Why do acquirers switch financial advisors in mergers and acquisitions?

Transcription

1 Why do acquirers switch financial advisors in mergers and acquisitions? Xiaoxiao Yu 1 and Yeqin Zeng 2 1 University of Texas at Arlington 2 University of Reading September 14, 2017 Abstract Using a sample of 1, 149 acquirers who announced two consecutive mergers and acquisitions within three years between 1987 and 2015, we find that about 54% of the acquirers do not retain any first deal financial advisors in their second deals. This paper examines why acquirers switch their financial advisors and the effect of the switch on the second deal s performance. In contrast to the performance-based explanation, we find little evidence that acquirers switch financial advisors because of the low abnormal announcement returns in their first deals. Several factors related to the financial advisor switch are: investment bank reputations, time between two consecutive deals, and some firm and deal characteristics. Furthermore, we find that acquirers have greater abnormal announcement returns in their second deals if they switch to financial advisors with higher reputations. Keywords: Mergers and Acquisitions; Financial Advisor Switch; Investment Bank Reputation We would like to thank Bin Srinidhi, Christos P. Mavis, Nandu Nagarajan, Ramgopal Venkataraman, Yaqin Hu and seminar participants at University of Texas at Arlington, 2017 EFMA annual conference, 2017 AAA annual conference, and 2017 FMA annual conference for their insightful and constructive comments. The financial support from ICMA Centre is gratefully acknowledged. Yeqin Zeng. y.zeng@icmacentre.ac.uk. Phone: +44(0) ICMA Centre, Henley Business School, University of Reading, Reading, Berkshire RG6 6BA, U.K. Xiaoxiao Yu, xiaoxiao.yu@mavs.uta.edu, Accounting Department, College of Business, University of Texas at Arlington, U.S.

2 1 Introduction Mergers and acquisitions (M&As) are important activities in corporate finance and require complex decision making by firm managers. According to Thomson Reuters SDC Platinum, the most recent merger wave peaked in 2015 when the announced U.S. domestic M&A deal value reached a record high of $2.3 trillion, which was around 13% of GDP. 1 Investment banks usually are hired by firms as M&A deal financial advisors. The advisory services provided by investment banks fall into three categories (McLaughlin, 1990). First, investment banks help M&A acquirers and targets to find potential counter-parties. Second, investment banks make an effort to complete bidding offers, seek higher bids, and negotiate deal terms. Third, investment banks offer advice on bidding strategies, offer prices, decisions of accepting or rejecting offers, and evaluate potential competitive bids. Bao and Edmans (2011) document a significant investment bank fixed effect in the M&A announcement returns, which supports the skilled-advice hypothesis that investment banks help clients identify synergistic targets and negotiate favorable terms. In 2015, U.S. acquirers paid over $849 million in advisory fees to investment banks, as recorded by SDC. Given the dollar amount involved and the importance of such takeovers for acquirers, a misguided acquisition may destroy shareholder value and even lead to CEO turnover. Therefore, the choice of investment banks as financial advisors in corporate acquisitions is of great importance to firm managers. Previous literature on M&A financial advisors focuses on how observable investment bank characteristics affect an acquirer s decision to employ these banks in a single acquisition. Servaes and Zenner (1996) compare acquisitions completed with and without financial advisors and study the determinants of using an investment bank in M&As. Allen et al. (2004) investigate whether commercial banks of acquirers are hired as merger advisors. Both Rau (2000) and Bao and Edmans (2011) find that investment banks market shares of the M&A advisory industry are not related to their client s announcement cumulative abnormal returns (CARs). Contrary to these two studies, Sibilkov and McConnell (2014) 1 The previous peak was $2.1 trillion in

3 find a significantly positive relationship between prior client performance and the likelihood that an investment bank will be chosen as the advisors by potential acquirers in the future. Golubov et al. (2012) document a positive effect of financial advisor reputation, measured by their advisory market shares, on the advised deal performance. In this paper, we study the acquirer s choice of financial advisors from a new perspective. We identify 1, 149 U.S. firms that announced and completed two consecutive domestic deals within a three-year window over the period of About 54% of deal acquirers in our sample did not retain any investment banks hired in the first deal as financial advisors in the second deal. For example, The Priceline Group hired Wells Fargo as its financial advisor for a $1.8 billion acquisition of KAYAK Software Corp. in After two years, The Priceline Group announced a $2.6 billion acquisition of OpenTable Inc., but the deal advisor was switched to Goldman Sachs. When an investment bank provides M&A advice to an acquirer, the investment bank is likely to invest in collecting firm-specific information, for example, acquirer s business operation, management team characteristics, intangible assets, and any inside information that may affect firm future growth opportunities. If the information is useful in future advisory business with the same acquirer, the investment bank will have economies of scale through repeated deal advisory relationships. The investment in firm-specific information is defined by Williamson (1979) as a durable transactor-specific asset that is required for the transaction but not transferable to transactions involving different parties. From the acquirers perspective, they may also invest in the business relationship with their investment banks. In two consecutive M&As, it would be less expensive for the acquirer to employ the same investment bank and the investment bank may do a better job than others. The investment in a transaction-specific asset creates a lock in effect by making it costly for an acquirer to switch financial advisors. Previous M&A financial advisor studies usually focus on the choice of financial advisors in a single deal. In our paper, we investigate the individual decisions acquirers make to switch financial advisors between two consecutive M&As. The first research question 2

