO is disclosed in the proxy statements that are sent to shareholders for the M&A transaction, and companies also state the existence of an FO during the announcement of the deal. Generally, only the FO letter, which is typically one or two pages long and contains the conclusion of whether a particular advisor views the pending deal as fair or not or if the price falls within a certain range, is provided to the board for inclusion in legal documentation. We manually search (through M&A documents and filings) and read FO letters for each deal in our sample. Almost all the letters we find state that the FO advisor considers the pending deal as fair from a financial point of view. The actual valuation specifics and related documents, while sometimes described verbally to the board, are typically not given to the board or disclosed. More recently, however, the disclosure of FOs has been more substantive, with valuation ranges and methodologies, due to increased regulatory scrutiny. 7 In many FOs, the investment banks also include both general 7 With a sample of negotiated mergers between 1998 and 2005, Cain (2007) examines whether investment banks valuation analysis disclosed in FOs contains information that is not available to the market.

D.J. Kisgen et al. / Journal of Financial Economics 91 (2009) 179 207 183 and specific disclaimers that they do not intend to evaluate non-pricing aspects of the transaction. Regulators have more recently begun to scrutinize the use of FOs (see, e.g., Davidoff, 2006). The SEC evaluated the FO procedure, but ultimately decided only to require increased disclosure regarding the FO. The disclosure requirements set forth in SEC Schedule 13e-3 require additional disclosure in the proxy statement including disclosure regarding whether the firm obtained an FO, the method of arriving at the FO, and whether there are any potential conflicts of interest in providing the opinion. Regarding the method for arriving at fairness, however, firms still vary substantially in what they disclose, with many still providing only boilerplate disclosure. The NASD also examined the use of FOs, but its conclusions also focused only on increased disclosure and not on recommendations regarding the use or execution of the FO. Rule 2290 of the NASD stipulates several additional disclosure requirements, including whether the advisor will receive compensation contingent on successful completion of the underlying transaction, whether the advisor has had a material relationship with the company in the last 2 years, and what information was provided to the advisor by the company in deriving the opinion. The Sarbanes-Oxley Act of 2002 (SOX) has also increased the importance and scrutiny of FOs. SOX directed the SEC to adopt rules that required the disclosure of potential conflicts of interest between investment banks and a firm. SOX also prohibits a firm s auditors from providing an FO. Section 404 of SOX requires the majority of publicly traded companies to include a report on the effectiveness of their internal controls, which when viewed from an M&A perspective indicates a need for greater due diligence that could include an independently provided FO. To summarize, the use of an FO is not required by law but it significantly reduces the risk of lawsuits triggered by the outcome of an M&A transaction. While merging firms and their boards face more stringent disclosure requirements on the use of FOs and potential conflicts of interest on the part of their financial advisors, they have significant discretion in choosing which advisors will provide the FO as well as the timing, structure, and content of the opinion. 3. Hypotheses development on the use, impact, and structure of fairness opinions The use of an FO in an M&A transaction represents a contractual agreement between the merging firm and the FO provider (investment bank). The use of an FO thus suggests that the benefits of the FO outweigh the costs of providing the FO for both of the contracting parties. The benefits of an FO to a merging firm, described in more detail below, are legal protection only or transaction improvement. From the perspective of the investment bank(s), the primary benefit of providing an FO is to build or strengthen the relationship with the firm, which can potentially increase business opportunities for the investment bank(s). As described in Section 2, FOs can be used to provide legal protection for merging firms boards and managements without any positive wealth effect for shareholders (the legal protection only hypothesis). For example, in Smith v. Van Gorkom, the board was found liable under the gross negligence standard because it did not undertake a duty of care in evaluating the terms of the deal. The court cited that the board did not have adequate information regarding the transaction. In contrast, in Treadway v. Care Corp. (1980) concerning a friendly sale of stock, the board was exonerated because it had received a written FO and report detailing the transaction. Since the Van Gorkom case, boards have felt that an FO is particularly useful in major corporate control decisions and transactions that involve conflicting interests (Gac and Shaw, 1995). FOs not only provide legal protection to merging firms boards and managements, they also provide value to the shareholders (the transaction improvement hypothesis). As mentioned before, FOs might be particularly beneficial for acquiring firms shareholders since many acquiring firms appear to experience significant losses in mergers (Moeller, Schlingemann, and Stulz, 2005). Anecdotal evidence also indicates that FO advisors prevent firms from making bad acquisitions (e.g., Sechler and Sweeney, 1999). An FO can lead to better deal negotiation and renegotiation. Prior to the deal being announced, an FO can assist management in negotiating, and it might prevent management from undertaking unfavorable transactions. Similarly, after the deal is announced, the existence of an FO could limit price revisions in subsequent renegotiations. Alternatively, investment banks might decline to provide FOs for deals that they deem to be of low quality because they are unwilling to risk their reputation, and thus those transactions that proceed without an FO might on average be of lower quality. In either case, the FO provides value to shareholders. The cost of an FO to the firm includes management time, since the process of producing an FO involves significant interaction between the advisor and the firm. This cost is higher for an advisor that is less familiar with the company to begin with, which can explain why firms often choose FO providers that have an existing relationship with the firm despite the potential conflict of interest. Additional costs to the firm include the potential for disclosure of competitive secrets in the process and the time to produce an FO, which could delay the deal at different points in the process. These costs are again greater with an investment bank not involved in advising the transaction. Preparing an FO does not take much time for an investment bank that is already an advisor to the transaction, since the advisory role typically includes valuation work, but for a new, independent advisor the valuation process might take several months. The cash cost of the opinion, with the median around $300,000 but ranging from $75,000 to $500,000 or even higher, is an expense for the firm, but it represents a revenue source for the investment bank, albeit small compared to the revenues from M&A advisory services. Perhaps the most significant cost of FOs for investment banks is the potential loss of their reputation from certifying

184 D.J. Kisgen et al. / Journal of Financial Economics 91 (2009) 179 207 low-quality deals. Since investment banks have significant reputational capital, providing an opinion for a questionable deal carries a large potential cost. In the rest of this section, we develop testable predictions on the use of FOs and their impact on various deal outcomes based on the two main hypotheses. These predictions guide us in interpreting our empirical results in the next section. To sort out the opposite effects of FOs implied by the two hypotheses, we develop additional predictions for the impact of various structures of FOs and types of FO advisors on deal outcomes. Since there is more variation in the use of FOs for acquirers, we focus more on acquirer FOs in our empirical analysis, and Appendix A provides a summary of the implications regarding acquirer FOs. 3.1. The use of fairness opinions (FO as a dependent variable) Our first set of predictions (top panel in Appendix A) concerns how the use of FOs relates to firm and deal characteristics. If merging firms use FOs for legal protection, FOs are more likely to be included in transactions in which the perception of wrongdoing is greatest and the legal consequences are most severe. Specifically, the legal protection only hypothesis predicts that FOs are more likely to be used by both the target and acquirer firms in transactions in which the deal size is large, the transaction is friendly, or the mode of payment is the acquirer s stock. Larger deals have a greater impact on shareholders and are more likely to raise the attention required to initiate a lawsuit. Larger deals could also be more likely to have an FO assuming a fixed cost of an FO and a benefit to shareholders of an FO that increases with the size of the deal. We examine this in the empirical tests by conducting the tests with and without small deals (less than $100 million). For a target selling the entire firm (as compared to selling a division), for example, the relative deal size is by definition large and the duty of care of the target board to the shareholders is clearly to ensure that the final sale price is fair or sufficiently high. For the acquiring firm, both relative and absolute sizes are relevant: a higher relative size has more impact on shareholder returns and a higher absolute size draws more attention. However, larger acquiring firms could have more resources to conduct internal due diligence for an M&A transaction, and hence the importance of an outside due diligence procedure, in the form of an FO, might be less than that for smaller acquirers. In a friendly deal, it is difficult for rival bidders to acquire the target without the support of target management. A friendly deal is thus more likely to raise questions than a hostile deal, in which the target management is more reluctant and the pricing is determined by market forces. Therefore, the acquiring firm s board must be more cautious in approving a friendly deal, because the motive for such a deal is more likely to be driven by agency problems, and the risk of being regarded as having neglected the duty of care is higher. If the acquirer s stock is the main method of payment, the deal is more complicated and its outcome is more uncertain, as stock prices and in particular, the acquirer s stock prices can fluctuate during the merger process. Hence, the benefits of using FOs to hedge legal risk for both the target and acquirer firms managements and boards can be significantly higher than the costs of providing the opinion. While the transaction improvement hypothesis also predicts that the use of FOs by both merging firms is more likely in large transactions since the impact on the shareholders payoff is greater, the prediction regarding the likelihood of an acquirer FO in friendly or stock deals is unclear. On the one hand, since these deals are either more questionable with respect to conflicts of interest or more complex with respect to pricing, potential advisors are perhaps more reluctant to endorse them in order to protect their reputational capital. On the other hand, thorough and objective due diligence in the form of an FO can be more valuable for shareholders in these deals. The transaction improvement hypothesis does suggest that an FO is more likely to be used in transactions outside the industry of the acquirer, since the acquirer would benefit more from an expert opinion. On the other hand, it is not clear whether the legal risk of being accused of negligence of duty of care or misconduct is higher in a related or diversifying merger. The use of FOs also depends on the corporate governance of the merging firms. Merging firms with strong internal governance tend to implement a thorough and effective due diligence process; see Holmstrom and Kaplan (2001) for a review of governance and M&A activities. If this process requires external certification, firms will use an FO (according to the transaction improvement hypothesis). Managers in weakly governed firms might pursue a questionable acquisition for private benefits, and will either be unable to solicit an FO from a reputable advisor or will retain a less reputable advisor to provide an FO to remove potential legal risk (the legal protection only hypothesis). 3.2. Impact of FOs on transaction outcomes (FO as an explanatory variable) Our second set of predictions relates to the impact of FOs on various deal outcomes (second panel of Appendix A). If an FO is used for legal protection only, it could facilitate deal completion because managements and boards are willing to push more strongly for completion knowing that they have legal protection. The legal protection only hypothesis also implies that an acquirer FO will not improve transaction pricing since it is only a rubber stamp. Furthermore, since the likelihood of a lawsuit against an acquirer is positively related to the magnitude of the deal premium (over the target price), the use of acquirer FOs could be associated with higher deal premiums. Finally, if the market holds the belief that an FO is used by the acquirer s management and board to facilitate the completion of a questionable deal, the acquirer s announcement-period return will be negative and lower than that of deals without an acquirer FO

D.J. Kisgen et al. / Journal of Financial Economics 91 (2009) 179 207 185 because the FO indicates that the transaction is overall of lower quality. The transaction improvement hypothesis also has several implications regarding deal outcomes. First, if FOs provide value to shareholders there should be a positive impact on deal pricing. Acquirer and target FO advisors, however, have different objectives in evaluating the sale price: while target advisors attempt to push for the highest price, acquirer advisors hope to buy the target on the cheap. Accordingly, target FOs will be positively associated with the deal premium, but the opposite is true for acquirer FOs. Second, the transaction improvement hypothesis implies higher announcement-period returns for transactions that have an acquirer FO, since higherquality transactions are associated with FOs. Although FOs are generally obtained prior to the deal being announced, in some cases the existence of an FO is not made public until the proxy statements are sent, and sometimes the FO is provided after deal announcement. This data limitation works against finding a significant relation between the use of an FO and the acquirer s announcement-period returns. To account for this to some extent, we measure the announcement return using both a shorter and longer time window. Finally, the transaction improvement hypothesis could imply a higher or lower probability of deal completion with an FO. It could imply a lower probability of deal completion because lower-quality transactions are more likely to be rejected with the due diligence provided by the FO advisors. Given data limitations, however, we are unlikely to observe the transactions that have been rejected due to the FO process, since we only observe deals after they have been announced and the FO has already been provided. However, if the FO reduces the potential for renegotiation after the announcement of the deal, the likelihood of deal completion is still reduced. On the other hand, if transactions with FOs are viewed to be stronger deals due to the presence of an FO, shareholders might be more likely to approve transactions with an FO, implying that an FO increases the probability of deal completion. 3.3. Impact of FO structure and advisor type on transaction outcomes To analyze the cross-sectional variations of the effects of FOs among transactions with at least one acquirer FO, we develop a final set of predictions (bottom panel of Appendix A) regarding the impact of various FO structures and types of FO advisors on deal outcomes. We begin by discussing which structures and advisor types are more likely to provide transaction improvement, and conclude by outlining the specific implications for transaction outcomes. Conflicts of interest for the financial advisor can obstruct the fairness of FOs (e.g., Bebchuk and Kahan, 1989; Davidoff, 2006). Accordingly, we categorize two types of FO advisors as unaffiliated for which the conflict is minimized: either independent advisors (e.g., consulting firms) that specialize in providing FOs without being involved in other aspects of M&A or investment banking businesses in general, or advisors (e.g., investment banks) that are not a part of the advisory group hired by one of the merging firms managements despite providing those services in general. We define an affiliated advisor as one directly involved with the advisory group and paid an additional fee. The conflict of interest is also related to the fee structure, and we postulate that the conflict is reduced when the advisor is separated from other fee-providing aspects of the transaction. If a larger fraction of the total advisory fee is contingent on the completion of the deal, then the advisor has a stronger incentive to complete the deal in order to receive the contingent fee, and thus a bad merger is more likely to be consummated. Therefore, we further distinguish affiliated advisors as those with additional non-contingent fees and those with fees that are contingent upon deal completion. The transaction improvement hypothesis implies that unaffiliated advisors are more likely to provide high-quality FOs because they do not face a conflict of interest problem. In its acquisition of Bank One Corp., JP Morgan Chase chose its own bankers to provide the advisory service, and not surprisingly, the same advisors also endorsed the fairness of the deal. While these arrangements do not imply that the deal is necessarily of low quality, it is understandable why shareholders might be suspicious of the objectivity and quality of the opinion. Regarding the number of advisors, the transaction improvement hypothesis implies that a multi-advisor structure could be particularly beneficial to acquirer shareholders. With two or more advisors, the negative impact of the potential conflict of interest of each of the advisors on the overall objectivity of the FO can be significantly reduced, as the reputation concerns of each investment bank lower the likelihood that all of the banks produce similarly biased opinions. We identify transactions with two or more advisors providing an FO, which we argue is best for transaction improvement; transactions with two or more advisors but only one providing an FO; and transactions with only one advisor that also provides an FO, which we argue is the least attractive for transaction improvement. We also examine the quality of the FO provider. The reputation of a financial advisor has a positive impact on the outcome and performance of IPOs/SEOs and M&A transactions (e.g., Chemmanur and Fulghieri, 1994; Rau, 2000; Kale, Kini, and Ryan, 2003). The transaction improvement hypothesis posits that firms use more reputable advisors because they are interested in improving the quality of the FOs, while lower-quality advisors are more willing to provide biased, or at least less informative, opinions. Further, despite conflicts of interest, higherquality advisors might be more likely to provide highquality FOs because reputation concerns can overcome conflicts of interest, whereas a low-quality advisor could issue a biased opinion to generate fees even if it is unaffiliated. Given these arguments, we examine whether the structure of the FO and the quality of FO advisors affect deal outcomes on the acquirer side of the transaction.

