The Choice of Sale Method and its Consequences in Mergers and Acquisitions Current Version: 03/22/2016

Inga Chira California State University – Northridge College of Business and Economics 18111 Nordhoff St Northridge, CA 91330

Nikanor Volkov Mercer University Eugene W. Stetson School of Business and Economics 3001 Mercer University Drive Atlanta, GA 30341

We examine the method by which firms are sold, auctions or one-on-one negotiations. We define and describe a subset of transactions that result from auction failure (i.e., target-attempted auctions that secure only one bidder). Controlling for endogeneity, firm, and transaction specific characteristics, we show that attempted auctions that resulted in one-on-one negotiations are associated with lower final premiums and higher acquirer returns compared with both successful auctions and pure negotiations (negotiations with only one bidder from the outset to the conclusion of the transaction). We find that several target, acquirer, and deal-specific characteristics affect the likelihood of auction failure. The loss of latent (perceived) competition that results from an unsuccessful attempt to auction the target partially shifts the wealth created by a merger or acquisition from targets’ to acquirers’ shareholders. To maximize shareholders’ wealth, targets should carefully consider the likelihood of securing more than one interested bidder prior to initiating an auction. JEL classification: D44, G34 Keywords: Auctions, Negotiations, Failed Auctions, Takeover Process

1.

INTRODUCTION Many sellers in corporate mergers and acquisitions (M&As) aim to identify a buyer that is willing

to pay the highest price. To this end, the target considers two alternative methods of sale: (1) approach multiple potential buyers and attempt to conduct an auction, or (2) engage in a one-on-one negotiation with a single bidder (i.e. pure negotiation). If the seller decides to negotiate with only one potential buyer, it generally reserves the right to approach other potential bidders in the future and convert the negotiation into an auction, without any significant loss of bargaining power. The decision to auction the asset from the outset is more complex because a failed attempt to secure more than one interested bidder may lead to a loss of bargaining power and thus result in a lower sale price. In this study, we identify M&A transactions that were intended as multi-bidder auctions but concluded in one-on-one negotiations. The review of takeover filings reveals that although about 60 percent of M&A transactions are bidder-initiated (consistent with Masulis and Samsir, 2013), the decision to engage in multi-party auction is always target-initiated. If a potential bidder approaches a target to initiate a purchase, target’s management may make a decision to reach out to other potential bidders.1 Thus, targets ultimately control the method of sale decision. Using the sample selection methodology proposed by Boone and Mulherin (2007)2, we identify transactions where target’s management approached more than one party that may have had interest in acquiring the target (i.e. attempted an auction), but only attracted one bidder, who later became the buyer.3 We label these transactions as attempted auctions and show that lack of competitiveness, both real and latent (see Aktas et al., 2010), in such transactions results in target and acquirer wealth effects that differ from both successful auctions and pure negotiations. Controlling for endogeneity, firm-, and transaction-specific characteristics, attempted auctions exhibit a target premium that is 10.38 percentage

1

There are no transaction in our sample where two independent bidders simultaneously approach the target with an offer to acquire it without a prompt by the target. 2 Refer to section 3, Sample Description, for a detailed explanation of the sample collection methodology. 3 For example, over the course of several months in early to mid-2010, Caliper Life Science, Inc., approached seven different parties that it believed might be interested in acquiring the company. The preliminary conversations identified a single interested party, PerkinElmer, Inc. Following a one-to-one negotiations, PerkinElmer purchased Caliper for about $600 million in cash.

2

points lower than that of successful auctions and 9.20 percentage points lower than that of pure negotiations. Furthermore, we find that attempted auctions result in significantly higher acquirer announcement returns when compared to both successful auctions (0.90 percentage points higher) and pure negotiations (0.69 percentage points higher).4 Attempted auctions represent 8.7% of our sample and account for over $130 billion in deal value over the seven-year period (2006-2012) examined in this study. We investigate several possible factors that may affect the likelihood of auction failure: First, we find that attractiveness of the target, proxied for by the level of target’s financial constraint, its market to book ratio, and liquidity, does not affect the probability of auction failure. Second, we show that the likelihood of auction failure is increased following economic crises. This finding is intuitive as the deteriorated credit conditions make it more difficult for the acquirers to finance the acquisition. Third, we show that targets in the high-tech industry are more likely to fail in an attempted auction. This finding can be explained by the high-tech industry being generally associated with more opaqueness and lack of public information (Luo, 2005). Fourth, we demonstrate that low liquidity acquirers are more likely to succeed in attempted auctions. This observation suggests that targets may not give enough consideration to potential lower liquidity acquirers as possible one-on-one negotiation counterparts prior to auction initiation. Finally, we find that attempted auctions are associated with aggressive bidding practices (i.e. preemptive bidding, see Fishman, 1988 and Eckbo, 2009). Attempted auctions are an important sub-set of M&A transactions. Failure to secure more than one interested bidder in an attempted auction results in a significantly lower wealth creation for target’s shareholders. We find some evidence that auction failure implies a loss of latent (perceived) competition that surrounds the transaction and results in a partial wealth shift from the target to the acquirer

4

The reported statistic is for the (-1, +30) window. The results are 0.81 percentage points higher for the (-1, +1) window when compared to successful auctions.

3

shareholders.5 Thus, target’s management should carefully consider the likelihood of auction failure prior to contacting multiple potential bidders. The remainder of the paper is organized in the following way: in Section 2 we discuss relevant literature and outline our main hypotheses; Section 3 describes the data; Section 4 covers the methodology; Section 5 presents the results; Section 6 reports the firm and deal characteristics that affect the likelihood of auction failure; Section 7 explores the pre-announcement bidding process, and Section 8 concludes. 2.