4 examined in this paper is: despite the transaction-specific asset, why do so many acquirers change their financial advisor in two consecutive M&As? We posit several reasons for the switch of investment banks as M&A financial advisors, and empirically test whether they are associated with the switch decision. The second research question which we are interested in is: does the switch of financial advisors improve the second M&A deal performance? The major objective of this paper is to improve our understanding of the economics underlying an acquirer s choice of financial intermediation services. The general transaction-specific asset in repeated contractual relations has been ascribed to relationship-specific capital by James (1992) and Burch et al. (2005). In a standard trade-off economic model, an acquirer chooses to switch investment banks if and only if the benefits of switching exceeds its costs. We study four main switch incentives in this paper. First, if an acquiring firm s managers attribute the first deal s poor performance to the deal financial advisors, then the benefit of switching is the improvement in the second deal announcement return and it may be above the replacement cost of relationship-specific capital. Second, if the benefits of hiring investment banks with high reputation outweighs the switch costs, then an acquirer may trade up to more prestigious financial advisors. This has been defined as the graduation effect in Krigman et al. (2001). Third, the relationship specific capital may depreciate over time. The more time it takes between two consecutive deals, the less valuable is relationship-specific capital, and the more likely the acquirer to switch financial advisors. Fourth, investment banks may have expertise in specific types of deals. If two consecutive deals differ greatly in terms of several important deal and firm characteristics, then an acquirer may choose appropriate investment banks with respect to these different characteristics. Our main findings are as follows. First, contrary to Sibilkov and McConnell (2014), who find that prior client performance measured by announcement returns is a significant determinant of the likelihood that an investment bank will be chosen as deal financial advisors by future acquirers, we find little evidence that acquirers switch to new financial advisors because of poor first deal performance. Both univariate and multivariate tests 3

5 show that acquirer CARs around the first deal announcement do not significantly change the likelihood that the first deal financial advisors are switched in the next deal. 2 Second, acquirers tend to trade up to more prestigious financial advisors in two consecutive deals. Following Rau (2000) and Golubov et al. (2012), we measure investment bank reputation by their market shares of the M&A financial advisory industry over a five-year window before the deal announcement. We find that the increase of investment bank reputation between the two deals is positively related to the probability of financial advisor switch. We obtain the similar results when we measure investment bank reputation as advisors being top-tier investment bank (Golubov et al., 2012). Third, acquirers are more likely to switch if the time between two consecutive deals is longer. This finding is consistent with the view that transaction-specific assets depreciate over time. The longer the time interval is between two M&As, the less costly it will be for the acquirers to switch their financial advisors. Fourth, acquirers tend to switch to new financial advisors when the second deals are different from the first ones in terms of relative deal size, tender offer, and payment methods. Furthermore, the changes in firm characteristics between two deals such as target industry, target public status, and acquirer leverage can also induce acquirers to change their prior financial advisors in the second deal. Finally, we find that acquirer CARs in the second deal will be improved if they switch to investment banks with a higher reputation. Neither the switching decision nor the reputation of investment banks in the second deal has a significantly positive effect on acquirer CARs by itself. Few papers in the literature focus on financial advisor switches between two consecutive M&As. To be best of our knowledge, Francis et al. (2014) is the only study that investigates the switch of financial advisors. There are three main differences between our study and Francis et al. (2014). First, Francis et al. (2014) mix the choice of financial advisors between equity issues and M&As, while we believe that equity underwriters provide 2 Sibilkov and McConnell (2014) define prior client performance as value-weighted or equal-weighted clients deal announcement returns in the past one-year and three-year periods. Instead of focusing on an investment bank s average prior client performance, we define the prior client performance as the performance of the first deal right before the second deal. These two deals are consecutive and announced by the same acquirer. 4