186 D.J. Kisgen et al. / Journal of Financial Economics 91 (2009) 179 207 We separate the transactions with acquirer FOs into subsamples in which any transaction improvement effects are likely to be more significant. According to the transaction improvement hypothesis, the hiring of a toptier FO advisor indicates that the pending deal is of high quality, which would lead to positive announcementperiod returns; multiple advisors or an unaffiliated advisor should also be associated with relatively better announcement returns. The transaction improvement hypothesis also predicts that better-structured FOs and more reputable FO advisors lead to better prices in the transaction, i.e., lower deal premiums for the acquirer. 4. Empirical tests and results In this section we perform empirical tests on a large sample of M&A deals. We focus on how the decision to use an FO relates to the characteristics of merging firms, including corporate governance measures, and on how an FO affects deal outcomes, particularly the deal premium and the acquirer s announcement returns. We emphasize those results that relate directly to the implications of the two hypotheses of the paper, as summarized in Appendix A. At the end of this section, we provide a discussion of the overall welfare effects of FOs on acquirer shareholders. 4.1. Description of data Our data collection begins with the SDC M&A database. The initial sample includes all mergers and acquisitions announced from January 1984 November 2003, with both merging firms publicly listed in the US and information on financial advisors of at least one of the merging firms available. SDC provides information on the identity and the contingency fees of advisors that perform the advisory task, but information on the FO advisors is incomplete. To ensure accuracy, for each deal from SDC we manually search and extract information from merging firms online SEC filings, which are available for deals announced in 1994 or later. We also rely on the Dow Jones News Retrieval Service, in particular the Dow Jones Interactive company filings database. As a result, M&A transactions announced prior to 1994 are excluded. In more than Table 1 Fairness opinions in M&A transactions. This table presents the use and structure of FOs and information on the investment banks providing the FOs in M&A deals during 1994 2003. We also split the whole sample into merger and tender offer subsamples. Information on FOs and investment banks is obtained from SDC s M&A database, and manually verified and augmented by searching through merging firms online SEC filings and Dow Jones News Retrieval Service. For the category Use of FO, NO means a transaction has no FO for the acquiring firm (or the target firm); YES means there is at least one FO for the acquiring firm (target firm). Under the category Structure of FOs, FO 11 indicates that the acquiring firm (target firm) has one financial advisor and the advisor also provides an FO; FO 21 means the acquiring firm (target firm) has two or more financial advisors and one of them provides an FO; FO 22 indicates that the acquiring firm (target firm) has two or more financial advisors and two or more of them provide an FO. For the category FO advisor affiliation, Unaffiliated means none of the acquirer s (target s) FO advisors is affiliated with the advisory group of the deal; Affiliated1 means each of the acquirer s (target s) FO advisors also receives an advisory fee that is not contingent on deal completion; and Affiliated2 indicates that one or more acquirer s (target s) FO advisors also receives an advisory fee that is contingent on deal completion. Finally, under the category FO advisor reputation, FO_Tier1 indicates that the ranking of the acquirer s (target s) FO advisor (see Table 2) is in the top 5 (top tier); FO_Tier2 indicates that the ranking of the FO advisor is between 6 and 20 (second tier); and FO_Tier3 means that the ranking is 21 and lower (third tier). In transactions with multiple FO advisors, the reputation of the FO advisor group is determined based on the highest-ranked FO advisor. Full sample (N ¼ 1,509) Merger subsample (N ¼ 1,175) Tender subsample (N ¼ 334) Acquiring firm Target firm Acquiring firm Target firm Acquiring firm Target firm (1) (2) (3) (4) (5) (6) Use of FO NO 947 307 643 211 304 96 YES 562 1,202 532 964 30 238 Structure of FOs FO 11 471 984 443 801 28 183 FO 21 54 138 53 105 1 33 FO 22 37 80 36 58 1 22 FO advisor affiliation Unaffiliated 50 72 42 60 8 12 Affiliated1 107 200 103 148 4 52 Affiliated2 405 930 387 756 18 174 FO advisor reputation FO_Tier1 211 373 205 301 6 72 FO_Tier2 172 371 159 298 13 73 FO_Tier3 179 458 168 365 11 93

D.J. Kisgen et al. / Journal of Financial Economics 91 (2009) 179 207 187 one-third of our initial sample of 2,072 deals, online documents revealed at least one FO and the issuing advisor(s) for the target and/or acquirer where the SDC database indicated no FO, or added an additional FO and issuing advisor to the information indicated on SDC. Our manually collected dataset thus provides a more accurate representation of FOs for deals announced during the period 1994 2003. Applying other standard selection criteria, our final merger sample includes 1,509 M&A transactions. 8 Table 1 summarizes the use and structure of FOs for the 1,509 transactions in our sample period. We also split the entire sample into merger and tender offer subsamples in Table 1 and in all of our empirical tests. As Column 2 indicates, in only 307 (20.3%) transactions is no FO issued for the target firm, indicating that these opinions are a standard component of M&A transactions on the target side, especially in mergers (Column 4 vs. Column 6). Conditional on having at least one investment bank as an advisor, the most common structure for the target firm is to have one advisor and one FO ( TFO 11 in 984 out of 1,202 deals; see Column 2), with the same investment banker providing both advisory services and the FO. The prevalence of this structure is perhaps surprising, since it is the one most likely to create a conflict of interest for the investment bank, as compared to an independent advisor providing the FO or two or more advisor/investment banks providing the opinions. Merging firms might choose this structure, however, because the investment bank that provides the advisory services already has substantial knowledge of the firm and the pending transaction, thereby making an FO a relatively straightforward task. Firms could also be reluctant to disclose confidential information to an additional advisor/investment bank. However, a significant number of target firms choose alternate structures (see Columns 4 and 6), which allows for the opportunity to test whether one particular structure is more effective in providing objective evaluation of a transaction. Contrary to the target FO structures, a significant number of acquiring firms (62.8%) choose not to have an FO (Column 1). Similar to target firms, the majority of acquiring firms with at least one FO (471 out of 562 deals) have one advisor and one FO ( AFO 11 ). However, acquiring firms also choose different FO structures; for example, 54 firms hire two or more investment banks but only one of them conducts an FO ( AFO 21 ; Column 1). In the AOL- Time Warner merger, a well-known deal that took place in the late 1990s, AOL hired three investment banks (Salomon Smith Barney, Merrill Lynch, and Goldman Sachs) as advisors but only Salomon Smith Barney 8 Deals are also excluded from the final sample if the value of the transaction is lower than $5 million or if deals have pending status. Both the target and acquirer must be listed on the NYSE, AMEX, or NASDAQ when the deal is announced and included in the Center for Research and Security Prices (CRSP) database and Compustat Annual files for at least one year prior to the announcement date. The acquirer cannot be a unit trust, REIT, or ADR, must have one year pre-acquisition stock returns available from CRSP, and cannot have a toehold of more than 50% prior to the acquisition. provided an FO for the acquisition. Time Warner also chose the two or more advisors, one FO structure ( TFO 21 ), with Morgan Stanley and Wasserstein Perella Group Inc. as advisors but only Morgan Stanley provided an FO. In 37 out of 562 deals with at least one acquirer FO, the acquirer hires two or more investment banks and two or more of them provide an FO ( AFO 22 ). With this multiopinion structure, it is more likely that at least one FO advisor is not affiliated with the advisory group and receiving an additional fee. With two or more advisors conducting an FO, reputation effects are also likely to be present. An advisor will be less likely to provide an unsubstantiated opinion if it knows that the report will be compared to another investment bank s report, and two or more advisors are less likely to collude with the client to produce an incorrect or biased FO. Table 1 provides more information on the potential conflict of interest of the FO advisors (and their investment banks). In only 72 out of 1,202 deals (Column 2) on the target side and 50 out of 562 deals (Column 1) on the acquirer side, the FO advisor is not affiliated with the advisory group; in transactions with multiple FO advisors, each of the FO advisor must be unaffiliated with the advisory group in order for the FO advisors group to be qualified as unaffiliated. In the majority of deals (930 deals on the target side and 405 deals on the acquirer side), the FO advisor is affiliated with the advisory group and receives a fee contingent on deal completion ( TFO_Affiliated2 and AFO_Affiliated2 ). Notice that in deals with multiple FO advisors, as long as one FO advisor receives a contingency fee the FO advisors group is classified as Affiliated2. For example, in the AOL-Time Warner deal, both Salomon Smith Barney (advisor and FO provider for AOL) and Morgan Stanley (advisor and FO provider for Time Warner) are affiliated advisors and were paid an additional fee contingent on deal completion. This structure is most likely to create a conflict of interest for the advisor. In another 200 deals on the target side and 107 deals on the acquirer side, each of the FO advisors is also affiliated with the advisory group but received a constant fee that is not contingent on deal completion ( TFO_Affiliated1 and AFO_Affiliated1 ). In our subsequent tests, we include dummy variables to examine how the structure of FOs and the conflict of interest of FO advisors affect the quality of the opinions. Finally, Table 1 sorts the deals by the reputation of FO advisors (sorted into three tiers); an advisor is ranked based on the number of deals in which this advisor provides an FO. Table 2 provides information on the identity of FO advisors and the rankings of these advisors based on our M&A sample. Previous studies have provided several measures of the reputation of financial advisors and investment banks. Consistent with prior studies, in our empirical tests we use a discrete ranking variable based on a three-tier ranking system of the FO advisors. 9 Unlike 9 For example, Megginson and Weiss (1991) use market shares of investment banks to measure reputation. Bowers and Miller (1990) and Servaes and Zenner (1996) divide financial advisors into two tiers, while Rau (2000) uses a three-tier ranking system based on the market shares

188 D.J. Kisgen et al. / Journal of Financial Economics 91 (2009) 179 207 Table 2 Rankings of fairness opinion advisors. This table lists the top 30 financial advisors that have issued FOs in M&A transactions over the period 1994 2003. We use a discrete, three-tier system, with the ranking of an advisor based on the number of M&A deals in our sample in which the advisor provides an FO. The market share of an FO advisor is the number of deals for which this advisor provided FOs divided by the total number of deals in our sample (1,509). We have two sets of rankings of FO advisors, one on the acquirer side and the other on the target side. Third-tier advisors include all the advisors not ranked in the top 20, but the table only lists the 10 highest-ranked advisors within this tier. In transactions with multiple FO advisors, the reputation of the FO advisor group is determined based on the highest-ranked FO advisor. Rank Acquirer s FO advisor No. of acquisitions Market share (%) Target s FO advisor No. of acquisitions Market share (%) Top tier (rank 1 5) Top tier (rank 1 5) 1 Merrill Lynch & Co 50 3.31 Goldman Sachs & Co 93 6.16 2 Salomon Smith Barney 45 2.98 Merrill Lynch & Co 71 4.71 3 Morgan Stanley & Co 44 2.92 Donaldson Lufkin & Jenrette 71 4.71 4 Donaldson, Lufkin & Jenrette 40 2.65 Morgan Stanley & Co 70 4.64 5 Goldman Sachs & Co 32 2.12 Salomon Smith Barney 68 4.51 211 14.0 373 24.7 Second tier (rank 6 20) Second tier (rank 6 20) 6 Bear Stearns 31 2.05 Credit Suisse First Boston 53 3.51 7 Credit Suisse First Boston 28 1.86 Lehman Brothers 40 2.65 8 Lehman Brothers 21 1.39 Bear Stearns & Co 39 2.58 9 Alex Brown & Sons Inc. 11 0.73 Alex Brown & Sons 37 2.45 10 Lazard Freres 10 0.66 Sandler O Neill 28 1.86 11 Keefe Bruyette & Woods 10 0.66 Dillon, Read & Co 26 1.72 12 Piper Jaffray 10 0.66 Keefe Bruyette & Woods 23 1.52 13 Sandler O Neill 8 0.53 Piper Jaffray 20 1.33 14 JP Morgan & Co 8 0.53 PaineWebber 17 1.13 15 Hambrecht & Quist 7 0.46 McDonald Investments 16 1.06 16 Stifel Nicolaus & Co 6 0.40 Broadview 15 0.99 17 Dillon, Read & Co 6 0.40 JP Morgan & Co 15 0.99 18 PaineWebber 6 0.40 Hambrecht & Quist 15 0.99 19 UBS 5 0.33 Lazard Freres 14 0.93 20 Wasserstein Perella Group 5 0.33 Robinson-Humphrey 13 0.86 172 11.4 371 24.6 Third tier (rank 21 and lower) Third tier (rank 21 and lower) 21 Banc of America Securities 5 0.33 Dain Rauscher 11 0.73 22 Chase Manhattan Bank 5 0.33 Wasserstein Perella Group 10 0.66 23 Robinson-Humphrey 5 0.33 CIBC 10 0.66 24 CIBC 5 0.33 Deutsche Bank AG 10 0.66 25 Jefferies & Co 4 0.27 Montgomery Securities 9 0.60 26 Ryan Beck & Co 4 0.27 Robert W Baird & Co 9 0.60 27 Montgomery Securities 4 0.27 Charles Webb & Co 9 0.60 28 Needham & Co 4 0.27 Robertson Stephens & Co 9 0.60 29 NationsBanc Corp 4 0.27 SG Cowen Securities 8 0.53 30 Robert W Baird & Co 4 0.27 Prudential Securities 8 0.53 some previous rankings of advisors that are based on the total size (in dollars) of deals in which the investment banks provide advisory services, we rank FO advisors based on the number of deals in which they provide FOs. Our ranking mechanism makes sense because, while advisory fees (in dollars) are usually contingent on deal completion and increase with deal size, the fees paid to each FO advisor are much lower than the advisory fees. (footnote continued) of investment banks. In our tests below, we also use continuous rankings for FO advisors based on the number of deals (so that an advisor has a higher ranking if the total number of deals in which he provides FOs is larger). Our main results are robust to this alternative ranking system. Moreover, as Bao and Edmans (2006) have shown, reputation measures based on market shares of advisory services (total dollar fees earned) are problematic, because investment banks have an incentive to push deals through, especially large deals for which they earn higher total fees, to boost their rankings. Therefore, a better measure for a more reputable FO advisor is how often they are asked to provide FOs in M&A deals. Our ranking procedure results in a list of 118 unique investment banks and consulting firms (FOs conducted by different subsidiaries of the same bank are credited to that bank) that serve as FO advisors for acquiring firms in our M&A sample, and 194 investment banks/consulting firms acting as FO advisors for target firms. The market share of