HYPOTHESES DEVELOPMENT Recent studies that focus on the method by which firms are sold, auction or negotiation, examine

transactions from their outset (i.e. the time when the target first considers the sale). As such, Boone and Mulherin (2007) find that the method of sale has no significant effect on the target announcement returns. In other words, the wealth created for the targets’ shareholders is not affected by the method of sale. The authors’ findings contrast with the auction theory, which suggests that auctions should always generate higher purchase prices (Bulow and Klemperer, 1996). Aktas et al. (2010) demonstrate that the lack of difference in target premiums between auctions and negotiations is a product of latent (unobserved) competition. Latent competition in negotiations (and, to some extent, auctions) arises from the perception that other interested parties may enter the bidding process. Failure to secure more than one interested bidder in an attempted auction may result in a loss of latent competition and thus, affect both target’s and acquirer’s shareholders wealth. Bulow and Klemperer (1996) make several key assumptions in developing their model of auctions and negotiations, including that every bidder is willing to make an opening bid that equals the seller’s current value. The expected revenue from an auction then can be written as the expectation of the maximum of the marginal revenues of a certain number of bidders. If the marginal revenue of N bidders is

5

The results are robust controlling for recessionary periods (see Aktas et al., 2010, who use recessions as a proxy for latent competition).

4

greater than or equal to the current value of the target for at least two bidders, it is always more beneficial for the target to hold an auction; no amount of bargaining power is as valuable to the seller as attracting one extra bidder to the sale process. The model thus implies that to maximize the marginal revenue of its sale, a seller should run an open auction and never consider a negotiation. Kirkegaard (2006) provides additional proof for this proposition. The empirical expectation that follows from theoretical predictions is that the final target premiums associated with auctions should exceed those of negotiations. Boone and Mulherin (2007) determine that about a half of the M&A transactions at the end of the last century were settled via one-on-one negotiations.6 The Boone and Mulherin study demonstrates that the method of sale does not affect target announcement returns or premiums. In a follow-up paper (Boone and Mulherin, 2009), the authors explain the lack of difference in target returns by citing the higher costs of auctions (see also French and McCormick, 1984) and the unwillingness of serious bidders to bid against potentially un- or less informed auction participants. Aktas et al. (2010) investigate the presence of latent competition and its impact on the bid premium; They show that, even when a seller engages in a negotiation, the potential for competition (auction) pushes the acquirer to offer a more competitive price. The large number of transactions that settle via one-on-one negotiations may result from targets’ awareness of presence of latent competition that pushes the price up, and (or) targets’ inability to secure more than one interested bidder. Boone and Mulherin (2007) and Gentry and Stroup (2014) both note that less than a half of invited bidders participate in auctions, which may imply that the uncertainty that surrounds the level of latent competition is reduced as a result of auction initiation. In the case of a successful auction, the latent competition converts into a real one; when auction fails, the latent competition disappears. Thus, the wealth effects of intentional (pure) and unintentional negotiations (attempted auctions) may differ significantly. We argue that the characteristics of a specific subset of negotiations—namely, attempted auctions, differ from both successful auctions and pure negotiations. We propose that, because of the loss of the latent competition, attempted auctions are associated with the

6

Our sample, which covers the period of 2006-2012 consists of 70% auctions and 30% negotiations.

5

lowest target premiums and higher acquirer announcement returns when compared to successful auctions and pure negotiations. The loss of latent competition in an attempted auction reduces target’s negotiating power and thus results in the partial wealth shift from the target to acquirer shareholders.7 According to Bulow and Klemperer (1996), the expected revenue from an auction with n + 1 bidders is equal to E{max(MR1, MR2,…MRn, MRn+1)}, whereas the expected revenue from a negotiation is E{max (MR1, 0)}, where MRn is the marginal revenue from a given bidder. This calculation assumes no latent competition, which may affect (improve) the target’s bargaining power in a negotiation (Aktas et al., 2010). We propose a modified definition of the expected revenue of a negotiation with one bidder, in presence of latent competition—that is, a pure negotiation: E {max (MR1, 0)} + ρ [E {max (MR2, MR3, …, MRn, MRn+1)} – E{max (MR1, 0)}], where ρ ≥ 0 is a parameter (measure) of latent competition. The ρ [E{max(MR2, MR3, …, MRn, MRn+1)} – E {max (MR1, 0)}] expression offers a component of the expected revenue due to latent competition. Assuming a special case of ρ = 1 (the effects of latent completion equal those of real competition) and using the properties of expected value, the above expression is reduced to the following: E {max (MR2, MR3,…MRn, MRn+1)} The above special case illustrates that presence of latent competition in a one-on-one negotiation may result in a target expected revenue that equals that of an auction with n + 1 participants. The definition of expected revenue in a negotiation proposed by Bulow and Klemperer, E {max (MR1, 0)}, may thus be viewed as a special case where ρ = 0. This special case represents transactions where the target attempted to hold an auction, but attempted to secure more than one interested party thereby eliminating the latent competition component. In an attempted auction, the lack of either real or latent competition places more of the negotiating power into the hands of the bidder. By attempting and failing an auction, the target may give up the potential for a higher premium from a pure negotiation. Thus, attempted auctions are distinct cases,

7

It is reasonable to expect that some firms recognize their disadvantageous position as a result of attempted auctions and do not proceed to a negotiation with a single interested party, abandoning the sale process.

6

which may lead to the lowest target premiums compared with either successful auctions or pure negotiations.8 A separate strand of finance literature focuses on the effects of M&A announcements on acquirer stock prices in both the short-term (e.g., Asquith, 1983; Moeller et al., 2005) and the long-term (Agrawal et al., 1992; Megginson et al., 2004). Although the results are mixed, the majority of such studies show that both announcement and long-term returns for acquirers are generally negative or, at best, zero (e.g., Schwert, 1996; Loughran and Vijh, 1997; Duchin and Schmidt, 2013). Investors’ recognition that acquirers might pay more for targets in auctions than in negotiations implies the potential for differential acquirer announcement and post-announcement returns between the two methods of sale. If, from a theoretical perspective, negotiated transactions are associated with lower acquisition premiums, market should recognize it and reward (or not punish as much) acquirers when they engage in negotiations. However, in their study of the competitiveness of the 1990s takeover market, Boone and Mulherin (2008) find no significant differences in acquirer announcement returns between auctions and negotiations. The authors attribute this result to the competitive landscape, or latent competition, for targets before the formal announcement of mergers. We argue that investors recognize that attempted auctions are associated with the least transaction competitiveness (actual or latent) and lowest premiums, and, thus, reward (or do not punish as much) acquirers in such transactions. Additionally, the greater information asymmetry (or Hansen’s (2001) competitive information effect) associated with auctions should be diminished significantly in attempted auctions. The acquirer should be able to execute the negotiating power it has gained because of auction failure and obtain better information than it would have obtained had the transaction proceeded as a competitive auction. We expect that, due to the lower initial transaction premium and lower information

8

Just as some transactions initiated as auctions result in negotiations (attempted auctions), transactions intended as negotiations could result in auctions. Such transactions would be controlled by the target, with no shift in the bargaining power between the parties. We do not separately analyze this subset of transactions because the move from a negotiation to an auction simply turns the latent competition into real competition.