6 different services from the ones offered by financial advisors in M&As. 3 Second, Francis et al. (2014) find that previous deal performance is positively related to the likelihood of retaining financial advisors. Our findings suggest that investment bank reputation along with other deal and firm characteristics, not prior deal performance, are correlated with financial advisor switches. Third, Francis et al. (2014) find that acquirers have better second deal performance if they retain their previous financial advisors. However, we find that an advisor switch itself does not improve the second deal performance. The second deal performance is improved only when switching firms trade up to investment banks with a higher reputation. Besides the studies on the choice of financial advisors in M&As, our paper is related to another strand of the literature that examines the underwriter s switch decision in equity issues. Firms regularly use investment banks for underwriting new security issues to the public. James (1992) shows that the longer the time between an initial public offering (IPO) and the follow-on offering, the more likely a firm is to switch underwriters as the value of its firm-specific information degrades. Krigman et al. (2001) provide evidence that investment bank reputation and analyst coverage, but not prior underwriter performance, are the determinants of underwriter switch. Burch et al. (2005) find that loyalty to an underwriting bank is associated with lower (higher) fees for common stock (debt) offers. Our paper contributes to the literature on the choice of M&A financial advisors by documenting the possible factors explaining why acquirers switch their financial advisors hired in the prior deals. Our study also adds to literature examining the choice of financial advisors and M&A deal performance. Rau (2000), Bao and Edmans (2011), and Sibilkov and McConnell (2014) have examined the relationship between M&A deal performance and the characteristics of financial advisors, such as investment reputation, prior client 3 Equity underwriters evaluate the value of a firm, confirm whether the firm qualifies to be publicly traded, purchase the firm shares at a discount price, and then sell the shares at the market price to investors. M&A buy-side financial advisors concentrate on valuing the target and determining a competitive bid price. The advisors may also prepare and coordinate official deal documents, assess the proposed deal from strategic and financial perspectives, recommend the method of payment, scout rival bidders, help the acquirer to market the deal to the target shareholders, obtain the information on the market s reaction to the deal, and participate in deal term negotiations. 5

7 performance, advisory fees, etc. None of these papers, however, have examined the consequences of financial advisor switch. Golubov et al. (2012) find that the financial advisor s reputation measured by investment bank market share is positively related to the deal performance. We further show that when an acquirer trades up its financial advisors, its deal performance will be improved. The rest of the paper is organized into five sections. Section 2 reviews the institutional background for advisor choice and motivates our hypotheses. Section 3 describes the data and discusses the definition of advisor switch indicator variables and advisor reputation change indicator variables. Section 4 presents empirical evidence on the determinants of an advisor switch. The robustness tests are discussed in section 5. Section 6 provides a summary of the results and major implications. 2 Hypotheses and empirical predictions In this section, we propose proxies for five incentives that may be associated with an acquirer s financial advisor switch decision: prior deal performance, investment bank reputation, time between two consecutive deals, differences in the two deals, and differences between the two target firms. 2.1 Prior performance hypothesis In many cases, a firm hires an investment bank to explore strategic opportunities for maximizing shareholders value or expanding its products and services. Investment banks, serving as M&A financial advisors, may identify potential targets and propose deals with positive synergies to acquiring firms. Acquirers also hire financial advisors to assist them in evaluating targets and obtaining better deal terms than they would have obtained without financial advisors. The existence of M&A financial advisors reduces the liability risk of acquirer managers and directors, because they may attribute the poor deal performance to financial advisors rather than themselves due to self-attribution bias. 6

8 However, the actual role of investment banks in M&As is controversial, because the advisory fees they receive may be contingent on the deal completion and there exists a principal agent problem between the banks and their client firms. Previous literature studying the effect of financial advisors on M&A deal performance provides mixed evidence. Rau (2000) as well as Bao and Edmans (2011) report that the shares of investment banks in the corporate acquisition advisory market are unrelated to the value created for their clients in their clients prior acquisition attempts. Rau (2000) finds that an advisor s reputation has a positive effect on the likelihood of deal completion but not on the stock price of its client. One possibility is that when acquirers choose financial advisors in M&As, they do not take the value created by the advisors in their prior takeover attempts into account. Contrary to these studies, Sibilkov and McConnell (2014) find that prior client performance is a significant determinant of the likelihood that an investment bank will be chosen as the advisor by future acquirers and of the changes in investment banks shares of the advisory business over time. 4 Given the large M&A transaction value and the importance of such takeovers to acquirers, a natural presumption is that a value-maximizing acquirer will choose its advisor based on the advisor s demonstrated performance in M&As. When an acquirer makes two consecutive M&As, the first deal performance may be a good proxy for the advisor s quality. If the first deal performance is poor, the acquirer may choose not to retain its first deal financial advisors in the second deal. Therefore, we form the following hypothesis: Hypothesis (H1): Acquirers are more likely to change financial advisors in the second M&A if the performance of the first deal is worse. 4 One concern about the findings in Sibilkov and McConnell (2014) is that if acquirers choose advisors based on their prior client performance, clients would always choose the advisors who have created the greatest value for them in the previous period and the advisory market would quickly devolve into one dominated by a single best advisor. This is inconsistent with the actually observed multi-participant market for advisory services. 7