D.J. Kisgen et al. / Journal of Financial Economics 91 (2009) 179 207 189 an FO advisor is defined to be the number of deals for which this advisor provided FOs divided by the total number of deals in our sample (1,509). 10 We thus have two sets of rankings of FO advisors, one on the acquirer side and the other on the target side. Table 2 lists the top 30 financial advisors that have issued FOs in M&A transactions over the period 1994 2003. These rankings are similar to other rankings of investment banks based on advisory services in M&A (e.g., League Tables from SDC; Rau, 2000). First-tier FO advisors on the target side include Goldman Sachs, which provided FOs in 93 deals and has a market share of 6.16%, Merrill Lynch (71 deals), Donaldson Lufkin & Jenrette (71 deals), Morgan Stanley (70 deals), and Salomon Smith Barney (68 deals). On the acquirer side, Merrill Lynch, which provided FOs in 50 deals and has a market share of 3.31%, is the highest-ranked FO advisor, while the other top spots are occupied by the same four banks that are ranked in the top five as target FO advisors. Finally, third-tier advisors include all the advisors not ranked in the top 20, but Table 2 only lists the ten highest-ranked advisors within this tier. Notice that in transactions with multiple FO advisors, the reputation of the FO advisor group is determined based on the highest-ranked FO advisor. Table 3 shows firm and transaction statistics based on the structure of FOs conducted for the target and acquirer firms. Larger deals (average deal size) in general have more FOs; for example, deals with two or more FOs on either side are on average four times as large as those with one advisor and one FO. The acquirer is on average much (103.5 times) larger than the target in transactions with no FO on the acquirer side than in deals with at least one acquirer FO (2.26 6.87 times the target size). A significant number of deals with no FO on either side are tender offers or cash deals, defined to be transactions in which at least 50% of the consideration is cash, as compared to deals with at least one FO. From the top panel of Table 3, acquirers with two or more FOs (AFO 22 ) are more likely to be paired with targets that have high market-to-book ratios and hence are more likely to be growth rather than value firms, as compared to acquirers with no more than one FO, which are likely to be paired with targets that have low market-to-book ratios. From the last two columns of Table 3, average M&A fees for the overall transaction, including contingency fees, do not significantly differ for transactions with one FO versus no FO; however, transactions with two or more FOs have substantially larger overall fees. The median fee for an acquirer or target FO ($0.3 million) is much smaller compared to the median total fees for a completed merger ($2.39 million for the acquirer and $2.37 million for the target). The median FO fee also increases when the target or acquirer has two or more FOs instead of one conducted. Finally, acquirers pay the highest premium (46.52%) for the target when no acquirer FO is used, although most of the deals in this group are tender offers; acquirers pay the lowest premium (28.11%) when two or more FOs are conducted on the acquirer side. On the other hand, targets 10 Rau (2000) finds that annual rankings of investment banks providing advisory service in M&A deals are stable over time. with two advisors and two FOs receive the lowest premium (37.3%; bottom panel) while targets with one advisor and one FO receive the highest (44.06%). Some of the preliminary evidence shown in Table 3 is consistent with the legal protection only hypothesis. On the target side, larger deals tend to have one or more FOs, perhaps because the risk of target management and/or boards being accused of insufficient due diligence and other forms of negligence is higher given that there is more at stake. On the acquirer side, acquisitions that are larger, are supported by target management (most stock deals are friendly ), or have a complicated structure (stock instead of cash) are associated with one or more FOs. In these types of acquisitions, the acquirer management and board are more likely to be accused of negligence or perverse motives for mergers, and thus the use of FOs becomes more prevalent. However, the relation between deal premiums and the use of FOs on the acquirer side is consistent with the transaction improvement hypothesis and contradicts the legal protection only hypothesis. 4.2. Empirical tests on the use of FOs Table 4 examines the determinants of the use and structure of FOs as well as the reputation and affiliation of FO advisors on either side using multivariate regressions. Appendix A summarizes the main implications for acquirer FOs for these tests given the two hypotheses of the paper. Panel A presents results from probit regressions in which the dependent variables indicate whether the acquirer and target had at least one FO for the transaction (dummy equals 1; default outcome is no FO). We report marginal effects rather than probit coefficients in this panel. Thus, the reported coefficients represent the change in the probability per unit change in the relevant explanatory variables; for indicator variables, the coefficient represents the change in the probability associated with moving the indicator from 0 to 1. We find that when the deal attitude is friendly instead of hostile in a tender offer, the probability of an acquirer FO increases by about 2.6% (Column 4 of Panel A), and the probability of a target FO increases by 21.4% (Column 9). The friendly dummy is positive but statistically insignificant for the use of FOs on either side in the merger subsample (Columns 1 3 and 6 8). During the sample period, hostile transactions were not prevalent (4.1% of deals). However, the significance of this variable, especially on the target side, is striking. This result supports the legal protection only hypothesis, since friendly deals are arguably more likely to be viewed as non-arm s-length transactions and to trigger a lawsuit. Panel A also indicates that the coefficients on deal size and the method of payment are statistically and economically significant for the use of acquirer FOs in mergers: as deal size doubles, the probability of an acquirer FO in a merger increases by 2.4 4.8% (Columns 1 3). This effect is also robust to excluding small deals (under $100 million), which suggests that the effect is not driven only by the fixed costs of obtaining an FO. The result is consistent

190 Table 3 Deal characteristics by use of fairness opinions. This table presents summary statistics of the 1,509 M&A deals and merging firms, sorted by the structure of FOs. Data are for M&A deals announced between January 1994 and November 2003, in which either the target or acquirer has one or more financial advisors. Dollar amounts ($) are in millions. Acquirer size is the market value of equity of the acquirer measured 2 months prior to deal announcement; relative size is defined as the ratio of acquirer size to target size. Acquirer (target) M/B is defined as the ratio of the market value of equity relative to the book value of equity of the acquirer (target) for the prior fiscal year (COMPUSTST items 24 25/60). A stock offer deal is defined as deals in which the consideration is at least 50% stock. % Tender is the percentage of deals where a tender offer was used. Premium is the percentage difference between the offer price and target share price four weeks prior to the announcement date. The number of non-missing observations for a particular variable is listed below the statistic. Acquirer FO structure Average deal size ($) Average acquirer size ($) Average relative size Average acquirer M/B Average target M/B % Hostile % Tender %Stock offer Average premium Mean (median) acquirer total fee ($) Mean (median) acquirer FO fee ($) a AFO ¼ 0 976.55 15,425.62 103.50 4.28 3.04 6.02% 32.10% 57.02% 46.52% 4.95 (2.77) N/A 947 947 901 934 840 947 947 947 822 178 947 AFO 11 2,505.00 4,394.17 6.87 3.49 2.85 0.85% 5.94% 91.51% 39.86% 4.09 (1.75) 0.68 (0.30) 471 471 461 462 430 471 471 471 430 419 168 AFO 21 6,734.21 6,884.31 2.26 4.97 3.31 0% 1.85% 100% 30.67% 7.21 (3.75) 1.04 (0.50) 54 54 52 51 52 54 54 54 51 53 22 AFO 22 10,632.44 19,945.36 4.22 4.19 5.15 2.70% 2.70% 86.49% 28.11% 13.65 (8.50) 3.63 (1.50) 37 37 36 37 32 37 37 37 36 34 16 Total 1,896.42 11,787.57 66.69 4.06 3.05 4.11% 22.13% 70.05% 43.28 5.03 (2.39) 0.95 (0.30) 1509 1509 1450 1484 1354 1509 1509 1509 1339 684 206 Target FO structure Average deal size ($) Average acquirer size ($) Average relative size Average acquirer M/B Average target M/B % Hostile % Tender %Stock offer Average premium Mean (median) target total fee ($) Mean (median) target FO fee ($) TFO ¼ 0 1,284.14 18,202.34 84.43 4.31 3.42 7.49% 31.27% 61.89% 43.64% 5.10 (2.35) N/A 307 307 286 302 267 307 307 307 258 270 370 TFO 11 1,297.78 9,395.39 69.08 3.94 2.89 2.44% 18.60% 73.37% 44.06% 3.90 (2.00) 0.51 (0.30) 984 984 956 970 894 984 984 984 886 953 469 TFO 21 5,499.44 14,998.90 42.89 4.31 3.34 5.07% 23.91% 65.21% 40.40% 8.86 (3.83) 0.79 (0.40) 138 138 132 132 123 138 138 138 123 136 74 TFO 22 5,394.06 11,055.26 11.14 4.10 3.11 10.00% 27.50% 68.75% 37.30% 12.90 (9.00) 2.17 (1.00) 80 80 76 80 70 80 80 80 72 75 41 D.J. Kisgen et al. / Journal of Financial Economics 91 (2009) 179 207 ARTICLE IN PRESS Total 1,896.42 11,787.57 66.69 4.06 3.05 4.11% 22.13% 70.05% 43.28 5.07 (2.37) 0.66 (0.30) 1509 1509 1450 1484 1354 1509 1509 1509 1339 1434 584 a In some deals, the FO fee and the advisory fee (done by the same advisor) are reported as one single fee.

D.J. Kisgen et al. / Journal of Financial Economics 91 (2009) 179 207 191 Table 4 Determinants of the use and structure of FOs and reputation and affiliation of FO advisors. We examine the determinants of the use of FOs on the acquirer side and target side (Panel A), and the structure, advisor reputation, and affiliation of FOs on the acquirer side (Panel B) and on the target side (Panel C). Data are for mergers and acquisitions announced between January 1994 and November 2003, where either the target or acquirer has advisor(s). Tender ( Friendly ) is a dummy that equals 1 if a tender offer is used (deal attitude is friendly). Same_Industry is a dummy that equals 1 if the acquirer and target are in the same industry (with the same first two-digits of SIC). Toehold is the fraction of target shares held by the acquirer prior to deal announcement. Target ROE is measured as the ratio of earnings to average equity for the prior fiscal year (COMPUSTAT items 20/(60+60(t 1)). Target sales growth is measured as the proportional change in sales over the prior fiscal year. Target D/E is measured as the ratio of debt to equity of the target for the prior fiscal year (COMPUSTAT items 6/60). A_BoardSize (T_BoardSize) is the number of directors on the board of the acquirer (target) prior to acquisition announcement; A_Board_Indep (T_Board_Indep) is the ratio of the number of noninsider board members to the total number of directors on the board of the acquirer (target); A_CEO_Chair (T_CEO_Chair) is a dummy variable equal to 1 if the CEO of the acquirer (target) is the chair of the board of directors; A_InsiderShrPct (T_InsiderShrPct) is the security ownership of officers and directors as a group (in %) reported in the proxy statement of the acquirer (target). Year dummies are included in all models. All other explanatory variables are defined in previous tables. P-values are presented in parentheses below the coefficients. Panel A Probit regressions on whether FOs are used (marginal effects are reported) The dependent variable is a dummy indicating whether one or more acquirer (target) FOs is used; the default is no acquirer (target) FO used. AFO TFO (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) Merger Merger Merger Tender Tender Merger Merger Merger Tender Tender Relative size/100 0.481 0.470 0.264 0.021 0.003 0.001 0.001 0.004 0.001 0.001 (0.106) (0.103) (0.184) (0.311) (0.352) (0.946) (0.782) (0.540) (0.826) (0.905) Ln (deal size) 0.035 0.034 0.069 0.007 0.004 0.004 0.004 0.008 0.013 0.004 (0.000) (0.001) (0.000) (0.126) (0.010) (0.539) (0.565) (0.328) (0.530) (0.859) Stock 0.278 0.307 0.353 0.050 0.030 0.032 0.022 0.007 0.057 0.168 (0.000) (0.000) (0.000) (0.130) (0.077) (0.350) (0.530) (0.844) (0.559) (0.145) Toehold 0.160 0.150 0.063 0.365 0.112 0.189 0.234 0.150 2.471 2.226 (0.608) (0.641) (0.874) (0.158) (0.072) (0.399) (0.286) (0.474) (0.046) (0.088) Friendly 0.205 0.218 0.241 0.026 0.011 0.087 0.094 0.188 0.214 0.221 (0.119) (0.111) (0.157) (0.069) (0.005) (0.415) (0.371) (0.137) (0.016) (0.025) Acq. M/B 0.001 0.001 0.005 0.011 0.008 (0.773) (0.825) (0.109) (0.207) (0.418) Same industry 0.014 0.019 0.023 0.008 0.003 0.039 0.036 0.038 0.041 0.037 (0.649) (0.570) (0.536) (0.531) (0.509) (0.106) (0.139) (0.148) (0.453) (0.545) Target M/B 0.001 0.001 0.008 0.004 0.003 0.004 0.006 0.014 0.014 (0.820) (0.818) (0.241) (0.163) (0.040) (0.234) (0.099) (0.005) (0.024) Target ROE 0.007 0.013 0.002 0.000 0.009 0.013 0.030 0.032 (0.392) (0.279) (0.811) (0.891) (0.347) (0.205) (0.011) (0.013) Target sales growth 0.015 0.015 0.008 0.001 0.004 0.018 0.010 0.007 (0.265) (0.224) (0.365) (0.861) (0.559) (0.328) (0.592) (0.716) Target D/E 0.011 0.009 0.004 0.004 0.000 0.010 0.038 0.032 (0.223) (0.374) (0.583) (0.071) (0.960) (0.191) (0.129) (0.255) A_BoardSize 0.028 0.001 (0.000) (0.476) A_Board_Indep 0.787 0.029 (0.000) (0.018) A_CEO_Chair 0.028 0.017 (0.498) (0.034) A_InsiderShrPct 0.003 0.000 (0.031) (0.166) T_BoardSize 0.011 0.004 (0.005) (0.797) T_Board_Indep 0.206 0.521 (0.000) (0.002) T_CEO_Chair 0.024 0.051 (0.332) (0.401) T_InsiderShrPct 0.002 0.001 (0.005) (0.593) Observations 1049 998 913 295 270 1112 986 878 290 260 Pseudo R-square 0.213 0.206 0.272 0.137 0.252 0.083 0.092 0.133 0.118 0.158