7

asymmetry between parties, attempted auctions are associated with better acquirer performance compared to successful auctions and pure negotiations. There exist several possible reasons for auction failure. The most obvious one relates to the attractiveness of the target. If the target is not (less) attractive, it is more likely to fail in securing multiple bidders in an attempted auctions. Second, macroeconomic conditions may also affect the likelihood of auction failure. Aktas et al. (2010) use an indicator variable for the financial crises arguing that macroeconomic conditions affect the level of latent competition and the M&A premiums. Third, it is possible that targets underestimate the likelihood of a successful one-on-one negotiation with some possible buyer. Finally, auction failure may be driven by acquirer’s aggressive bidding practices. Aware of the target’s predisposition to favor auctions, bidders in negotiations may engage in preemptive bidding (place high initial offers to deter the target from running an auction, see Fishman, 1988 and Bulow and Klemperer, 2009). Evidence in support of the expectation of higher opening bids and lower overall premiums in negotiations and attempted auctions would illustrate the effectiveness of preemptive bidding. We investigate the possible reasons for auction failure in Sections 6 and 7. 3.

SAMPLE DESCRIPTION We use the manual data collection methodology proposed by Boone and Mulherin (2007) and

examine the Background of Merger section of the SEC Form 14A filings (mergers), 14D filings (tender offers) and S-4 filings to gather detailed information about the transactions that precede merger announcement for public targets. In addition to assessing the true competitiveness of the bids, this method enables us to identify and record the timing of each step of the sale process and obtain detailed information on the bidding practices. We concentrate on a sample of M&As announced between January 1, 2006, and December 31, 2012. The sample includes public targets and public and private acquirers. More specifically, we construct the sample as follows. First, using the SDC database, we identify 779 announced mergers categorized as majority acquisitions (i.e., M, A, or MA). We then manually look for the SEC Form 14A, 14D, or S-4 filings in the EDGAR Security and Exchange Commission system.

8

We could not locate 14A, 14D, or S-4 for 143 of these transactions, reducing the sample to 636. For these transactions, we identify five main process characteristics from 14A, 14D, and S-4 filings: (1) the date of the first discussion of a potential transaction; (2) the number of potential buyers contacted; (3) the number of potential bidders that enter into a confidentiality agreement with the target, providing a non-binding indication of value, or that receive an invitation to the next round of negotiations without a confidentiality agreement; (4) the number of potential buyers that make offers to acquire the firm; and (5) the date of the first offer. Following Boone and Mulherin (2007), we classify auctions as transactions in which more than one party has signed a confidentiality agreement to conduct further due diligence or placed a non-binding offer to acquire the target. To classify the transactions further, we also note the numbers of firms contacted and proceeding to the next step (i.e., entering confidentiality agreements, providing non-binding offers, or being invited to the next step). With this classification, we break the sample of negotiations down into pure negotiations (deals that started and concluded as negotiations) and attempted auctions (deals that were intended as auctions but resulted in one-on-one negotiations). Finally, we obtain firmand deal-specific characteristics (as described in Section 4, Methodology) from COMPUSTAT and SDC databases. We further trim the sample by eliminating deals in which the acquirer was neither a public nor a private firm (e.g., joint ventures, subsidiaries) and for which we could not calculate the premium or locate deal-specific characteristics. Our final sample thus consists of 469 transactions: 343 auctions and 126 negotiations. The subsample of negotiated deals further consists of 41 attempted (attempted) auctions and 85 pure negotiations. Attempted auctions represent 8.74% of the overall sample, for a total dollar value of $133.26 billion. Table 1 provides the sample description and a breakdown of the sample by year. Unlike Boone and Mulherin (2007), who find a roughly equal number of auctions and negotiations in the 1990s, the late 2000s appear dominated by auctions, such that they account for approximately two-thirds of the total takeover market. Boone and Mulherin (2007) also report an average of 20.67 firms contacted during an

9

auction process, 6.83 firms proceeding to the next step, and 1.57 firms bidding to acquire the target (these results are also consistent with Gentry and Stroup (2014), who show that less than a half of invited bidders chose to participate). For our 2006–2012 sample, the averages indicate that 22.21 potential bidders were contacted in an auction process, 7.89 potential acquirers moved on to the next step, and 2.99 bidders placed offers to acquire the target. The targets that attempted to conduct an auction but attempted to secure more than one interested bidder approached 10.39 parties on average. Our observations suggest greater competitiveness in the recent takeover market compared with that reported in Boone and Mulherin (2007) for the 1990s. Although at first sight, it may appear that attempted auctions result from targets not contacting enough potential acquirers, a closer examination does not support this explanation: Approximately 10% of the targets in the successful auction subsample contacted more than 40 parties. That is, they approached virtually every possible company that might have been interested in acquiring them (including potential corporate and financial acquirers). If we eliminate them from our sample, the remaining sample of targets in successful auctions approached 11.69 parties, which is statistically indistinguishable from the 10.39 parties approached by targets in attempted auctions. Table 2 displays the deal-, target- and acquirer-specific characteristics for our sample. The description of the variables is presented in Section 4. For all indicator variables, we report the proportion of the sample where the indicator variable equals 1; for all continuous variables, we present the means. All data are winsorized at the 1% level. On average, negotiations are larger deals, more likely to involve a public acquirer, and more likely to involve a corporate buyer. Negotiations are also less likely to be target-initiated. It takes 130 days to receive the first bid in pure negotiations and 167/178 days to receive a bid in attempted/successful auctions. Finally, it takes 112 days from the receipt of the first bid to transaction announcement in a negotiation and 224 days in a successful auction (attempted auctions take 115 days from first bid to announcement).