9 2.2 Investment bank reputation hypothesis Previous studies on the choice of investment banks have specifically focused on the reputation of financial advisors. McLaughlin (1990) argues that reputation-building concerns bring investment banks an incentive to improve their reputation so that they will complete deals at any cost and protect the interests of their clients. Kale et al. (2003) examine the relative reputation between acquirer and target financial advisors and find that both the deal synergies and the acquirer share of total takeover wealth gains are positively associated with the reputation difference between acquirer and target financial advisors. Golubov et al. (2012) document that better financial advisor reputation is positively associated with the deal performance. However, Servaes and Zenner (1996) investigate an acquirer s choice between external financial advisors and in-house expertise but find no relation between advisor reputation and bidder wealth. The quality of investment banks is a unobservable firm characteristic. Yet high advisor quality may be signaled by an investment bank s market share of the M&A advisory industry, which is a common measure of investment bank reputation in the previous studies (Rau, 2000). An acquirer may select an investment bank with a higher reputation in the hope of achieving better deal performance. We posit that acquirers naturally flock to financial advisors with higher reputation: Hypothesis (H2): Acquirers tend to hire a financial advisor with a better reputation in the second M&A. 2.3 Transaction-specific asset hypothesis Williamson (1979) defines the durable transaction-specific asset as an asset that is required for the transaction but not marketable or transferable to transactions involving different parties. 5 During repeated transactions, both suppliers and customers may invest 5 According to James (1992), training new employees firm-specific skills in labor markets and firmspecific information associated with establishing a credit relationship with external creditors are both the examples of transaction-specific assets. 8

10 in durable transaction or relationship-specific assets because the average cost of durable transaction-specific assets per transaction is negatively related to the number of transactions. However, the investment in transaction-specific assets makes it costly for customers to switch suppliers during repeated transactions. This kind of lock in effect in the optimal contract pricing theory has been extensively analyzed in the previous literature (e.g., Klein et al., 1978; Farrell and Shapiro, 1989). The nature of financial advisor services in M&As suggests that both acquirers and financial advisors may invest in durable, transaction-specific assets if they expect that the advisory service will resume in the future M&A deals. In particular, in the process of providing advice on the deal, the investment bank obtains information concerning the acquiring firm s operations and management team that would be useful in advising subsequent deals. The investment in the transaction-specific assets may reduce the transaction costs in consecutive deals. To the extent that the value of transaction-specific assets depreciates over time, the longer the interval between transactions, the less costly it will be to switch advisors. The above discussion leads to the following predication: Hypothesis (H3): Acquirers have a higher probability of switching financial advisors if the time interval between two consecutive M&A deals is longer. 2.4 Deal difference hypothesis Hayward (2003) posits that acquirers are more likely to hire investment banks on stock-financed acquisitions when they have previously used these banks, because investment banks may have specialized expertise to help their clients with certain types of deals, such as a stock-financed acquisition. He provides empirical evidence that when an acquirer hires investment banks in a cash-financed acquisition, it will be less likely for the acquirer to use the same investment banks in the subsequent stock-financed acquisition. Therefore, firms may switch financial advisors if the payment methods in two consecutive deals are different. We state this hypothesis below: 9

11 Hypothesis (H4a): Acquirers tend to switch financial advisors if the method of payment in the first deal is different from the one in the second deal. In addition, transaction costs may affect an acquirer s decision to switch financial advisors. One important component of the M&A transaction costs is the financial advisor fees paid by acquirers. M&A advisory fees are usually 1% of the deal transaction value, which may increase as the transaction value decreases (Kosnik and Shapiro, 1997). The lucrative M&A advisory fees offer investment banks a strong incentive to pitch M&A ideas to their current or prospective clients, often pushing them into unnecessary deals of dubious value (Eccles and Crane, 1988). However, if a financial advisor charges high advisory fees in the first deal, the acquirer may not stay with the same advisor in future deals. It is worth noting that 80% of advisory fees are contingent on deal completion (Rau, 2000). Our sample consists of only completed deals, which avoids the comparison of financial advisor fees between completed deals and withdrawn deals. Our next hypothesis is summarized as: Hypothesis (H4b): Acquirers tend to switch financial advisors if the ratio of advisory fees to deal transaction value is higher in the first deal. In our empirical analyses, we focus on the acquirers choices of financial advisors between two consecutive deals. Buy-side advisors may help acquirers to evaluate the deal and determine a competitive bid price. They usually prepare and coordinate documentation, value the target, assess the proposed acquisition from strategic and financial perspectives, recommend how to finance the acquisition, scout rival bidders, help the acquirer market the acquisition to the target s shareholders, obtain feedback from stock market participants, and may participate in negotiations with the target or its representatives. The buy-side advisors may also recommend an offer price and other deal terms, estimate a final price that includes fees and expenses related to the merger, recommend a method of payment, and suggest negotiating strategies (Fleuriet, 2008). All these roles may require that investment banks have specialized expertise in a certain type of deal. If two consecutive 10