192 D.J. Kisgen et al. / Journal of Financial Economics 91 (2009) 179 207 Table 4 (continued) Panel B Multinomial logit regressions of the structure, advisor reputation, and affiliation of acquirer FOs (merger subsample; logit coefficients are reported) The dependent variable in Model 1 is the choice of one of three possible FO structures of the acquirer: the default group is AFO 11, or one advisor, one FO. The dependent variable in Model 2 is the choice of one of three tiers of FO advisors hired by the acquirer: the default group is AFO_Tier3, or third (lowest) tier AFO advisor. The dependent variable in Model 3 is the choice of one of three types of affiliation FO advisors of the acquirer: the default group is AFO_Unaffiliated, or unaffiliated AFO advisor (not affiliated with the advisory group). (1) (2) (3) AFO structure AFO advisor reputation AFO advisor affiliation AFO 21 AFO 22 AFO_Tier1 AFO_Tier2 Affiliated1 Affiliated2 Relative size 0.357 0.049 0.029 0.031 0.016 0.002 (0.001) (0.086) (0.083) (0.058) (0.463) (0.610) Ln (Deal size) 0.158 0.155 0.819 0.531 0.190 0.030 (0.095) (0.326) (0.000) (0.000) (0.217) (0.834) Stock 1.579 3.405 0.192 0.492 0.365 0.784 (0.048) (0.001) (0.817) (0.576) (0.722) (0.359) Friendly 2.100 1.385 3.895 3.882 0.935 0.105 (0.050) (0.253) (0.001) (0.001) (0.543) (0.936) Target M/B 0.004 0.290 0.044 0.004 0.095 0.337 (0.959) (0.000) (0.390) (0.933) (0.851) (0.449) Same industry 0.434 1.139 0.373 0.496 0.061 0.009 (0.228) (0.019) (0.249) (0.136) (0.539) (0.915) Target ROE 0.074 0.895 0.163 0.230 0.722 0.620 (0.811) (0.001) (0.473) (0.403) (0.037) (0.034) Target sales growth 0.057 1.736 0.185 0.210 0.308 0.669 (0.721) (0.049) (0.163) (0.138) (0.451) (0.060) Target D/E 0.117 0.118 0.141 0.137 0.454 0.506 (0.063) (0.199) (0.126) (0.165) (0.074) (0.041) A_BoardSize 0.003 0.077 0.078 0.016 0.004 0.026 (0.945) (0.207) (0.043) (0.696) (0.961) (0.635) A_Board_Indep 1.453 1.847 0.436 0.240 0.536 0.916 (0.198) (0.228) (0.588) (0.763) (0.696) (0.438) A_CEO_Chair 0.284 0.284 0.421 0.333 0.727 0.627 (0.451) (0.608) (0.173) (0.288) (0.155) (0.174) A_InsiderShrPct 0.010 0.084 0.001 0.004 0.025 0.014 (0.274) (0.001) (0.981) (0.648) (0.183) (0.438) Observations 423 423 423 Pseudo R 2 0.575 0.151 0.395 Panel C Multinomial logit regressions on the structure, advisor reputation, and affiliation of target FOs (merger subsample; logit coefficients are reported) The dependent variable in Model 1 is the choice of one of three possible FO structures of the target: the default group is TFO 11, or one advisor, one FO. The dependent variable in Model 2 is the choice of one of three tiers of FO advisors hired by the target: the default group is TFO_Tier3, or third (lowest) tier TFO advisor. The dependent variable in Model 3 is the choice of one of three types of affiliation FO advisors of the target: the default group is TFO_Unaffiliated, or unaffiliated TFO advisor (not affiliated with the advisory group). (1) (2) (3) TFO structure TFO advisor reputation TFO advisor affiliation TFO 21 TFO 22 TFO_Tier1 TFO_Tier2 Affiliated1 Affiliated2 Relative size 0.001 0.034 0.002 0.001 0.001 0.001 (0.587) (0.021) (0.233) (0.871) (0.404) (0.317) Ln (deal size) 0.295 0.536 0.944 0.580 0.034 0.317 (0.002) (0.000) (0.000) (0.000) (0.826) (0.018) Stock 0.439 0.224 0.343 0.051 0.888 1.274 (0.235) (0.635) (0.241) (0.869) (0.050) (0.001) Friendly 2.509 2.560 3.846 3.236 0.046 0.557 (0.000) (0.001) (0.000) (0.000) (0.964) (0.440) Acq. M/B 0.003 0.043 0.031 0.098 0.092 0.037 (0.897) (0.469) (0.403) (0.001) (0.043) (0.333) Same industry 0.139 0.139 0.243 0.031 0.960 0.818 (0.632) (0.705) (0.293) (0.891) (0.029) (0.040) Target M/B 0.037 0.080 0.083 0.100 0.139 0.084 (0.479) (0.234) (0.057) (0.023) (0.092) (0.086)

D.J. Kisgen et al. / Journal of Financial Economics 91 (2009) 179 207 193 Table 4 (continued ) Panel C Multinomial logit regressions on the structure, advisor reputation, and affiliation of target FOs (merger subsample; logit coefficients are reported) The dependent variable in Model 1 is the choice of one of three possible FO structures of the target: the default group is TFO 11, or one advisor, one FO. The dependent variable in Model 2 is the choice of one of three tiers of FO advisors hired by the target: the default group is TFO_Tier3, or third (lowest) tier TFO advisor. The dependent variable in Model 3 is the choice of one of three types of affiliation FO advisors of the target: the default group is TFO_Unaffiliated, or unaffiliated TFO advisor (not affiliated with the advisory group). (1) (2) (3) TFO structure TFO advisor reputation TFO advisor affiliation TFO 21 TFO 22 TFO_Tier1 TFO_Tier2 Affiliated1 Affiliated2 Target ROE 0.890 0.343 0.022 0.049 0.254 0.167 (0.067) (0.538) (0.844) (0.261) (0.095) (0.264) Target sales growth 0.206 0.133 0.040 0.038 0.387 0.419 (0.376) (0.348) (0.568) (0.470) (0.374) (0.334) Target D/E 0.150 0.174 0.047 0.048 0.184 0.192 (0.126) (0.013) (0.481) (0.448) (0.110) (0.075) T_BoardSize 0.014 0.082 0.052 0.041 0.145 0.067 (0.775) (0.176) (0.211) (0.311) (0.126) (0.454) T_Board_Indep 0.095 3.244 0.337 0.196 0.818 0.740 (0.900) (0.000) (0.520) (0.696) (0.401) (0.381) T_CEO_Chair 0.122 0.471 0.110 0.268 0.412 0.228 (0.694) (0.208) (0.638) (0.218) (0.287) (0.494) T_InsiderShrPct 0.002 0.015 0.006 0.001 0.013 0.009 (0.808) (0.189) (0.414) (0.838) (0.200) (0.336) Observations 723 723 723 Pseudo R-square 0.567 0.186 0.489 with both the legal protection only and transaction improvement hypotheses. When the transaction is financed by the acquirer s stock (dummy equals one if at least 50% of the transaction value is paid by stock) as compared to cash, the probability of an acquirer FO increases by 27.8% 35.3% (Columns 1 3). This result is consistent with the legal protection only hypothesis, but does not contradict the transaction improvement hypothesis. The latter hypothesis does suggest that an FO is more likely to be used if a transaction is outside the industry of the acquirer, since the acquirer would benefit more from an expert opinion, but this implication is not confirmed in Table 4, as same-industry deals are not more or less likely to have an acquirer FO. To the extent that investment banks are less likely to provide FOs for low-quality deals due to reputation concerns, the transaction improvement hypothesis implies that deals that proceed with an acquirer FO are of higher quality, on average. However, none of the characteristics that proxy for the quality of the target firm (e.g., return on equity, sales growth, or leverage prior to deal announcement) positively affect the use of acquirer FOs in mergers. We also expect the decision by merging firms managements and boards to obtain an FO to be affected by corporate governance. We include several commonly used measures of governance, including board size, percentage of outside board members, share ownership by insiders (defined to be officers and directors as disclosed in proxy statements), and whether the CEO is also the chairman of the board. Data for these variables are hand collected from firms proxy statements (using the date closest and prior to deal announcement); sample size for different models thus varies depending on data availability for these measures. Since each of these variables is an imperfect measure of effective governance (see, e.g., Hermalin and Weisbach, 2003), we are careful in drawing any causal conclusions. 11 From Panel A, the use of an FO for acquirers and targets becomes more likely when the size of the board is smaller or when the board has fewer inside members (significant in the merger subsample, Columns 3 and 8; board size is negative but statistically insignificant in the tender offer subsample, Columns 5 and 10). Smaller boards with more outside members can be more effective in monitoring managers (e.g., Boone, Field, Karpoff, and Raheja, 2007) and consistent with the transaction improvement hypothesis, these boards are more likely to approve or request the use of an FO as part of the due diligence process. We also find the acquirer s inside ownership to be positively related to the use of an FO (Column 3), but the opposite is true for target insider ownership (Column 8). It is unclear whether a larger ownership stake provides 11 We also use the G-index based on Gompers, Ishii, and Metrick (2003) as an additional measure for governance, which significantly reduces sample size (after merging the G-index with our sample firms). Overall, there is no significant relation between the G-index and the use of FOs on either side, while our other results regarding the use and structure of FOs do not change qualitatively in models with the index. For brevity we do not report these results in the tables.