10

When we consider the negotiation subsample to compare pure negotiations with attempted auctions, we find that attempted auctions are more likely to include a go-shop provision,9 be smaller, be announced during the crisis, be a high-tech deal, be target initiated, and be finalized with a same-industry acquirer. The last observation is interesting as it suggests that targets in attempted auction transactions may not give enough consideration to potential one-on-one negotiation counterparts from their own industry. Furthermore, it appears that firms that have not properly assessed the likelihood of auction failure at the outset of the transaction tend to include go-shop provisions in their merger documents. The go-shop provision may be included in an attempt to create a perception that latent competition is present in the transaction. When comparing successful and attempted auctions, we find that it a public acquirer is more likely to acquire a target in an attempted auction and that an attempted auction is more likely to conclude as a tender offer. Similarly to the negotiation results, attempted auctions are more likely in the high-tech industry. Furthermore, attempted auctions appear to be less likely when transaction is target-initiated. This observation may hint at opportunistic management behavior. Following being approached by a bidder, management attempts to initiate a multi-party auction but fails to secure additional bidders. Next, we examine the timing of bid submission. It takes 130 days to receive the first bid in a pure negotiation, whereas in both auctions and attempted auctions it takes over 160 days. It appears that, in an effort to insure that the target does not initiate an auction, bidders present their first offers rather rapidly in pure negotiations. We use the date when the board of directors first discusses its interest in pursuing a sale as the transaction initiation date and the date on which the first bid was submitted as the first bid date. Thus, if we account for the time that it takes to prepare necessary due diligence documentation to initiate the sale process, the potential presence of preemptive bidding in negotiations becomes apparent.10 This

9

A go-shop provision allows the target to approach other potential acquirers during a set period (typically, 30 days) after the completion of a transaction negotiation. This provision is designed to help the target ensure no other parties in the market are willing to acquire it at a price higher than the price negotiated with the current bidder (acquirer). 10 Preparation of due diligence documentation follows from the board of directors’ decision to initiate the sale. We interviewed several law firms and investment banks to confirm that it takes one to four months to prepare the necessary information; the time needed varies with the size and complexity of the target firm.

11

result is consistent with Gentry and Stroup (2014), who show presence of deterrence (preemptive) bidding in their study. Finally, it takes 224 days on average from the first bid to the announcement of a merger in a successful auction, but it takes 115/112 days to move from a first bid to an announcement in an attempted auction/pure negotiation. Since the targets in attempted auctions are aware of the lack of latent competition surrounding the transaction, they may be more motivated to close quicker compared to successful auctions. A legitimate question pertains to the possibility of introducing selection bias by examining negotiations and attempted auctions as subsamples. Companies that proceed with a negotiation after an auction fails may be in financial distress. Targets in attempted auctions may also be inherently different, or simply less attractive than those in successful auctions. If that were the case, one would expect the premiums to be lower for the less attractive targets, regardless of the method of sale. However, we find no major difference in financial characteristics of the targets between the successful auctions, pure negotiations and attempted auctions. Target liquidity or target growth opportunities do not differ between the methods of sale examined. Firms that undergo attempted auctions are not distressed and, arguably, do not have to continue the negotiation with a single bidder after failing to secure additional bidders. Furthermore, to alleviate the potential issue of targets in attempted auctions being less attractive, we calculate the KZ index.11 The p-value of .4984 indicates that the two groups do not differ significantly in their degree of financial constraint. Additionally, we compare the Herfendahl industry concentration indexes for firms that merged through auctions, negotiations, and attempted auctions. No significant differences arise between auctions and negotiations or between auctions, pure negotiations and attempted auctions. An examination of SIC codes of targets acquired in attempted auctions also does not indicate dominance of any specific industries or that the attempted auctions are concentrated in industries that are undergoing consolidation.

11

The KZ index (Kaplan and Zingales, 1997) derives a single measure of liquidity constraints by combining different firm characteristics: cash flow/net capital, market-to-book ratio, debt (short- and long-term), dividends/net capital, and slack/net capital.

12

4.

METHODOLOGY Moeller et al. (2005) classify the competitiveness of a takeover by the number of public bidders.

Boone and Mulherin (2007) define an auction as a process with more than one participant at the confidentiality stage. We adopt Boone and Mulherin’s (2007) methodology to examine the competitiveness of the M&A market, with slight changes. Specifically, we identify an initiated auction as a process in which the target or target’s representative approached more than one potentially interested party with an offer to participate in the sale process. Because more than one potential bidder has been approached,12 all bidders should know that they are not the only potential bidder in the transaction, an assumption which is consistent with Aktas et al., (2010). We define transactions in which multiple parties are approached, but only one bidder ultimately proceeds to the next stage of the merger process, as attempted auctions. We use two main dependent variables to test our hypotheses: final premium and acquirer announcement returns. To find the final premium, we follow standard methodology and calculate it relative to the target price four weeks prior to the transaction announcement. We use a standard event study methodology to calculate acquirer announcement returns using the market-adjusted model, based on two event windows, (–1, +1) and (–1, +30). In addition, we employ a number of other dependent variables to further investigate the differences between auctions, negotiations, and attempted auctions. To identify the differences in the preannouncement period, we look at the initial bid premium and the initial bid premium band. The Initial (Low) Bid is the percentage premium offered to the target originally, relative to the price of the target one day before the first bid. For deals in which bidders relay a range instead of a point bid when communicating the first offer, we calculate the initial bid band as the percentage difference between the initial low and initial high premium bid, relative to the target price one day before the initial offer.

12

Theoretically, it is possible that several bidders could simultaneously approach a target. From reading the background of mergers and acquisitions, however, we do not identify such instances.