12 deals differ in terms of some important deal characteristics, then the acquirer may choose different financial advisors who have deal-specific specialized expertise. We predict that: Hypothesis (H4c): Acquirers tend to switch financial advisors if the characteristics of the first deal are different from those of the second deal, such as tender offer, toehold, competition, and hostile deal. 2.5 Firm difference hypothesis Financial advisors may accumulate industry-specific expertise through repeatedly advising M&As with targets in a certain industry, which enables them to better assess target firm value and deal synergies, execute complex deals, and reduce transaction costs. When choosing advisors among all candidate investment banks, acquirers may attach importance to an investment bank s expertise in industries that are of interest to them (Chang et al., 2016). It is possible that acquirers may switch their financial advisors when the second deal targets are different from the first ones. For example, if two targets in a deal pair are in the manufacturing and business equipment industries, the acquirer may hire two investment banks who have expertise in the corresponding industries: Hypothesis (H5a): Acquirers tend to switch financial advisors if the target firm characteristics in the first deal are different from those in the second deal. Investment banks provide technical and tactical assistance to acquirers throughout the takeover process by evaluating acquirer firm characteristics (Bodnaruk et al., 2009). Acquirers also invest in the cooperative relationship with their deal advisors through repeated deals. The costs of establishing the relationship between acquirers and financial advisors may be taken as transaction-specific assets. The value of transaction-specific assets is negatively related to the difference of acquirer characteristics between two consecutive deals. Therefore, acquirers may switch their financial advisors because their own financial condition, growth opportunities, and risk premium at the announcement of the second deal 11

13 are different from those at the announcement of the first deal. Our last hypothesis is stated as: Hypothesis (H5b): Acquirers tend to switch financial advisors if their own firm characteristics in the second deal are different from those in the first deal. 3 Data and variable construction In this section, we discuss how our M&A sample is constructed. We also illustrate the definitions of financial advisor switch variables and investment bank reputation change variables. 3.1 Sample selection We start with all U.S. domestic M&As with announcement dates between 1984 and 2015 from the Thomson Reuters Securities Data Company (SDC) Platinum Mergers and Acquisitions database. 6 We require acquirers to be public firms and targets to be public, private, or subsidiary firms. Deals without disclosed transaction value and small transactions with deal value less than $1 million are excluded from our sample. In line with previous M&A studies, the percentage of target shares acquired by acquirers must be higher than 50%. Applying the standard filters used in the literature, we then exclude all transactions that are labeled as a minority stake purchase, acquisition of remaining interest, privatization, repurchase, exchange offer, self-tender, recapitalization, or spinoff. Because investment banks may have a closer connection with financial firms than others, we exclude deals with acquirers or targets in the financial industry. We further limit our sample to deals in which the acquirer has daily stock return data from the Center for Research in Security Prices (CRSP) and the annual accounting data from Compustat for at least one year prior to the deal announcement. 6 The sample begins in 1984 because the information in the SDC database is less reliable before this date (Chen et al., 2007). 12

14 Because we examine the switch of acquirer financial advisors between two consecutive deals in this paper, the following three sample selection criteria are applied. First, we only keep completed deals in our sample so that both consecutive deals have the same deal status. The financial advisor switch due to the deal status is not the focus of our paper. We also mitigate the confounding effect of mixed deal status on an acquirer s decision to switch financial advisors. Second, we only keep deals announced by the same acquirer within three years. 7 The three-year window not only leaves us a large sample of deal pairs but also excludes deal pairs with a very long time period between their announcement dates. Third, deals without acquirer financial advisor information are also excluded from our sample. Our final sample includes 1,149 paired deals. Panel A of Table 1 presents the distribution of our sample M&As by deal announcement year. Consistent with the merger wave literature (e.g., Harford, 2005; Duchin and Schmidt, 2013), the frequencies of first deals, as well as second deals, peaked in the late 1990s and dropped in the early 2000s. 8.62% of the first deals and 9.02% of the second deal were announced in The deal frequencies also significantly dropped following the 2007 financial crisis. The pattern of the first deals and second deals across years are similar to each other over our sample period. We also report the frequency and percentage of the switchers in the second deal by year. We define switchers as acquirers that do not retain any first deal financial advisors in their second deals. Acquirers of 617 deal pairs switch their financial advisors in our sample, about 53.7 % of our sample deal pairs. We also report the frequency and percentage of switchers in the second deal by year. The last column of Panel A shows that acquirers have switched to new financial advisors more and more frequently over our sample period. This increasing trend peaked at 66.7% in Panel B of Table 1 presents the distribution of acquirers and targets in our sample by industry. We assign deals into Fama French 10 industries (Fama and French, 1997) by their SIC codes. Business Equipment is the industry that includes the most number of 7 We identify a year as 360 days. Our results are qualitatively similar for the samples of deals announced within 2 years and 1 year. 13