194 D.J. Kisgen et al. / Journal of Financial Economics 91 (2009) 179 207 better governance: with a larger ownership stake, the incentives of acquirer insiders can be more aligned with those of shareholders in pursuing a high-quality transaction, but insiders can benefit from low-quality transactions, possibly at the cost of shareholders, thus reducing the positive impact of inside ownership. 12 The incentive effect of ownership for target insiders can also depend on whether the target insiders ultimately retain their positions in the joined firm. In Panels B and C of Table 4 we evaluate which factors are related to the structure, advisor reputation, and affiliation of FOs given the decision to use an FO on the acquirer side (Panel B) and target side (Panel C). While there are a number of findings from these tests, to save space we only present the results that will be used later to provide identification in our two-stage tests on the impact of FOs on deal premiums and the acquirer s announcement returns. These panels indicate that acquirers and targets are more likely to retain higher-quality advisors when the deal size is larger and when the deal is hostile. Acquirers are more likely to use a multi-advisor, multiopinion structure when pursuing a diversifying merger or when there is less insider ownership, and are less likely to hire a high-quality advisor when the board size is large. Targets are more likely to use a multi-opinion structure when the deal size is larger and the transaction is hostile and are less likely to use this structure when the board is more independent. In addition, targets are more likely to use an affiliated advisor if they operate in a different industry from the acquirer or if the mode of payment is stock. We will revisit these results in our analysis on deal premiums and announcement returns. 4.3. The impact of FOs on transaction outcomes Examining the impact of FOs on transaction outcomes is the main contribution of the paper. We first examine the impact of FOs on deal completion. We then examine the impact of FOs on the quality of a transaction, measured by deal premium and the acquiring firm s announcementperiod returns. We exclude governance variables as controls in all regressions on deal outcomes (Table 5, deal completion, Table 6, deal premium, and Tables 7 and 8 on acquirer announcement returns). As discussed before, we view these variables as determinants for the use and structure of FOs. Results from Tables 6 8 are robust to the inclusion of governance variables, which generally do not have a significant impact on deal outcomes themselves. Appendix A summarizes the main implications for acquirer FOs for these tests given the two hypotheses of the paper. When examining the effects of FOs on transaction quality, we recognize that the use of an FO is itself 12 For example, Stulz (1988), Morck, Shleifer, and Vishny (1988), McConnell and Servaes (1990), and more recently, Coles, Lemmon, and Meschke (2007) all argue and find a non-monotonic relation between CEO ownership and firm value and performance. Grinstein and Hribar (2004) document that the CEO of some acquirers receives a cash bonus for completing an M&A transaction regardless of the subsequent performance of the merged firm. endogenously determined by merging firms. As shown in Table 4, the use and structure of FOs on either side are correlated with characteristics of merging firms and the M&A transaction. This endogenous selection process therefore has the potential to bias estimates of the impact of FOs on the merger outcomes that we evaluate. To control for the potential self-selection bias, we employ both OLS regressions and a two-stage procedure when we examine the impact of FOs on deal premiums and the acquirer s announcement returns; we view these as the main tests of the implications of the two hypotheses. Our two-stage model consists of a treatment equation and a regression equation on the transaction outcome (see, e.g., Maddala, 1983, pp. 120 122). We assume that there is an unobservable underlying variable, FAIR*, that determines whether a firm obtains an FO. The treatment rule is that an FO is observed, or a particular type of FO structure or advisor is used, if FAIR* exceeds zero; otherwise, the firm does not obtain an FO or chooses another structure or advisor. Letting Z i denote a column vector of variables that predict whether a firm obtains an FO, the first-stage treatment rule is given by FAIR i ¼ jz i þ u i, (1) where FAIR i ¼ 1ifFAIR * i 40 and FAIR i ¼ 0 otherwise. We use several variables in Z i in the treatment equation (probits), motivated by the results of Table 4. We obtain probit estimates of the treatment equation, Pr (FAIRi ¼ 1 Z i ) ¼ F(jZ i ). From these estimates, the hazard rate, h i for each observation i is computed as h i ¼ fðjz i Þ= FðjZ i Þ, if FAIR i ¼ 1, or h i ¼ fðjz i Þ=f1 FðjZ i Þg, if FAIR i ¼ 0, where f and F are the density and cumulative distribution functions of the standard normal distribution. The second stage regression model is given by Deal premium i =announcement return i ¼ a þ bx i þ gfair i þ lh i þ i. (2) The difference between Eq. (2) and OLS is that the dummy variables on the use and structure of FOs in Eq. (2) are augmented by the hazard rate obtained from Eq. (1). The variables included in the vector X i in Eq. (2) are those shown to have a significant impact on deal premium or the acquirer s announcement-period returns in previous research. 4.3.1. Deal completion Table 5 shows the results (marginal effects) from probit regressions with a binary dependent variable equal to one if a deal is completed and zero otherwise. We control for several firm and deal characteristics that have been shown to affect deal status in these tests. 13 We find that having an FO on the acquirer side increases the likelihood of the completion of a merger (Column 1), but does not significantly impact the completion of tender offers (Column 7). The result for mergers is consistent with the legal protection only hypothesis described in 13 Consistent with previous studies, friendly mergers are significantly more likely to be completed compared to hostile takeovers, while the presence of a competing bid makes it much more difficult to complete the deal (e.g., Schwert, 2000; Betton and Eckbo, 2000).

D.J. Kisgen et al. / Journal of Financial Economics 91 (2009) 179 207 195 Table 5 The impact of FOs on deal completion (marginal effects are reported). We run Probit regressions with the dummy (equals 0 if deal is withdrawn) on deal completion as the dependent variable. In Models 1 through 6, we examine the merger subsample; in Models 7 through 10, we examine the tender offer subsample. Compete is a dummy that equals 1 if there is one or more competing bids for the target. We include year dummies in all models. All the other explanatory variables are defined in previous tables. P-values are presented in parentheses below the coefficients. Merger Tender (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) AFO 0.013 0.012 0.001 0.059 (0.057) (0.057) (0.944) (0.105) TFO 0.003 0.004 0.002 0.007 (0.742) (0.571) (0.694) (0.749) AFO 11 0.014 (0.028) AFO 21 0.006 (0.599) AFO 22 0.013 (0.464) TFO 11 0.000 0.004 (0.994) (0.857) TFO 21 0.006 0.009 (0.577) (0.793) TFO 22 0.016 0.022 (0.175) (0.433) AFO_Tier1 0.005 (0.497) AFO_Tier2 0.012 (0.117) AFO_Tier3 0.015 (0.054) TFO_Tier1 0.002 0.029 (0.835) (0.171) TFO_Tier2 0.005 0.004 (0.549) (0.842) TFO_Tier3 0.004 0.005 (0.641) (0.820) AFO_Unaffiliated 0.014 (0.169) AFO_Affiliated1 0.012 (0.176) AFO_Affiliated2 0.011 (0.116) TFO_Unaffiliated 0.011 0.038 (0.453) (0.540) TFO_Affiliated1 0.005 0.006 (0.663) (0.813) TFO_Affiliated2 0.002 0.012 (0.836) (0.603) Relative size/100 0.032 0.031 0.032 0.032 0.029 0.028 0.019 0.020 0.021 0.019 (0.043) (0.036) (0.042) (0.041) (0.096) (0.073) (0.219) (0.195) (0.172) (0.201) Ln (deal size) 0.001 0.002 0.000 0.001 0.001 0.001 0.003 0.006 0.008 0.005 (0.629) (0.366) (0.990) (0.730) (0.733) (0.631) (0.597) (0.363) (0.171) (0.401) Stock 0.015 0.014 0.015 0.014 0.014 0.012 0.030 0.029 0.026 0.029 (0.026) (0.026) (0.017) (0.031) (0.018) (0.025) (0.184) (0.185) (0.235) (0.226) Toehold 0.054 0.051 0.049 0.052 0.021 0.025 0.185 0.197 0.174 0.224 (0.302) (0.294) (0.320) (0.315) (0.561) (0.464) (0.490) (0.423) (0.451) (0.411) Friendly 0.132 0.135 0.141 0.131 0.112 0.107 0.323 0.302 0.277 0.306 (0.021) (0.020) (0.015) (0.022) (0.044) (0.044) (0.000) (0.000) (0.000) (0.000) Compete 0.175 0.173 0.173 0.172 0.152 0.145 0.104 0.098 0.090 0.099 (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.017) (0.018) (0.019) (0.018) Premium/100 0.025 0.008 (0.024) (0.532) AFO*(premium/100) 0.036 (0.069) Observations 1119 1119 1119 1119 1015 1015 321 321 321 321 Pseudo R-square 0.174 0.183 0.179 0.177 0.186 0.197 0.361 0.356 0.365 0.357

196 D.J. Kisgen et al. / Journal of Financial Economics 91 (2009) 179 207 Section 3.2. An FO on the target side, however, does not affect deal completion for either subset. For mergers, the effect of acquirer FOs on deal completion appears to be mainly driven by the one advisor, one FO structure: the AFO 11 dummy in Column 2 is significant at the 5% level, whereas coefficients on the multi-advisor, one FO (AFO 21 dummy) and multi-opinion (AFO 22 dummy) structures are insignificant relative to the default outcome of no AFO. With one advisor and one FO, the FO advisor is also the overall advisor on the transaction and therefore has other incentives to complete the deal. However, F-tests show no differences in the effect on deal completion for deals with one advisor and one FO versus multi-advisor and multi-opinion deals, perhaps due to the small sizes on the multi-advisor and multi-opinion subsamples. The effect of acquirer FOs on deal completion is also mainly driven by third-tier advisors, or advisors of lowest quality; the magnitude of the coefficient for a third-tier advisor is three times that for a top-tier advisor (Column 3). Finally, in Columns 8 10, we do not include dummy variables indicating acquirer FO structure and advisor characteristics due to a small number of observations in the tender offer subsample. Taken together, these results indicate that an acquirer FO leads to a higher probability of deal completion, particularly with a structure that is most likely to have a conflict of interest and an advisor that has the least reputational capital at risk. We also find that a transaction is more likely to be completed if the acquirer pays a higher premium for the target (Column 5). When these highpremium transactions also have a FO on the acquirer side, the probability of deal completion further increases (Column 6). We will revisit these results when we discuss the relation between the use of FOs and acquirer announcement returns below. 4.3.2. Deal premium Panel A of Table 6 presents results from OLS regressions with deal premiums, defined to be the percentage premium of the offer price over the target price four weeks prior to deal announcement, as the dependent variable. 14 After controlling for firm and deal characteristics, we find a negative and significant relation between the use of FOs on the acquirer side and the deal premium. The use of an acquirer FO leads to a reduction of 4.32% in the deal premium compared to deals without an acquirer FO in mergers (Column 1), and the reduction is 5.7% in tender offers but is not statistically significant (Column 5). This result is consistent with the implication of the 14 Consistent with prior studies, deal premium is significantly lower when the acquirer has a toehold (e.g., Schwert, 1996; Betton and Eckbo, 2000; Goldman and Qian, 2005). The acquiring firm s market-to-book ratio prior to deal announcement is positively related to the premium in mergers (Columns 1 4). Acquirers with high market-to-book ratios ( growth firms) may be overvalued, making the acquirer s stock an attractive method of payment in a merger. Thus, paying the target a higher premium so as to complete the acquisition can still be a good long-term investment strategy (Shleifer and Vishny, 2003; Rhodes-Kropf and Viswanathan, 2004; Rhodes-Kropf, Robinson, and Viswanathan, 2005). transaction improvement hypothesis, as described in Section 3.2 and summarized in Appendix A. In some model specifications not presented in tables, we find a positive and significant relation between the use of target FOs and the deal premium, consistent with the transaction improvement hypothesis for target firms and their shareholders. However, this relation is not stable and is sensitive to which firm and deal variables and whether acquirer FO variables are included in the regressions. By contrast, the negative and significant relation between the use of acquirer FOs and the deal premium is robust to the inclusion of target FO variables and other controls. To control for potential self-selection bias, we perform the two-stage procedure with the treatment equation on the endogenous use of an FO on the acquirer side as described in Eq. (1) above, and the results are shown in Panel B of Table 6. In the first stage we run probit regressions, and the dependent variable is a dummy indicating the use of an FO. For explanatory variables, we use deal size, relative size (of acquirer and target), the target firm s market-to-book ratio, a dummy variable indicating if the deal is friendly, a dummy variable indicating whether it is a related acquisition, a dummy indicating the method of payment (stock), and measures of the acquirer s corporate governance. All of these variables have been shown to have an impact on the use of an FO on the acquirer side (Table 4, Panel A). In the second-stage regressions on deal premiums, we include variables that have been shown to have a significant impact on deal premiums. These include some of the variables used in the treatment equation, e.g., deal size, stock, and friendly dummies, as well as variables that are not expected to influence the use of FOs but are expected to affect deal premiums, e.g., toeholds, competing bids, and the acquirer s market-to-book ratio. As mentioned above, the key variable in each of the four models of this equation that differentiates our two-step procedure from the OLS regression is the hazard rate of acquirer FOs obtained from the first-stage treatment equation. We find that the negative and significant impact of the use of an acquirer FO on the deal premium is robust in the premium regression (Model 1 of Panel B) after controlling for the potential self-selection bias. This result further supports the transaction improvement hypothesis that the use of FOs on the acquirer side has a positive impact on transaction outcomes. As discussed in Section 3.3 and summarized in Panel 3 of Appendix A, we also conduct tests to examine the relative effects of different FO structures and advisors on transaction outcomes. We find that the use of multiple advisors with one or more FOs leads to a further reduction in the deal premium in mergers. Acquirers with two or more advisors and one FO pay a premium that is 11.25% lower than that of acquirers with no FO, while acquirers with multiple advisors and multiple FOs pay 13.03% less (Column 2, Panel A). The deal premium of transactions in which the acquirer has one advisor and one FO is not statistically different from that of transactions with no acquirer FO. Moreover, F-tests comparing the coefficients of the multi-advisor structures show a significant

Table 6 The impact of FOs on deal premiums. Panel A presents OLS regressions of deal premium (% premium of offer price over target market price four weeks prior to deal announcement) on acquirer FO as well as other characteristics of the deal and target. In Models 1 through 4, we use the merger subsample; in Models 5 through 8, we examine the tender offer subsample. Panel B reports the results of two-step regressions of the deal premium and the determinants of acquirer FO following Maddala (1983, pp. 120 122). We include year dummies in all models. All the other control variables are defined in previous tables. P-values are presented in parentheses below the coefficients. Panel A: OLS (1) (2) (3) (4) (5) (6) (7) (8) Merger Merger Merger Merger Tender Tender Tender Tender AFO 4.317 5.704 (0.068) (0.368) AFO 11 2.779 5.895 (0.265) (0.380) AFO 21 11.249 20.369 (0.015) (0.221) AFO 22 13.033 8.767 (0.029) (0.388) AFO_Tier1 5.682 12.110 (0.067) (0.279) AFO_Tier2 3.705 0.985 (0.282) (0.919) AFO_Tier3 3.382 8.315 (0.338) (0.432) AFO_Unaffiliated 0.718 13.697 (0.909) (0.396) AFO_Affiliated1 4.491 16.887 (0.294) (0.234) AFO_Affiliated2 4.596 0.720 (0.069) (0.912) Ln (deal size) 2.111 1.904 1.989 2.112 1.065 0.968 1.185 1.297 (0.006) (0.013) (0.012) (0.006) (0.532) (0.576) (0.495) (0.453) Stock 1.886 1.631 1.851 1.858 7.576 7.038 8.218 7.783 (0.541) (0.600) (0.548) (0.548) (0.418) (0.461) (0.390) (0.408) Toehold 62.819 59.906 62.660 62.499 83.846 83.302 83.705 84.691 (0.000) (0.000) (0.000) (0.000) (0.002) (0.002) (0.002) (0.001) Compete 6.587 6.337 6.608 6.515 7.351 7.525 7.105 7.410 (0.259) (0.272) (0.259) (0.265) (0.275) (0.269) (0.296) (0.267) Friendly 2.843 2.861 2.689 2.938 2.116 2.008 2.295 1.861 (0.641) (0.628) (0.665) (0.631) (0.714) (0.730) (0.694) (0.749) Acq. M/B 0.801 0.827 0.811 0.802 1.046 1.053 1.042 1.027 (0.023) (0.019) (0.021) (0.023) (0.287) (0.287) (0.294) (0.297) Target M/B 0.094 0.142 0.084 0.094 0.774 0.779 0.777 0.772 (0.865) (0.797) (0.879) (0.866) (0.000) (0.000) (0.000) (0.000) Target ROE 2.039 2.060 2.033 2.055 4.504 4.519 4.511 4.489 (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) Target sales growth 0.252 0.305 0.260 0.242 0.127 0.113 0.106 0.141 (0.623) (0.547) (0.613) (0.638) (0.888) (0.901) (0.908) (0.877) D.J. Kisgen et al. / Journal of Financial Economics 91 (2009) 179 207 197 ARTICLE IN PRESS