13

The final announcement of M&A transactions is usually associated with significant leakage, the magnitude of which may vary depending on the method employed to sell the target. 13 Following Schwert (1996), we calculate leakage over the period of (–42,–1) days and a shorter window of (–30,–1) days. For the multivariate analysis, we use ordinary least squares regressions and apply the NeweyWest correction for heteroscedasticity in the residuals of each linear regression model. Depending on the test, our model is as follows: Final Premium/Acquirer Returns/Initial_Low_Bid (Bid_Band)/Leakage = α + β1Negotiation/AttemptedAuctionsi + β2GOSHOPi +β3DEALVALUEi +β4ASTATUSi+ β5RELATEDi+ β6TENDERi+ β7CRISISi+ β8POSTCRISISi+ β9TTECHi+ β10EQUITYi +β11ATYPEi+ β12TLIQi + β13TGROWTHi+ β14TINITi+β15RELSIZEi+β16ALIQi + β16IMRi + ei, (1) where:  Negotiation (AttemptedAuction) = dummy variable equal to 1 if the transaction was consummated via a negotiation (attempted auction) and 0 otherwise.  GOSHOP = dummy variable equal to 1 if the transactions have the provision in the merger documents, and 0 otherwise. The existence of this provision may enable the target to obtain a higher premium because of the inherent option to approach other potential buyers prior to finalizing the transaction. .  DEALVALUE = a measure of the size of the transaction calculated as ln(deal value). The higher the value of the deal, the more capital the acquirer needs to purchase the target. Previous studies demonstrate that negotiations are associated with larger transactions (e.g., Boone and Mulherin, 2007).  ASTATUS = acquirer status, private or public. The acquirer status may result in different premium, leakage, and post-announcement returns. Bargeron et al. (2008) show that public acquirers pay a price 63% higher than private firms. We assign a dummy value of 1 if the acquirer is public, and 0 otherwise.

13

We check and do not find significant leakage prior to the first bid and thus investigate only the pre-announcement leakage.

14

 RELATED = acquirers in the same industry as the target. Same industry acquirers may have better knowledge and thus may derive at a more precise valuation of the target than unrelated acquirers. This may affect both, the initial offer and premium. Companies in the same industry also may have other reasons to acquire firms, such as synergies and competition reduction. Gorbenko and Malenko (2014) and Fidrmuc (2014) show that strategic bidders pay more for targets than financial acquirers, which leads them to conclude that different targets appeal to different acquirers. We use the SIC code as a measure of relatedness and assign a dummy value of 1 if the two firms take the same three (two)-digit SIC code, and 0 otherwise.  TENDER = the deal is a tender offer, equals to 1, and 0 otherwise. Rau and Vermaelen (1998) show that post-announcement returns differ between mergers and tender offers.  CRISIS = financial crisis period. We assign a dummy value of 1 to mergers announced during the official recession period (December 2007–June 2009), and 0 otherwise. During the crisis, buyers may have been more wary of an expensive acquisition and less willing to pay high premiums. Conversely, acquirers might have been more willing to pay a higher premium during the crisis, when target stock prices are depressed. Aktas et al. (2010) use a similar variable as a proxy for latent competition.  POSTCRISIS = post financial crisis period. We assign a dummy value of 1 to any acquisition announced after June 30, 2009. Additionally, the end of the financial crisis coincides with the time of Galleon case prosecution, which resulted in a change in the levels of insider trading and leakage (see Chira and Madura, 2013).  TTECH = target is in the high-tech industry. Targets in high-tech industries may prompt a higher premium because they signal higher potential growth, for which acquirers are willing to pay a higher price (Ang and Kohers, 2001). Kohers and Kohers (2001) also show that high-tech targets tend to

15

embody more asymmetric information than other firms. We assign a value of 1 if the target is in the high-tech industry and 0 otherwise.14  EQUITY = stock offers. Stock offers allow target’s investors to delay the payment of taxes, which may result in a lower premium demanded. We assign a value of 1 to firms that use 50% or more stock to finance the acquisition, and 0 otherwise.  ATYPE = private equity acquirers. Corporate buyers may be willing to offer higher premiums. Gorbenko and Malenko (2014) show that strategic (corporate) bidders are willing to pay more for targets than are financial acquirers. (Brigida and Madura, 2012) show that transactions where the acquirer is a private equity firm experience lower leakage. We assign a dummy value of 1 when the acquirer is a corporate buyer, and 0 otherwise.  TLIQ = target liquidity. Low target liquidity may result in less bargaining power for the target and more for the acquirer, which may affect the transaction premium. We measure target’s cash and marketable securities scaled by total assets in the quarter before the merger announcement as a proxy for liquidity.  TGROWTH = target growth options. The more attractive the target from the bidder’s perspective, the higher the price the bidder may be willing to pay. We use the market-to-book ratio as the proxy for target’s growth opportunities.  TINIT = a target-initiated transaction. Fidrmuc (2014) shows that bidder-initiated transactions are more likely to be of strategic nature and can generally be viewed as a better fit than those that are target-initiated. The author also demonstrates that the motivation to complete the deal is generally higher in bidder-initiated transactions. Thus, bidders that are approached by the target may not be willing to pay similarly high premiums as bidders who initiate the acquisition process. We read 14A

14

High-tech companies are in the following SIC codes: 3571, 3572, 3575, 3577, and 3578 (computer hardware); 3661, 3663, and 3669 (communications equipment); 3674 (electronics); 3812 (navigation equipment); 3823, 3825, 3826, 3827, and 3829 (measuring and controlling devices); 4899 (communication services); and 7370, 7371, 7372, 7373, 7374, 7375, and 7379 (software).