15 deals for both acquirers and targets. We observe a similar industry distribution pattern not only in acquirers and targets but also in first deals and second deals. Panel A and B of Table 1 also demonstrate that our sample includes deals from a wide range of time period and firms from diversified industries. 3.2 IB switch variable construction Next, we describe the definitions of financial advisor switch variables in our empirical analyses. 145 first deal acquirers and 153 second deal acquirers in our sample employ multiple financial advisors. The maximum numbers of financial advisors used are 5 in both the first and second deals. To thoroughly measure all the financial advisor switch scenarios, we adopt the following four definitions of advisor switch : 8 1. ALLIB, a binary variable that is equal to 1 if none of the financial advisors hired in the first deal is retained as the financial advisors in the second deal, and 0 otherwise. For example, ALLIB is equal 1 for a deal pair in which the first deals financial advisors are A, B, and C; the second deal s financial advisors are E and F. 2. HALF IB, a binary variable that is equal to 1 if more than half of the first deal s financial advisors are changed in the second deal, and 0 otherwise. For example, HALF IB is equal to 1 for a deal pair in which the first deal s financial advisors are A, B, and C; the second deal s financial advisors are B, C, D, and E ANY IB, a binary variable that is equal to 1 if the first deal and the second deal do not have exactly the same financial advisors, and 0 otherwise. For example, ANY IB is equal 1 for a deal pair in which the first deal s financial advisors are A, B, and C; the second deal s financial advisors are A and B. 8 We only consider the deal pairs in which financial advisors are hired by acquirers in both deals. We exclude deal pairs in which financial advisors are hired by acquirers in the first deals, but no financial advisors are hired in the second deals. 9 In this example, the total number of changes is three because the acquirer no longer hires A but adds D and E in the second deal. There are three financial advisors in the first deal. 1 > 1 2. number of advisor changes number of first deal advisors = 14

16 4. LEADIB, a binary variable that is equal to 1 if the lead financial advisors hired in the first deal is no longer the lead financial advisors in the second deal, and 0 otherwise. This definition follows the prior literature on the underwriter switch in IPOs and seasoned equity offerings (SEOs) (Krigman et al., 2001). For example, LEADIB is equal to 1 for a deal pair in which the first deal s financial advisors are A, B, and C; the second deal s financial advisors are B, A, and C; A is the lead financial advisor in the first deal; and B is the lead advisor in the second deal. Only ALLIB indicates that all financial advisors in the first deal are no longer used at all in the second deal. For the other three switch definitions, an acquirer might hire even more advisors in the second deal than in the first deal. SDC may report multiple codes for the same investment bank, so we manually check these codes and combine them into a single one if they refer to the same bank. On the other hand, there have been significant M&A activities in the investment banking industry during our sample period from 1987 to To account for this, we utilize the data provided in Corwin and Schultz (2005), Ljungqvist et al. (2006), and Chang et al. (2016), and combine them with those reported by SDC Platinum and other financial news sources. The effective dates of bank mergers are obtained from Corwin and Schultz (2005), supplemented by the other financial news sources. Following Krigman et al. (2001), we do not exclude deals advised by merged or acquired investment banks, because we are interested in the decision to change financial advisors. For example, if the first deal advisor A is acquired by B between the two deal announcements and B is appointed as the financial advisor in the second deal, we do not classify this choice as a switch decision. Panel A of Table 2 reports the summary statistics of our four switch variables. Among 1, 149 pairs of consecutive deals, 53.70% of acquirers switch their financial advisors according to the definition of ALLIB; 64.32% of acquirers switch their financial advisors according to the definition of HALF IB; 64.66% of acquirers switch their financial advisors according to the definition of ANY IB; and 57.96% of acquirers switch their financial advisors according to the definition of LEADIB. All the four definitions of advisor switch 15

17 suggest that more than 50% of deal acquirers use different financial advisors between two consecutive M&As. In the rest of our paper, our primary objective is to investigate the reasons that acquirers switch financial advisors. We control for various deal and firm characteristics in our multivariate analyses. The summary statistics of these control variables are presented in Panel B of Table 2. The detailed definitions of these variables are described in Appendix A. We report the number of observations, total sample mean, non-switcher sample mean, and switcher sample mean, respectively. The last two columns present the statistics of mean difference tests (tstatistics) and median difference tests (z-statistics) between the non-switcher and switcher samples. The summary statistics show that our M&A sample is similar to those used in previous studies of U.S. M&As. For the first deal, switchers have significantly lower transaction value, less toehold, and larger acquirer firm size than non-switchers. For the second deal, switchers have significantly higher transaction value, a higher possibility of tender offers, lower possibility of hostile offers, higher possibility of public targets, less completion time, and larger acquirer firm size than non-switchers. 3.3 IB reputation change variable construction Investment bank reputation is an important determinant of acquirers financial advisor choices. Because more than one investment bank may be employed as the financial advisor in M&As, we construct four investment bank reputation change indicator variables in our empirical analyses. We follow the M&A financial advisor literature (e.g., Rau, 2000; Golubov et al., 2012) and use investment banks market share of the M&A advisory industry to measure investment bank reputation. The market share is defined as the transaction value allocated to each advisor divided by the total transaction value over the previous five (three) years before the deal announcement. To begin with, we define the following four proxies for investment bank reputation: 1. Average reputation (IBRP T AV G): The average market share of investment banks 16