Table 6 (continued ) 198 Panel A: OLS (1) (2) (3) (4) (5) (6) (7) (8) Merger Merger Merger Merger Tender Tender Tender Tender Target D/E 0.466 0.470 0.468 0.487 0.804 0.817 0.661 0.772 (0.528) (0.518) (0.523) (0.511) (0.652) (0.648) (0.717) (0.667) Constant 53.533 52.887 52.918 53.640 61.340 60.643 62.165 62.637 (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) Observations 915 915 915 915 273 273 273 273 R-squared 0.057 0.061 0.057 0.057 0.152 0.153 0.154 0.156 Panel B: Two-equation treatment procedure In step 1 of each model, we obtain the probit estimates of the treatment equation. From these estimates, the hazard rate for each observation is computed following Maddala (1983, pp. 120 122). The coefficients of acquirer FO variables are further obtained by augmenting the regression equation of step 2 with the hazard rate. Treatment equation is for AFO in Model 1, AFO2 in Model 2, AFO_Tier1 in Model 3, and AFO_Affiliated2 in Model 4. AFO2 is a dummy variable that takes value of 1 if there are two or more advisors on the acquirer side (that is, either AFO 21 ¼ 1 or AFO 22 ¼ 1); 0 otherwise. (1) (2) (3) (4) Step 1: AFO Step 2: Premium Step 1: AFO2 Step 2: Premium Step1: A_Tier1 Step 2: Premium Step 1: AFO_Affiliated2 Step 2: Premium AFO* 10.837 (0.011) AFO2* 31.656 (0.027) AFO_Tier1* 16.075 (0.089) AFO_Affiliated2* 16.215 (0.006) Ln (deal size) 0.058 0.743 0.122 0.305 0.297 0.211 0.098 0.485 (0.048) (0.314) (0.012) (0.722) (0.000) (0.832) (0.005) (0.532) Stock 1.205 3.571 0.3940 0.919 0.546 1.406 1.095 4.310 (0.000) (0.356) (0.225) (0.800) (0.017) (0.704) (0.000) (0.278) Friendly 0.968 0.127 0.640 1.318 0.681 0.422 0.437 0.492 (0.024) (0.991) (0.350) (0.904) (0.152) (0.969) (0.338) (0.964) Relative size 0.052 0.083 0.024 0.030 (0.000) (0.000) (0.000) (0.000) Target M/B 0.002 0.049 0.020 0.012 (0.880) (0.014) (0.274) (0.477) Same industry 0.072 0.366 0.074 0.078 (0.508) (0.021) (0.570) (0.491) Toehold 64.218 63.945 64.969 64.581 (0.014) (0.014) (0.013) (0.013) Compete 10.291 9.717 9.918 10.266 (0.100) (0.119) (0.112) (0.099) D.J. Kisgen et al. / Journal of Financial Economics 91 (2009) 179 207 ARTICLE IN PRESS

D.J. Kisgen et al. / Journal of Financial Economics 91 (2009) 179 207 199 Acq. M/B 0.621 0.746 0.695 0.612 (0.019) (0.004) (0.008) (0.021) A_BoardSize 0.059 0.010 0.062 0.031 (0.000) (0.607) (0.000) (0.024) A_Board_Indep 1.956 0.629 0.920 1.030 (0.000) (0.127) (0.004) (0.000) A_CEO_Chair 0.077 0.097 0.118 0.064 (0.527) (0.581) (0.384) (0.581) A_InsiderShrPct 0.003 0.001 0.001 0.001 (0.396) (0.918) (0.815) (0.967) Hazard Rate 5.700 12.848 6.532 9.011 (0.055) (0.105) (0.352) (0.020) Observations 863 863 863 863 863 863 863 863 Rho 0.170 0.381 0.195 0.267 difference (at the 10% level) from the coefficient when there is one advisor and one FO. We also find that the reputation of the acquirer advisor providing the FO has a positive impact on the deal premium. From Column 3 of Panel A, when a firsttier advisor provides an FO for the acquirer, the deal premium is reduced by 5.68% as compared to deals with no FO. The use of third-tier advisors on the acquirer side, previously shown to be positively related to deal completion (Table 5), has no impact in terms of lowering the deal premium. We do not, however, find that the conflict of interest of the acquirer advisors has an adverse impact on the deal premium. In fact, when the acquirer s FO advisor also receives an additional fee contingent on deal completion (AFO_Affilated2), the deal premium is 4.6% lower than the case with no acquirer FO (Column 4, Panel A). There is no significant difference in premiums paid between deals with no acquirer FO and deals where the FO is provided by either an unaffiliated advisor or an affiliated advisor that receives an additional fee regardless of deal completion (AFO_Affilated1). Finally, F-tests comparing coefficients associated with different rankings (Column 3) and affiliations (Column 4) of acquirer FO advisors show that these coefficients are not significantly different from each other. Given the small number of mergers with the multiadvisor, one FO structure and the multi-advisor, multiopinion structure, along with the fact they both lead to significant reductions in premiums paid compared to deals with no acquirer FO, we combine these two groups and rename the new group AFO2 for purposes of conducting the two-stage estimation. Model 2 of Panel B indicates a negative and significant relation between this AFO2 structure and the deal premium. These results confirm that a multi-advisor/opinion structure is particularly valuable for acquiring firms shareholders in terms of the pricing of the transaction. From Model 3 of Panel B, the positive impact of advisor reputation on the deal premium is also robust in the two-equation procedure, although the statistical significance drops from the OLS model. This result is also consistent with the transaction improvement hypothesis: with their reputation on the line, top-tier FO advisors are willing to certify the quality of the transaction for the acquiring firm only when the price is fair. Finally, the negative relation between the deal premium and the use of affiliated FO advisors that receive a contingent fee is robust to controlling for the selection bias (Model 4). This result suggests that the potential conflict of interest of the acquirer FO advisor does not seem to adversely affect the pricing of the transaction. One possible explanation is that the reputation mechanism motivates an investment bank to exert effort in completing quality transactions on good terms despite the contingency fee. 4.3.3. Announcement-period returns of acquiring firms Panel A of Table 7 presents results from OLS regressions with the three-day ( 1, +1) cumulative abnormal returns of acquirers around the acquisition

200 D.J. Kisgen et al. / Journal of Financial Economics 91 (2009) 179 207 Table 7 The impact of FOs on three-day announcement-period returns of acquirers CAR( 1, +1). Panel A presents OLS regressions of the three-day ( 1, +1) cumulative abnormal returns of acquirers (in %) around the deal announcement date on the use of acquirer FOs as well as other characteristics of the deal and target. In Models 1 through 7, we use the merger subsample; in Model 8, we examine the tender offer subsample. Year dummies are included in all models. All the other explanatory variables are defined in previous tables. P-values are presented in parentheses below the coefficients. Panel A: OLS (1) (2) (3) (4) (5) (6) (7) (8) Merger Merger Merger Merger Merger Merger Merger Tender AFO 2.252 2.326 0.814 0.936 1.276 (0.000) (0.000) (0.235) (0.224) (0.317) AFO 11 2.315 (0.000) AFO 21 2.093 (0.105) AFO 22 1.594 (0.331) AFO_Tier1 1.965 (0.004) AFO_Tier2 1.839 (0.021) AFO_Tier3 2.893 (0.000) AFO_Unaffiliated 0.773 (0.617) AFO_Affiliated1 1.464 (0.133) AFO_Affiliated2 2.616 (0.000) High premium 1.408 0.135 (0.004) (0.814) AFO*High premium 3.305 (0.001) Premium 0.004 (0.486) AFO*Premium 0.033 (0.032) Relative size/100 0.048 0.047 0.045 0.049 0.053 0.054 0.055 0.075 (0.428) (0.434) (0.453) (0.424) (0.455) (0.421) (0.405) (0.185) Ln (deal size) 0.155 0.165 0.208 0.137 0.173 0.197 0.185 0.868 (0.331) (0.306) (0.205) (0.391) (0.286) (0.218) (0.256) (0.003) Stock 1.813 1.813 1.811 1.832 1.985 2.002 1.989 0.451 (0.005) (0.005) (0.005) (0.004) (0.004) (0.003) (0.004) (0.687) Toehold 7.370 7.277 7.122 7.171 5.083 4.640 5.171 7.248 (0.001) (0.001) (0.002) (0.002) (0.063) (0.055) (0.037) (0.222) Friendly 1.785 1.801 1.739 1.753 0.930 0.560 0.858 3.030 (0.448) (0.446) (0.457) (0.461) (0.739) (0.840) (0.754) (0.011) Target M/B 0.146 0.151 0.147 0.143 0.132 0.124 0.134 0.033 (0.121) (0.117) (0.122) (0.130) (0.187) (0.215) (0.178) (0.432) Target ROE 0.347 0.349 0.346 0.340 0.254 0.265 0.270 0.212 (0.015) (0.014) (0.016) (0.013) (0.013) (0.008) (0.005) (0.009) Target sales growth 0.133 0.130 0.126 0.125 0.149 0.138 0.135 0.138 (0.145) (0.151) (0.153) (0.158) (0.111) (0.152) (0.156) (0.456) Target D/E 0.194 0.194 0.183 0.200 0.205 0.237 0.213 0.172 (0.125) (0.124) (0.160) (0.113) (0.097) (0.055) (0.078) (0.643) Constant 0.358 0.408 0.669 0.281 1.943 1.581 1.202 3.599 (0.896) (0.881) (0.807) (0.919) (0.566) (0.643) (0.720) (0.144) Observations 998 998 998 998 935 935 935 295 R-squared 0.080 0.081 0.082 0.083 0.088 0.099 0.092 0.144 Panel B: Two-equation treatment procedure In step 1 of each model, we obtain the probit estimates of the treatment equation. From these estimates, the hazard rate for each observation is computed following Maddala (1983, pp. 120 122). The coefficients of acquirer FO variables are further obtained by augmenting the regression equation of step 2 with the hazard rate. Treatment equation is for AFO in Model 1, AFO 11 in Model 2, AFO_Tier3 in Model 3, and AFO_Affiliated2 in Model 4. (1) (2) (3) (4) Step 1: AFO Step 2: CAR Step 1: AFO1 Step 2: CAR Step1: A_Tier3 Step 2: CAR Step 1: AFO_Affiliated2 Step 2: CAR AFO* 4.798 (0.000) AFO 11 * 5.364 (0.000)

D.J. Kisgen et al. / Journal of Financial Economics 91 (2009) 179 207 201 Table 7 (continued ) Panel B: Two-equation treatment procedure In step 1 of each model, we obtain the probit estimates of the treatment equation. From these estimates, the hazard rate for each observation is computed following Maddala (1983, pp. 120 122). The coefficients of acquirer FO variables are further obtained by augmenting the regression equation of step 2 with the hazard rate. Treatment equation is for AFO in Model 1, AFO 11 in Model 2, AFO_Tier3 in Model 3, and AFO_Affiliated2 in Model 4. (1) (2) (3) (4) Step 1: AFO Step 2: CAR Step 1: AFO1 Step 2: CAR Step1: A_Tier3 Step 2: CAR Step 1: AFO_Affiliated2 Step 2: CAR AFO_Tier3* 9.052 (0.000) AFO_Affiliated2* 6.914 (0.000) Ln (deal size) 0.125 0.086 0.036 0.230 0.254 0.690 0.103 0.012 (0.000) (0.573) (0.279) (0.120) (0.000) (0.000) (0.002) (0.940) Stock 1.232 1.443 1.117 1.465 0.945 1.848 1.051 1.309 (0.000) (0.051) (0.000) (0.051) (0.001) (0.013) (0.000) (0.083) Friendly 0.973 3.017 0.734 2.930 4.942 2.196 0.448 2.862 (0.032) (0.145) (0.116) (0.162) (0.000) (0.303) (0.315) (0.176) Relative size 0.040 0.032 0.032 0.031 (0.000) (0.000) (0.000) (0.000) Target M/B 0.010 0.025 0.003 0.011 (0.565) (0.135) (0.880) (0.484) Same industry 0.030 0.143 0.125 0.071 (0.790) (0.183) (0.347) (0.513) Toehold 5.705 5.099 5.223 5.461 (0.276) (0.332) (0.326) (0.297) Acq. M/B 0.056 0.061 0.055 0.057 (0.295) (0.261) (0.313) (0.289) A_BoardSize 0.057 0.052 0.008 0.033 (0.000) (0.000) (0.655) (0.012) A_Board_Indep 1.785 1.222 0.481 0.900 (0.000) (0.000) (0.149) (0.000) A_CEO_Chair 0.120 0.157 0.222 0.086 (0.303) (0.159) (0.101) (0.438) A_InsiderShrPct 0.003 0.002 0.001 0.001 (0.470) (0.638) (0.856) (0.924) Hazard rate 2.116 2.615 4.245 3.218 (0.000) (0.000) (0.000) (0.000) Observations 949 949 949 949 Rho 0.292 0.356 0.567 0.434 announcement as the dependent variable. 15 We find that mergers with acquirer FOs have worse announcement returns (Column 1; negative but insignificant for tender offers in Column 8) than deals without an acquirer FO. An acquirer FO leads to an announcement return that is approximately 2.3% less than other transactions, and this is statistically significant at the 1% level. This result is 15 The three-day ( 1, +1) cumulative abnormal returns (CARs) around the acquisition announcement are computed using the market model. The estimation period is from 181 trading days to 22 trading days prior to announcement date. Similar to the findings from recent literature on the merger wave in the late 1990s (e.g., Moeller, Schlingemann, and Stulz, 2005), we find that transactions in which the target firm has a higher return on equity prior to the deal announcement (proxy for stock price runup) and transactions financed by the acquirer s stock have lower announcement-period returns, while cash deals, friendly deals (in tender offers), and deals with toeholds fare better. consistent with the implication of the legal protection only hypothesis and contradicts the transaction improvement hypothesis. This result also appears contradictory to the deal premium results, since the reduced deal premium with an acquirer FO should be associated with a better market reaction. We also employ a two-stage procedure to control for the potential self-selection bias in these tests, with the dependent variable in the second stage the acquirer s announcement-period returns. As shown in Panel B of Table 7, the negative and significant impact of the use of FOs on the announcement return is robust after controlling for the potential self-selection bias. The findings from both OLS regressions and the two-stage procedure are consistent with the legal protection only hypothesis. To further explore the seemingly contradictory evidence on announcement returns and deal premiums, we