16

and 14D filings to identify target-initiated transactions, then assign a dummy value of 1 when target initiated the sale, and 0 otherwise.  RELSIZE = relative size. Abnormal announcement returns relate to the size of the acquirer (Faccio et al., 2006), such that smaller targets that have weak bargaining power in the M&A negotiation (Pagano et al., 1998) accept lower premiums than bigger targets. Moeller et al. (2004) show that small acquirers obtain post-announcement returns that are 2% higher than those of large acquirers. We measure the relative size of the target to the acquirer as the ln (acquirer total assets/target r total assets) in the quarter before the merger announcement.  ALIQ = acquirer liquidity. The higher the acquirer’s liquidity, the more likely it is to offer a higher payment to the target. We measure acquirer liquidity as the cash and marketable securities scaled by total assets of the acquirer in the quarter before the merger announcement.  IMR = the Inverse Mills Ratio (control for endogeneity). The ratio is calculated using the probit model, which estimates the likelihood of the target initiating an auction using the target- and marketspecific characteristics (Heckman, 1979). The model, the coefficients, the detail of the estimation process used to calculate the estimates, and a discussion of the estimation results are provided in Appendix A. The estimates of the probit are included to control for endogeneity of the decision to use auction or negotiation as a sale method in an M&A transaction (see for example Gentry and Stroup, 2014 and Schlingemann and Wu, 2014). 5.

RESULTS

a. Univariate Results: Attempted Auctions, Successful Auctions, and Negotiations We present our univariate results in Table 3. In Panel A, we focus on the comparison of subsample of attempted auctions, successful auctions and pure negotiations. On average, attempted auctions result in final target premiums of 35.61%, compared with 41.71% for successful auctions and 43.42% for pure negotiations; the difference for both attempted auctions and successful auctions and attempted auctions and pure negotiations is statistically significant. In contrast, the difference in the final

17

premiums between successful auctions and pure negotiations is insignificant, which is consistent with the findings of Boone and Mulherin (2007). Furthermore, we find that acquirer announcement returns for attempted auctions are positive and statistically significant in both the (–1, +1) and (–1, +30) windows, at +1.34% and +2.16%, respectively. Successful auctions are associated with acquirer returns of –.02% and –.87% over these two windows. The difference between the subsamples of attempted and successful auctions is statistically significant. For pure negotiations, acquirer returns are +1.59% over the (–1, +1) window and +1.28% for the (–1, +30) window; they do not differ statistically from the acquirer returns in attempted auctions. The difference in acquirer returns in both windows between successful auctions and pure negotiations is not statistically significant. The univariate analysis of the sample of attempted auctions reveals lower premiums compared with both successful auctions and pure negotiations and higher acquirer returns compared with successful auctions. The results are consistent with our predictions that lack of latent competition reduces the target premiums and that investors recognize that attempted auctions are “better deals”. Our preliminary results indicate that attempted auctions may result in a partial wealth transfer, from target to acquirer shareholders. In Panel B we focus on the analysis of the pre-announcement bidding practices and the price runup. First, we find that attempted auctions are associated with significantly higher opening bids when compared to both successful auctions. This finding provides first indications of the presence of preemptive bidding in such transactions. The opening bids in attempted auctions are about 20 percent higher than in successful auctions. This finding, coupled with the finding of lower final premiums in attempted auctions, provides first evidence of successful utilization of preemptive bidding strategies by the acquirer. It is noteworthy that significant preemptive bidding is only present in the sample of attempted auctions, but does not appear in pure negotiations. The difference in opening bids between successful auctions and pure negotiations is not statistically significant. The investigation of differences in the width of the initial offer band between the methods of sale reveals that negotiations are associated with the most precise initial valuation (narrowest bands when compared to both attempted and successful

18

auctions). This result is consistent with the expectation that negotiations are associated with lower ex ante information asymmetry, which results in more precise initial offers. Furthermore, negotiations are more likely to be associated with exact bids and bidders in auctions are more likely to provide ranges on the initial stages of the bidding process. We move on to compare the levels of pre-announcement leakage between the three methods of sale. We find no statistical difference in the level of pre-announcement leakage between successful auctions and pure negotiations. About 15% of the final premium occurs during the 30-day period that precedes successful auctions and pure negotiations. In contrast, attempted auctions are associated with the lowest leakage when compared to both methods of sale. This result is both interesting and intriguing. It appears that transactions that result from attempted auctions are better kept secrets. A possible explanation of this finding may be the awareness of the acquirer about lower (under market) price paid for the target company, which in turn reduces the likelihood and the extent of informed trading. b. Multivariate Results: Attempted Auctions, Successful Auctions, and Negotiations To investigate the above results in greater depth, we perform multivariate analysis. In Table 4 we display the results for premiums. In Models 1 and 2, we compare auctions with negotiations in general and in Models 3 and 4, we compare the premiums paid in attempted auctions against those paid in successful auctions and pure negotiations. 15 Generally, we find no evidence that negotiations result in a different final premiums than do auctions.16 These results are consistent with Fidrmuc et al. (2012), who show that the way the firm is sold does not influence the premium. These results are also consistent with Boone and Mulherin (2007), who find no difference in the premium paid for targets in auctions or negotiations for a sample of M&A transactions during the 1990s.

In all model specifications used in this paper, we test for serial autocorrelation (using Stata’s Drukker’s xtserial test), for heteroscedasticity (using the LR test) and for multicollinearity (using VIFs). We do not find evidence that either of the three conditions are present and affect the reported results. 16 In alternative specifications, we use a calculated premium 30 days before the announcement, with similar results. 15

19

The final premium is significantly lower in attempted auctions than in either successful auctions or pure negotiations. Firms that attempt an auction but are unable to secure multiple bidders exhibit the lowest premiums. The final premium in an attempted auction is 10.4 percentage points (9.2 percentage points) lower than that in a successful auction (pure negotiation).17 A potential explanation for the differences is the shift of negotiating power from the target to the acquirer that results from the loss of latent competition. Furthermore, the findings provide first evidence that that preemptive bidding strategies aid in securing the target at a lower price. Given the evidence that the target premium does not differ in successful auction from the premium obtained in pure negotiations while the premium obtained in failed auctions is lower than in other methods of sale, one wonders why the target risks the potential of a failed auction and lower premiums. The existence of attempted auctions points to the fact that targets perceive a benefit from running the auction process. The results may suggest that the targets that choose an auction despite the possibility that it may fail suffer from possible agency issues.