18 hired by acquirers. 2. Maximum reputation (IBRP T M AX): The maximum market share of investment banks hired by acquirers. 3. Summation reputation (IBRP T SUM): The total market share of investment banks hired by acquirers. 4. Lead reputation (IBRP T LEAD): The market share of lead investment banks hired by acquirers. We then compare the value of these reputation proxy variables between two consecutive deals. If the value of a reputation proxy variable in the second deal is greater than its value in the first deal, we assign 1 as the value of the corresponding reputation change indicator variable, and 0 otherwise. 4 Empirical results 4.1 Why do acquirers switch financial advisors? In this section, we first present the univariate comparisons of deal pairs in which acquirers switch financial advisors and those in which acquirers do not. Then we apply multivariate probit estimations corroborating the variables that are statistically significant in the univariate tests Univariate tests We split the entire M&A deal pair sample into two sub-samples based on whether an acquirer switches all financial advisors between two consecutive deals. Non-switchers are deal pairs in which ALLIB is equal to 0 and switchers are deal pairs in which ALLIB is equal to 1. Table 3 presents the summary statistics for the financial advisor switch associated factors between these two sub-samples. The last two columns report the mean 17

19 and median differences in these factors between the non-switch deal pairs and switch deal pairs. To measure the first deal performance, we follow Brown and Warner (1985) and calculate cumulative abnormal returns (CARs) at the deal announcement. The benchmark is the CRSP value-weighted index return over the pre-announcement window [ 300, 46], where event day 0 is the deal announcement date. We measure acquirer CARs over five different event windows: [ 1, +1], [ 2, +2], [ 3, +3], [ 4, +4], [ 20, +5]. The difference between the first deal CARs of non-switchers and those of switchers is not statistically significant at the 10% level. Previous M&A studies also use post-acquisition long term operating performance and post-acquisition buy and hold abnormal stock returns as the alternative deal performance measures. In our sample of 1, 149 deal pairs, the deal effective dates of 215 first deals are later than the deal announcement dates of the corresponding second deals. Furthermore, the time difference between the effective dates of 563 first deals and the announcement dates of the corresponding second deals is less than one year. Therefore, we only use CARs as the measure of prior performance. To test the relationship between investment bank reputation and financial advisor switch, we compare the reputation of financial advisors between two consecutive deals. We assign 1 to the reputation change indicator variables if the reputation of financial advisors in the second deal is higher than those in the first deal, and 0 otherwise. The univariate tests suggest that the increase in investment bank reputation is more likely to be observed in the switcher sample than in the non-switcher sample. The mean and median difference tests are statistically significant at the 1% level. Our results support the view that acquirers choose investment banks with higher reputation when they switch financial advisors between two consecutive M&As. To measure the depreciation of transaction-specific assets over time, we use the number of calendar days between the two consecutive M&As. The value of transaction-specific assets invested by acquirers and financial advisors is negatively related to the time between two consecutive deals. The mean (median) number of calendar days between two deals is 18

20 281.7 (196) in the non-switcher sample, while the mean (median) number of calendar days between two deals is (326) in the switcher sample. The mean (median) difference between the switcher and non-switcher sample is (130) days, which is statistically significant at the 1% level. Finally, we compare the change in the deal and firm characteristics between the switcher and non-switcher sample. The difference tests are statistically significant for the following variables: the change in tender offer, the change in method of payment, the change in toehold, the change in target industry, the change in public target, and the change in acquirer leverage. The univariate tests also suggest that financial advisor switchers are not statistically different from non-switchers with respect to the change in acquirer advisor fee, the change in hostile deal, the change in deal competition, the change in completion time, and the change in relative size Multivariate tests We check the robustness of our findings on the factors associated with financial advisor switch decisions in probit and logit regressions by controlling for year fixed effects. We apply all four definitions of financial advisor switch as the dependent variables in these regressions. The independent variables of interest are the factors associated with financial advisor switch decisions. Hypothesis 1 predicts that the first deal performance is negatively related to the probability of switching financial advisors in the second deal. Table 4 presents the regression results of both probit and logit models in which the dependent variable is equal to 1 for acquirers switching their first deal financial advisors, and 0 otherwise. We use four switch indicator variables: ALLIB, HALF IB, AN Y IB, and LEADIB. The dependent variables are acquirer first deal CARs with five different event windows: [ 1, 1], [ 2, 2], [ 3, 3], [ 5, 5], [ 20, 5]. In all forty regressions reported in Table 4, we control for the year fixed effects. All coefficients of first deal CARs are not statistically significant at the 10% level and the signs of these coefficients are mixed. Table 4 indicates that acquirer CARs in 19