202 D.J. Kisgen et al. / Journal of Financial Economics 91 (2009) 179 207 Table 8 The impact of FOs on three-day announcement-period returns of acquirers (high premium dummy and more interaction terms included as controls). This table presents OLS regressions of the three-day ( 1, +1) cumulative abnormal returns of acquirers (in %) around the deal announcement date on the use of acquirer FOs as well as other characteristics of the deal and target for the merger subsample. We include year dummies in all models. All the other explanatory variables are defined in previous tables. P-values are presented in parentheses below the coefficients. (1) (2) (3) (4) (5) (6) Merger Merger Merger Merger Merger Merger High premium 0.131 0.136 0.134 (0.820) (0.813) (0.815) AFO*HighPrem 3.232 3.303 3.402 (0.001) (0.001) (0.001) AFO_Tier1 0.632 (0.436) AFO_Tier2 0.308 (0.745) AFO_Tier3 1.534 (0.100) AFO 11 0.832 (0.237) AFO 21 0.960 (0.483) AFO 22 0.337 (0.840) AFO_Unaffiliated 1.736 (0.291) AFO_Affiliated1 0.371 (0.727) AFO_Affiliated2 1.123 (0.122) AFO_Tier1*LowPrem 0.332 (0.687) AFO 22 *LowPrem 1.450 (0.507) AFO_Unaffiliated*LowPrem 3.041 (0.170) Relative size/100 0.051 0.053 0.054 0.075 0.074 0.075 (0.442) (0.424) (0.420) (0.367) (0.368) (0.367) Ln (deal size) 0.247 0.201 0.190 0.271 0.265 0.269 (0.134) (0.216) (0.235) (0.090) (0.100) (0.094) Stock 2.003 2.000 2.017 2.842 2.845 2.859 (0.003) (0.003) (0.003) (0.000) (0.000) (0.000) Toehold 4.261 4.605 4.648 6.076 5.925 6.257 (0.077) (0.058) (0.056) (0.005) (0.004) (0.004) Friendly 0.527 0.586 0.488 0.395 0.414 0.409 (0.849) (0.834) (0.862) (0.878) (0.872) (0.874) Target M/B 0.125 0.127 0.120 0.127 0.134 0.126 (0.213) (0.212) (0.230) (0.200) (0.182) (0.202) Target ROE 0.263 0.267 0.252 0.275 0.278 0.264 (0.009) (0.008) (0.008) (0.006) (0.005) (0.004) Target sales growth 0.130 0.136 0.130 0.120 0.117 0.115 (0.158) (0.155) (0.163) (0.232) (0.240) (0.241) Target D/E 0.226 0.239 0.251 0.183 0.180 0.188 (0.075) (0.053) (0.045) (0.152) (0.159) (0.141) Constant 1.859 1.586 1.630 2.289 2.261 2.242 (0.586) (0.643) (0.636) (0.461) (0.468) (0.471) Observations 935 935 935 935 935 935 R-squared 0.101 0.099 0.103 0.061 0.061 0.063 differentiate transactions with a high and low premium in Columns 5 7 in Panel A. Column 5 shows that lower premiums generally provide better announcement returns, and the impact of an FO on announcement returns independent of any premium impact is negative. Next, we find that the negative relation between the use of FOs and

D.J. Kisgen et al. / Journal of Financial Economics 91 (2009) 179 207 203 acquirer announcement returns is concentrated in transactions with high premiums: the coefficient on the interaction of acquirer FO and high-premium dummies (Column 6) as well as the interaction of acquirer FO and premium (continuous variable; Column 7) is negative and statistically significant. Both the magnitude and statistical significance of the coefficient on the acquirer FO dummy fall sharply once we control for deal premium (Columns 6 and 7). Furthermore, the coefficient on the high-premium dummy variable is no longer negative in these regressions. These results indicate that the market reacts particularly negatively toward transactions in which the acquirer pays a high premium for the target and uses an FO. In addition, univariate comparisons (not shown in tables) reveal that, compared to the low-premium deals, acquirers in the high-premium group are more (less) likely to employ advisors with the lowest (highest) rankings, more (less) likely to use the one advisor, one FO (multi-opinion) structure, and more (less) likely to employ an unaffiliated (affiliated) advisors. As discussed in Section 3.3, we also conduct tests to examine the relative effects of different FO structures and advisors on transaction outcomes. We find that the negative impact of FOs on announcement returns is significant with one advisor and one FO (Column 2 of Panel A) but not with multiple advisors and/or FOs. F-tests cannot reject the null that the impacts are equal, however, perhaps due to the small subsample sizes. While the acquirer announcement returns are lower regardless of the rankings of the FO advisors compared to deals with no FOs, the coefficients on the different dummies in Column 3 of Panel A indicate that the announcement returns are lowest if the acquirer hires an advisor with the lowest ranking. The negative impact of FOs on announcement returns also seems to be driven by the use of an affiliated FO advisor that also receives a contingency fee (Column 4), whereas the coefficients for an unaffiliated advisor and an affiliated advisor receiving a constant (additional) fee are negative but statistically insignificant. Since the same advisor provides both advisory and FO services and receives an additional fee contingent on deal completion that is much larger than the FO fee in these transactions, the potential conflict of interest could significantly reduce the objectivity of the FO, which also explains the negative reaction by the market and investors. Once again, however, the differences in coefficients are not statistically significant. We also examine the relation between FO structures and advisors and the acquirer s announcement returns using the two-stage procedure to control for the potential self-selection bias. Unlike in Panel B of Table 6 (Models 2 4), where the key explanatory variable is the FO structure and advisor types that have the largest impact in reducing the deal premium, we focus on the FO structure and advisor types that have the largest negative impact on the announcement returns in Panel B of Table 7. As shown in Models 2 4, all of the results from OLS regressions in Panel A are robust in the two-stage procedure. We further examine the impact of FO structures and advisors on announcement returns controlling for deal premium, and the results are presented in Table 8. These tests examine the impact of the acquirer structure and quality on acquirer announcement returns conditional on deal premiums. In these tests, the coefficient on the thirdtier FO advisor dummy is the only one that remains statistically significantly negative after controlling for the interaction of the acquirer FO and high-premium dummies (Column 1), and the magnitude of the coefficient on the third-tier FO advisor dummy is larger than those of the first- and second-tier FO advisor dummies. The coefficient for the unaffiliated FO advisor dummy is positive (but insignificant) in Column 3, and the coefficient on the unaffiliated FO advisor dummy interacted with the low-premium dummy is positive in Column 6. These results reinforce earlier evidence that the market reacts negatively to the use of a contingency fee when contracting with the FO advisor and the potential conflict of interest. The coefficient on the multi-advisor/opinion dummy is also positive but insignificant when interacted with the low-premium dummy (Column 5), consistent with our earlier result that this multi-opinion structure is relatively beneficial to acquirer shareholders. Finally, we re-examine announcement returns using a longer announcement window. As argued earlier in the paper, the existence of an FO might not be known immediately at announcement. Furthermore, the benefits of an FO could manifest after the announcement if the FO prevents management from pursuing potentially unfavorable renegotiation. We therefore calculate announcement returns for one month following the announcement (21 trading days) to account for this. These tests yield qualitatively similar results compared to those from the shorter window (Tables 7 and 8), and are not reported to save space. 4.4. Interpretation of results Combining the results on the acquirer s announcement returns with the earlier results on deal completion (Table 5) and deal premium (Table 6), we conclude that the use of FOs has mixed effects for acquiring firms and their shareholders. Neither the simple version of the legal protection only hypothesis nor the simple version of the transaction improvement hypothesis tells the full story of acquirer FOs. Consistent with the transaction improvement hypothesis, FOs for acquirers help reduce deal premiums paid; but consistent with the legal protection only hypothesis, transactions with an acquirer FO are more likely to be completed and to result in greater wealth loss for the acquiring firm s shareholders. The acquirer s announcement-period returns could represent the most comprehensive measure for deal quality, since the market and investors update their beliefs about the acquirer upon observing the announcement and associated information disclosure for the deal. Accordingly, our results from Tables 7 and 8 indicate that the market interprets the use of an FO negatively (on average) for the overall quality of the transaction as compared to deals without an FO, and the market reacts particularly negatively toward transactions in which the acquirer uses an FO and pays high premiums for the

204 D.J. Kisgen et al. / Journal of Financial Economics 91 (2009) 179 207 16 For example, consider if low-quality transactions (upon completion) decrease shareholder value on average by 8%, high-quality transactions increase shareholder value by 4%, the increase in the probability of a transaction being low quality at announcement is 20% given an acquirer FO, and an acquirer FO makes a low-quality transaction 3% more likely to be completed. These figures imply that the relative announcement return for transactions with an acquirer FO is equal to: 0.2 ( 0.08 0.04) +0.03 ( 0.08) ¼ 2.64% (this example assumes, for simplicity, that higher-quality transactions always get completed). target. In our sample, the average size of an acquirer is about six times that of a target, and thus the lower premium (4.3% from Table 6) associated with the use of an FO implies that the positive impact of an FO for the acquiring firm s shareholders (by lowering the cost of the acquisition) is about 0.7%, or about one-third of the negative announcement returns that the FO triggers ( 2.3% in Table 7, Panel A). Overall, the benefits provided by an FO in reduced premiums are dominated by the market s assessment of what the presence of an FO indicates about the deal, and the weight of all of our evidence falls in favor of the legal protection only hypothesis. Our analysis suggests why the market might react negatively to the presence of acquirer FOs. The use of FOs could increase the likelihood of deal completion for questionable transactions and subsequently lead to lower announcement returns, since mergers on average are associated with negative (acquirer) announcement returns. We find that transactions are more likely to be completed after announcement if they have an acquirer FO (see Table 5). Furthermore, while unobservable to us, an FO could also increase the probability that a deal makes it to the announcement stage, particularly for transactions that are more questionable and worse for the acquiring firm s shareholders. The announcement return for a transaction with an FO therefore incorporates both the increased probability that the deal is of lower quality for acquiring firm shareholders, and the increased probability that the transaction will eventually be completed. The relative negative announcement return of 2.3% for transactions with an acquirer FO can be derived given reasonable assumptions for these combined effects. 16 In addition, the FO provider is also part of the M&A advisory group and receives an additional fee in most cases, and hence the potential conflict of interest of the FO advisor can lead to the market s skepticism about the advisor s objectivity and value to the shareholders. To make matters worse, the disclosure in FOs, such as the almost uniformly fair assessment of the deal pricing, the disclaimer that the FO provider does not evaluate nonpricing aspects of the deal, and the legal interpretation that the existence of an FO letter can be sufficient proof of due diligence, could confirm the market s belief that FOs are rubber stamps. Examining subsets of FOs based on the structure and provider of the FO offers additional insights into the mixed evidence. Although some mixed results remain for certain subsamples for example, while the use of a toptier FO advisor significantly reduces the deal premium, the acquirer s announcement return remains lower compared to deals without an FO other more consistent stories emerge. In particular, for some transactions FOs appear to provide value to the acquiring firm s shareholders. For these acquirers, the managements and boards use an FO as part of their due diligence process. These firms in general retain a reputable investment bank and/ or multiple banks to provide the opinion. Further, Table 8 indicates that the incremental impact on the announcement returns is positive (although statistically insignificant) if the deal premium is low and the FO is provided by an advisor that is unaffiliated. Therefore, for certain subsets of transactions, the use of FOs improves the quality of the transactions and benefits shareholders, in that these deals have a significantly lower acquisition cost and/or do not trigger negative announcement-period returns. This evidence provides a glimmer of hope for the long-term viability of the use of FOs. 17 5. Summary and conclusion This paper empirically examines the use of fairness opinions in M&A transactions. It is the first to examine the effects of these opinions, as well as the reputation of issuing advisors, on the completion and performance of transactions. Approximately 80% of deals over the period 1994 2003 have at least one FO on the target side, while 37% of deals have one or more FOs on the acquirer side. The use of an FO is less likely when the deal size is small, when the transaction is hostile, or when it is a cash transaction as opposed to stock. The use of an FO is more likely when the board of either merging firm is smaller and has more outside members. FOs are used to provide legal protection for management and boards in M&A transactions, particularly in those transactions in which the potential legal risk is higher. From the target shareholder perspective, we find little evidence that FOs provide any value beyond this role. On the other hand, we find evidence that FOs on the acquirer side significantly affect transaction outcomes. An acquirer with an FO pays a lower premium for the target. This effect on premiums is larger if the FO provider is of high quality or if multiple FOs are obtained, indicating that these structures are more favorable to acquirer shareholders. On the other hand, the use of an acquirer FO increases the likelihood of deal completion and is 17 In unreported tests, we also examine whether the use and structure of FOs is related to the long-term stock performance of the merged firm. The transaction improvement hypothesis predicts a positive (negative) relation for an acquirer (target) FO, as long as information contained in the FOs is not fully revealed to the market immediately after deal completion. However, this type of analysis, performed in many other studies, is built on the assumption of some form of market inefficiency. We find that merger transactions with no acquirer FO outperform those with the one advisor, one FO (AFO 11 ) structure; mergers with the multi-opinion structure (AFO 22 ) also outperform the AFO 11 group. These results again suggest that this multiple-fo structure is valuable for acquirer shareholders. We also find that mergers with no target FO outperform those with the one advisor, one FO structure (TFO 11 ); mergers in which a top-tier FO advisor is retained by the target underperform those without a target FO. These results support the transaction improvement hypothesis for the target firms.