Other deal and parties’ characteristics that influence the final premium are crisis periods, hightech deals, target liquidity, and target initiation. Specifically, the final premium is higher during the crisis and when the target’s liquidity is high. It is lower for high-tech targets and when the target initiates the takeover. We present the results for the acquirer announcement and short-term returns in Table 5. Consistent with the univariate results, no difference arises in acquirer announcement returns between auctions and negotiations in general. Similarly, Boone and Mulherin (2008) note that acquirer returns are not associated with the level of takeover competition. Any differences in announcement returns observed in the univariate analysis appears to be associated with other transaction characteristics, such as deal value and acquirer and target liquidity. Highly liquid acquirers and targets are associated

17

Our results are robust to use of year and industry (2 digit SIC) fixed effects models.

20

with higher acquirer announcement returns. Furthermore, large takeovers and deals announced during the financial crisis are associated with lower acquirer announcement returns. Attempted auctions instead are associated with significant positive acquirer returns over both (1;+1) and (-1;+30) windows. In economic terms, attempted auctions result in one-month announcement acquirer returns that are 0.90 percentage points higher than those of successful auctions (3 day announcement returns are 0.81 percentage points higher).Similarly, the 30 day announcement returns for the acquirers in attempted auctions are 0.69 percentage points higher compare to pure negotiations. Thus, the acquirer seemingly benefits from attempted auctions. The superior performance of acquirers around M&A transactions that settled via attempted auctions could be attributed to investors’ recognition of the wealth transfer from the target to the acquirer, which results from the lower premiums paid. In addition to the method of sale, we find that the returns relate to a number of control variables, such as tender offers and related deals. Additionally, when the target and the acquirer are in the same industry, the market values the deal more than if they came from different industries. As Chevalier (2004) documents, mergers between firms in the same industry are associated with higher event returns, due to their higher operational synergies (Devos et al., 2009). Finally, tender offers are associated with lower returns. 6.

A CLOSER LOOK AT ATTEMPTED AUCTIONS Our findings that attempted auctions exhibit lower premiums and higher acquirer returns

compared with both successful auctions and pure negotiations imply that acquirers benefit but targets suffer from attempted auctions. Identifying mergers that could become attempted auctions would be beneficial to both acquirers and targets. In this section, we investigate characteristics associated with attempted auctions. We carefully examine the 14A and 14D filings of transactions that consummated via attempted auctions and do not find any significant similarities (or specific patterns) between such transactions. Furthermore, the filings do not reveal any significant differences between attempted and successful

21

auctions in the period preceding the submission of bids. The information provided in the filings is rather limited and it does not provide any clear indications that the high initial offers observed in attempted auctions cause the management of the target to immediately cease the ongoing conversations with other potential acquirers. There is also no clear indication in the filings that target’s management discloses the current (high) initial offer to other possible participants and that the disclosure causes other participants to drop out from the bidding process, causing the auctions to fail. Next, we use logistic regression to study if deal, target and acquirer-specific characteristics associated with attempted auctions differ from those that describe pure negotiations and successful auctions. We test different specifications of the following logit model and present the results in Table 6: Auction/Negotiation/Attempted Auction= α+ β1Initial (Low) Bid + β2GOSHOPi +β3DEALVALUEi +β4ASTATUSi+ β5RELATEDi+ β6TENDERi+ β7CRISISi+ β8POSTCRISISi+ β9TTECHi+ β10EQUITYi +β11ATYPEi+ β12TLIQi + β13TMTBi+ β14TINITi+β15RELSIZEi+β16ALIQi +ei, (2) where the dependent variable is either a binary 1 or 0 for auctions, negotiations, and attempted auctions depending on the pair examined (for a binary logit regression in Table 6 Panel A) and 0, 1, 2 for the multinomial logit regression in Table 6 Panel B. We add the explanatory variable Initial (Low) Bid, which is the initial premium offered to the target with the first bid, calculated as is the percentage premium offered to the target originally, relative to the price of the target one day before the first bid. The rest of the explanatory variables are the same as in Equation (1). Fidrmuc et al. (2012) show that targets choose among available sale methods by considering their own firm-specific characteristics, such that high profitability and low leverage tend to characterize firms sold in auctions. They also show that deals initiated by the target that involve private equity buyers are also more likely to be associated with auctions. With a more recent sample, we replicate their analysis and find that negotiations are more likely to be associated with large deals and larger relative size of the parties. Negotiations are also more likely between firms in the same industries. We also show that higher initial bids are more likely to occur in negotiation, a finding that supports the existence of preemptive bidding. Consistent with previous literature findings, we document that negotiations are less likely when

22

target liquidity and growth opportunities are high and when the target initiates the deal. In addition, we find that the likelihood of auctions increases in the post-crisis period. Next, we identify characteristics associated with attempted auctions—those intended to proceed as auctions but that resulted in a negotiation. The comparison in Model 2, of attempted auctions with successful auctions, suggest that attempted auctions are closely related to successful auctions. The central difference that is evident in the attempted auction sample is the higher starting bid. This finding once again points to the success of preemptive bidding. We also show that target-initiated deals are more likely to result in attempted auctions, a finding that is consistent with the notion that auctions in general are more likely to be target initiated. Consistent with our univariate results, go-shop provisions are more likely to appear in transactions consummated via an attempted auction. Although the go-shop provisions are incorporated in the merger documents, they do not cause an increase in the premium paid. Our investigation of the filings reveals that inclusion of these provisions does not necessarily imply that the target acts on such provisions. Furthermore, there is generally a very limited time span during which the target can act on the provision. Thus, the inclusion of the provision is either an attempt by the target to create an illusion of latent competition being present in the transaction to entice the acquirer to offer a higher price, or is a formality included in the documents for legal reasons. As demonstrated by the lower final deal premiums, acquirers are not influenced by the inclusion of these provisions. We also find that lower liquidity acquirers appear more likely to consummate a target in a attempted auction. This finding may suggest that targets do not give enough credit to low potential liquidity bidders as possible counterparts in one-on-one negotiations. Finally, we show that high-tech targets are more likely to fail in their auction attempt. Our results are consistent with the idea that the bidder has some say in the bargaining process, particularly through preemptive bidding. Under certain conditions, a target-initiated auction increases the bargaining power of the bidder. We also investigate how attempted auctions differ from pure negotiations. If we were to find no differences, it would indicate that attempted auctions are truly similar to negotiations. Model 3 presents the results. Consistent with our prior results, we find that high-tech firms are more likely to be associated