21 the first deal are not significantly different between the switcher and non-switcher samples. Therefore, the first deal performance, at least as measured by acquirer announcement abnormal returns, does not contribute to the financial advisor switch decisions. We do not find any evidence supporting Hypothesis 1. Hypothesis 2 relates to the reputation of financial advisors. We follow the previous M&A literature and measure investment banks reputation as their market shares of the M&A advisory industry. We predict that the increase of financial advisor reputation from the first deal to the second deal is positively related to the probability of financial advisor switch. Table 5 reports the regression results of advisor switch variables on the change in financial advisor reputation from the first deal to the second deal. Both probit and logit regressions are estimated. The dependent variable is equal to 1 for acquirers that change their first deal financial advisors, and 0 otherwise. As in Table 4, we adopt four different definitions of advisor switch. Because some of our sample acquirers use multiple financial advisors, we measure the change in investment bank reputation from the first deal to the second deal by ten different definitions. Year fixed effects are controlled for in all regressions. In Panel A of Table 5, the investment bank reputation change indicator variables are defined according to the investment bank market shares over five years before the deal announcement. The results in Panel A show that the coefficients of all ten investment bank reputation measurements are positive and statistically significant at the 1% level. In Panel B of Table 5, we measure the investment bank reputation change indicator variables according to the investment bank market share over three years before the deal announcement. The results reported in Panel B are qualitatively similar to those reported in Panel A. We further check if our results remain robust after controlling for the reputation of the financial advisors in the first deal. In untabulated tests, we add the reputation of the financial advisors in the first deal as control variables in all regressions reported in Table 5. All the coefficients of reputation change indicator variables remain positive and statistically significant, suggesting that the reputation incentive of switching financial advisors 20

22 does not depend on the reputation level of first deal advisors. Overall, our test results show that acquirers switch their financial advisors used in the first deals to obtain the services of financial advisors with a higher reputation in the second deals. Hypothesis 2 that investment bank reputation is associated with financial advisor switch is supported. Hypothesis 3 states that the transaction-specific assets invested by acquirers and investment banks depreciate over time, so acquirers are more likely to switch financial advisors when the time interval between two consecutive deals is longer. Table 6 presents the probit and logit regression results of investment bank switch variables on the number of days between two consecutive deals announcement dates. The regression models are similar to those reported in Table 4. The results in Table 6 show that the coefficients of the days between two deals are all positive and statistically significant at the 1% level. When the time interval between two M&As is longer, an acquirer has a higher probability to switch financial advisors in the second deal. This result is consistent with the findings in James (1992) that the likelihood of firms changing underwriters in a subsequent SEO is positively related to the time between the IPO and SEO. Because the transaction-specific asset depreciates over time, the longer the expected interval between two transactions, the less costly it will be to replace the transaction-specific asset. In M&As, the longer time between the two consecutive deals, the less valuable will be the connection between acquirers and their financial advisors hired in the first deal, therefore the less costly for acquirers to switch financial advisors in the second deal. Hypothesis 3 is supported by our empirical test results. Hypothesis 4 predicts that the change in deal characteristics may affect an acquirer s decision to switch financial advisors. An investment bank may have special expertise in a certain type of deal, so an acquirer may pick its financial advisors with respect to important deal characteristics. The results in Table 6 show that the coefficients of the change in tender offer and the change in cash payment are positive and statistically significant. When the second deal is different from the first in terms of the tender offer and cash payment, the acquirer tends to switch its financial advisors. Our cash payment result is consistent with 21

23 the findings in Hayward (2003) that acquirers tend to change their financial advisors when the payment method changes. All the coefficients of the change in toehold are negative and they are statistically significant when the dependent variables are HALF IB and AN Y IB. Betton and Eckbo (2000) report a significantly negative relation between toeholds and target premiums. Greater bidder toeholds reduce the probability of competition and target resistance and are associated with both lower bid premiums and lower pre-bid target stock price runups. The expected payoff to target shareholders is decreasing in the bidder s toehold. Table 6 also shows that the change in advisory fees is not significant in any of our advisor switch definitions, indicating that an acquirer does not change its financial advisors because of advisory fees. This is consistent with the findings in McLaughlin (1990) and McLaughlin (1992) that more than 80% of the advisory fees are contingent on deal completion. All the deals in our sample are completed. The change in completion time is not associated with the switch decision. Finally, the competition in M&As does not lead an acquirer to switch its financial advisors. We document mixed evidence on Hypothesis 4. Hypothesis 5 predicts that both acquirer and target characteristics will affect the switch of acquirer financial advisors. Table 7 presents regression results of investment bank switch variables on the change of firm characteristics from the first deal to the second deal. The results show that acquirers are more likely to switch financial advisors in the second deal when two deal targets are in different industries or when two target listing statuses are different. In addition, an acquirer tends to switch its financial advisors when its own leverage has changed between two consecutive deals. The change in relative size is positive and statistically significant in three advisor switch proxy variables: HALF IB, ANY IB, and LEADIB, suggesting that the change in relative size is also an important factor when an acquirer makes its advisor switch decision. 22