D.J. Kisgen et al. / Journal of Financial Economics 91 (2009) 179 207 205 associated with lower announcement-period returns as compared to deals without an FO. Both of these effects are more significant in transactions in which the acquirer pays a high premium for the target, suggesting that the market is skeptical about the quality of the FO and the transaction. Based on the relative size difference between the acquirer and the target in our sample, it appears that the benefits associated with an FO in terms of improved pricing are offset by the negative signal that the existence of an FO conveys about the overall merits of the deal. We find that the multi-opinion FO structure on the acquirer side is positively related to deal quality and performance, yet this structure is used infrequently. Among the acquirers that have at least one FO, only 6.6% of them adopt this structure. Furthermore, we also find relatively positive outcomes for transactions in which the advisor that provides the FO is independent of the overall transaction. Instead of choosing these structures, most acquiring firms choose to have one advisor that also provides an FO. This decision could be justified for two reasons: (1) sensitive business information is not revealed to additional outsiders if the advisor also provides the FO, and (2) using the same advisor is less costly in terms of management time and information processing. However, the checks and balance system provided by the multiopinion structure can be crucial in determining the quality of a large and complicated acquisition and in overcoming any potential agency and conflict-of-interest problems. Since the cost of an FO is small compared to the advisory fee, our results suggest that more acquiring firms should consider using additional, preferably unaffiliated, advisors to provide an FO, in addition to the FO provided by the advisors affiliated with the advisory group. Appendix A. Summary of implications of competing hypotheses regarding acquirer fairness opinions The table below summarizes the implications (in terms of the predicted signs of variable effects in regressions) regarding acquirer FOs based on the two competing hypotheses, Legal Protection Only and Transaction Improvement. The first two sets of implications reflect different predictions on the incidence of FOs and the opposite effects of FOs on various deal outcomes as implied by the two hypotheses. The final set of predictions, concerning the structure of FOs and FO advisors, indicates the heterogeneity of the effects of FOs across transactions with at least one acquirer FO. Sections 3.1 3.3 of the text provide detailed descriptions of these implications and predictions. Legal protection only Transaction improvement 1. Use of FO (AFO as dependent variable) Explanatory variable: Size of the target/transaction (+) (+) Friendly deals (+) (+/ ) Stock payment (+) (+/ ) Related (same-industry) acquisition (+/ ) ( ) Strong corporate governance ( ) (+) 2. Impact of FOs on transaction outcomes (AFO as explanatory variable) Dependent variable: Deal completion (+) (+/ ) Deal premium (+) ( ) Announcement return ( ) (+) 3. FO structure and advisor type (conditional on at least one FO) (we examine which FO structures and advisor types are more likely to provide transaction improvement and their impact on deal outcomes; explanatory variables include independent advisors, e.g., AFO_Unaffiliated, more advisors, e.g., AFO 22, or higher-quality advisors, e.g., AFO_Tier1) Dependent variable: Deal completion (N/A) (+/ ) Deal premium (N/A) ( ) Announcement return (N/A) (+) Appendix B. List of variables used in tests 1. AFO (TFO): dummy variable equal to 1 if one or more FO advisors are used, 0 otherwise 2. AFO ij (TFO ij ): dummy variable that takes the value of 1 if the acquiring firm (target firm) has i advisors and j of them conducted an FO. For example, the dummy variable AFO 21 is equal to 1 if the acquiring firm hires two or more financial advisors but only one of them conducts an FO, and AFO 22 is another dummy variable equal to 1 if the acquirer hires two or more financial advisors and two or more of them conduct an FO

206 D.J. Kisgen et al. / Journal of Financial Economics 91 (2009) 179 207 3. AFO_Tier1 (TFO_Tier1): a dummy variable indicating the rank of the acquirer s (target s) FO advisor, 1 if rank is from 1 to 5 (top tier); in the case of multiple FO advisors, the rank of the FO advisor group is based on the highest-ranked FO advisor 4. AFO_Tier2 (TFO_Tier2): a dummy variable indicating the rank of the acquirer s (target s) FO advisor, 1 if rank is from 6 to 20 (second tier); in the case of multiple FO advisors, the rank of the FO group is based on the highest-ranked FO advisor 5. AFO_Tier3 (TFO_Tier3): a dummy variable indicating the rank of the acquirer s (target s ) FO advisor, 1 if rank is 21 or over (third tier); in the case of multiple FO advisors, the rank of the group is based on the highest-ranked FO advisor 6. AFO_Unaffiliated (TFO_Unaffiliated): dummy that equals 1 if the acquirer s (target s) FO advisor is not affiliated with the advisory group; in the case of multiple FO advisors, each of the advisor must be unaffiliated with the advisory group in order for the FO advisors group to be classified as Unaffiliated 7. AFO_Affiliated1 (TFO_Affiliated1): dummy that equals 1 if each of the acquirer s (target s) FO advisors also receives an advisory fee that is not contingent on deal completion 8. AFO_Affiliated2 (TFO_Affiliated2): dummy that equals 1 if the acquirer s (target s) FO advisor also receives an advisory fee that is contingent on deal completion; in the case of multiple FO advisors, as long as one FO advisor receives a contingent fee the FO advisors group is classified as Affiliated2 9. Deal size: the transaction value reported by SDC 10. Relative size: the ratio of acquirer size relative to target size 11. Acquirer size: the market value of equity of the acquirer measured 2 months prior to the acquisition announcement 12. Acq. M/B (Target M/B): the acquirer s (target s) market-to-book ratio, measured as the ratio of the year-end market value of common stock to the book value of equity for the prior fiscal year (COMPUSTAT items 24 25/60) 13. Same_Industry: a dummy variable equal to 1 if the acquirer and target have the same first two digits in their SIC codes 14. Target ROE: the ratio of earnings to average equity for the prior fiscal year (COMPUSTAT items 20/(60+60(t 1)) 15. Target sales growth: the proportional change in sales over the prior fiscal year 16. Target D/E: the ratio of debt to equity of the target for the prior fiscal year (COMPUSTAT items 6/60) 17. Stock: a dummy variable that takes the value of 1 if at least 50% of the transaction is paid by stock, 0 otherwise 18. Friendly: a dummy variable indicating deal attitude: 1 if friendly, and 0 if hostile 19. Tender: a dummy variable indicating that deals are identified as a tender offer by SDC 20. Toehold: the fraction of the target s stock held by the acquirer prior to the merger announcement; deals in which the toehold is larger than 50% are dropped 21. Compete: a dummy variable equal to 1 if there is at least one competing bidder for the same target after the deal announcement and before the deal completion/termination 22. Premium: the percentage difference between the offer price and the target s share price four weeks prior to the announcement date 23. High premium/low premium: dummy variables indicating whether the deal premium of a particular transaction is higher or lower than the sample median 24. A_BoardSize (T_BoardSize): the number of directors on the board of the acquirer (target) prior to acquisition announcement 25. A_Board_Indep (T_Board_Indep): the ratio of the number of non-insider board members to the total number of directors on the board of the acquirer (target) 26. A_CEO_Chair (T_CEO_Chair): a dummy variable equal to 1 if the CEO of the acquirer (target) is the chair of the board of directors 27. A_InsiderShrPct (T_InsiderShrPct): the security ownership of officers and directors as a group (in %) reported in the proxy statement of the acquirer (target). References Allen, L., Jagtiani, J., Peristiani, S., Saunders, A., 2004. The role of bank advisors in mergers and acquisitions. Journal of Money, Credit, and Banking 36, 197 224. Bao, J., Edmans, A., 2006. How should acquirers select investment banks advisors? Unpublished Working Paper, Massachusetts Institute of Technology. Bebchuk, L., Kahan, M., 1989. Fairness in opinions: how fair are they and what can be done about them? Duke Law Journal 1989, 27 53. Betton, S., Eckbo, E., 2000. Toeholds, did jumps, and expected payoffs in takeovers. Review of Financial Studies 13, 841 882. Boone, A., Field, L., Karpoff, J., Raheja, C., 2007. The determinants of board size and composition: an empirical analysis. Journal of Financial Economics 85, 66 101. Bowers, H., 2002. Fairness opinions and the business judgment rule: an empirical investigation of target firms use of fairness opinions. Northwestern University Law Review 96, 567 578. Bowers, H., Latham, W., 2004. Value of fairness opinions in US mergers and acquisitions, 1980 2002. Working Paper, University of Delaware. Bowers, H., Miller, R., 1990. Choice of investment banker and shareholders wealth of firms involved in acquisitions. Financial Management 19, 34 44. Cain, M., 2007. The information content of fairness opinions in negotiated mergers. Working Paper, Purdue University. Chemmanur, T., Fulghieri, P., 1994. Investment bank reputation, information production, and financial intermediation. Journal of Finance 49, 57 79. Coles, J., Lemmon, M., Meschke, F., 2007. Structure models and endogeneity in corporate finance: the link between managerial ownership and corporate performance. Working Paper, Arizona State University. Davidoff, S., 2006. Fairness opinions. American University Law Review 55, 1557 1625. Gac, E., Shaw, B., 1995. Fairness opinions in leveraged buyouts: should investment bankers be directly liable to shareholders? Securities Regulation Law Journal 23, 27 40.

D.J. Kisgen et al. / Journal of Financial Economics 91 (2009) 179 207 207 Gevurtz, F., 2000. Corporation Law. West Group, St. Paul, MN, pp. 284 290. Goldman, E., Qian, J., 2005. Optimal toeholds in takeover contests. Journal of Financial Economics 77, 321 346. Gompers, P., Ishii, J., Metrick, A., 2003. Corporate governance and equity prices. Quarterly Journal of Economics 118, 107 155. Grinstein, Y., Hribar, P., 2004. CEO compensation and incentives: evidence from M&A bonuses. Journal of Financial Economics 73, 119 143. Hermalin, B., Weisbach, M., 2003. Boards of directors as an endogenously determined institution: a survey of the economics literature. Economic Policy Review 9, 7 26. Holmstrom, B., Kaplan, S., 2001. Corporate governance and merger activity in the US: making sense of the 1980s and 1990s. Journal of Economic Perspectives 15, 121 144. Kale, J., Kini, O., Ryan, H., 2003. Financial advisors and shareholder wealth gains in corporate takeovers. Journal of Financial and Quantitative Analysis 38, 475 501. Maddala, G., 1983. Limited-dependent and Qualitative Variables in Econometrics. Cambridge University Press, New York. McConnell, J., Servaes, H., 1990. Additional evidence on equity ownership and corporate value. Journal of Financial Economics 27, 595 612. McLaughlin, R., 1990. Investment-banking contracts in tender offers: an empirical analysis. Journal of Financial Economics 28, 209 232. McLaughlin, R., 1992. Does the form of compensation matter? Investment banker fee contracts in tender offers. Journal of Financial Economics 32, 223 260. Megginson, W., Weiss, K., 1991. Venture capitalist certification in initial public offerings. Journal of Finance 46, 879 903. Moeller, S., Schlingemann, F., Stulz, R., 2005. Wealth destruction on a massive scale? A study of acquiring-firm returns in the recent merger wave. Journal of Finance 60, 757 782. Morck, R., Shleifer, A., Vishny, R., 1988. Management ownership and market valuation: an empirical analysis. Journal of Financial Economics 20, 293 316. Rau, R., 2000. Investment bank market share, contingent fee payment and the performance of acquiring firms. Journal of Financial Economics 56, 293 324. Rhodes-Kropf, M., Viswanathan, S., 2004. Market valuation and merger waves. Journal of Finance 59, 2685 2718. Rhodes-Kropf, M., Robinson, D., Viswanathan, S., 2005. Valuation waves and merger activity: the empirical evidence. Journal of Financial Economics 77, 561 604. Sechler, C., Sweeney, P., 1999. Who says it s a fair deal? Journal of Accountancy 188, 44 51. Servaes, H., Zenner, M., 1996. The role of investment banks in acquisitions. Review of Financial Studies 9, 787 815. Schwert, W., 1996. Markup pricing in mergers and acquisitions. Journal of Financial Economics 41, 153 192. Schwert, W., 2000. Hostility in takeovers: in the eyes of the beholder? Journal of Finance 55, 2599 2640. Shleifer, A., Vishny, R., 2003. Stock market driven acquisitions. Journal of Financial Economics 70, 295 311. Stulz, R., 1988. Managerial control of voting rights: financing policies and the market for corporate control. Journal of Financial Economics 20, 25 54.