23

with attempted auctions. Luo (2005) argues that market informativeness is lower for high-tech firms, and our results affirm this reasoning. Mergers involving high-tech firms are opaque, which leaves the market less knowledgeable about them. Other characteristics that distinguish attempted auctions from negotiations include deal size, the relative sizes of the two parties, and acquirer liquidity. Again, it appears that targets may underestimate the likelihood of a successful transaction via a one-on-one negotiation with potential low liquidity acquirers. It is possible that low liquidity acquirers “win” because they are the only survivor at the bidding table. High liquidity acquirer may have more opportunities for better investments. Attempted auctions also are more common in the post-crisis period.18 Our results for the comparison of attempted auctions to both pure negotiations and successful auctions reveals are generally robust to the use of multinomial logit regression (see Table 6 Panel B). Overall, we find that attempted auctions differ from both negotiations and successful auctions. Accounting for these differences may result in higher (lower) premiums for the target (acquirer). Prior to auction initiation, both the target and the bidder should estimate the possibility of auction failure carefully. The proper choice of a sale method translates into significant economic gains for target’s shareholders. 7.

PRE-ANNOUCEMENT PROCESS Generally, a central objective of a bidder is to acquire the target company at the lowest cost

possible. Since the current M&A market features transactions that concluded via one-on-one negotiations, multi-party auctions and attempted auctions, we expect some differences in the bidding practices employed by the targets. We expect that, in some instances, acquiring firms employ preemptive bidding to entice targets to complete the transaction without executing an open auction (see Fishman, 1988 and Eckbo, 2009). If preemptive bidding is successful in avoiding auctions, higher first bids are likely to occur in negotiations and attempted auctions. Furthermore, in line with previous studies (e.g. French and McCormick, 1984 and Boone and Mulherin, 2009, Fidrmuc, 2014), we assume that negotiations are

18

Possibly, these deals could have attempted during the crisis but been finalized after the crisis.

24

associated with lower information asymmetry between parties, thus, they should also be associated with narrower first bid ranges (when first bids are provided in a form of ranges and not point estimates) compared to auctions. Likewise, greater information asymmetry between the target and the bidders in an auction should result in greater divergences of opinion and thus a wider first offer range (Miller, 1977). The same reasoning may also apply to attempted auctions. The competitive information effect proposed by Hansen (2001) also supports this reasoning: Target managers can justify limiting the information flow in an auction and effectively increase the divergence of opinion associated with the target’s value. The higher first bids and narrower bid range in negotiations result from preemptive bidding and the acquirer’s ex ante superior knowledge of the target. We expect attempted auctions to be associated with preemptive bidding. If a bidder places a very high offer at the outset of the bidding process, it may deter the target from continuing the bargaining process with other potential acquirers. Thus, we would expect higher first bids and narrower bid ranges (bands) in pure negotiations and attempted auctions than in successful auctions. The results for the above propositions are reported in Table 7. The tests of an initial low bid (Model 1) and band (range) of the initial bids (Model 2) are included as two separate specifications. Specification A includes the characteristics of the transaction and the target company; specification B incorporates acquirer characteristics, which reduces the sample size (because some acquirers are private companies). In line with our univariate results, we find attempted auctions are associated with significantly higher opening bids when compared to successful auctions. This finding provides further support to the notion that attempted auctions, at least in part, are caused by aggressive initial bidding practices employed by the acquirer. The multivariate results also provide evidence that bidding in pure negotiations starts at a higher point than in successful auctions, but is lower than in attempted auctions. Furthermore, the results provided in Model 2 provide support to the idea of negotiations being associated with lower information asymmetry than auctions. Negotiations display

25

narrower first bid bands and thus, are associated with a more precise initial valuation by the acquirer. This result is supported by the Hansen (2001) proposition discussed earlier. These results are consistent with two propositions: First, in an attempt to avoid an auction, bidders employ preemptive bidding,19 and second, narrower bid bands (ranges) in negotiations suggest that the transactions are characterized by lower information asymmetry. In addition to the method of sale, we identify several characteristics that influence the first bid and the takeover bid range. The larger the deal, the lower is the original bid, which is consistent with previous literature that indicates that larger transactions are associated with less competitiveness (Aktas et al., 2010). Furthermore, larger transactions are associated with larger, better known firms, such that they also are characterized by lower information asymmetry. During the financial crisis, the initial offer was higher compared with after the crisis, which might be an effect of generally lower (i.e., depressed) stock prices. If these deals looked like bargains, they likely attracted higher initial prices than they would in other, non-crisis periods. In addition, corporate acquirers are more likely to present higher initial offers. More liquid acquirers are more likely to present higher initial offers. The offer band is narrower when there is a go shop provision associated with the deal and to a lesser degree, when acquirer liquidity is high. Anilowski et al. (2009) show in a sample of M&A deals from 1998-2005 that the method of sale (auction versus negotiation) influenced the degree of earnings management. Specifically, the authors argue that the target’s decision to sell via an auction increases as the degree of earnings management increases. This is due to the fact that in a negotiation, the bidders are more apt at uncovering earnings management as compared to bidders in an auction setting. We extend the authors’ analysis to our paper to examine the impact of earnings management on attempted, but failed auctions. Using a standard measure of earnings management (discretionary accruals in the prior year), we calculate the current accruals20. In

19

The results reported in Tables 4 and 6 provide evidence that preemptive bidding practices help acquirers to purchase targets at lower premiums. 20 Following Anilowski et al. (2009), we use the following equation to calculate current accruals: CAit = α + β(ΔSales – ΔARit ) it + εit, defining the variables as in the above paper. ΔSales is the change in sales for a specific firm year, and ΔAR is the change in accounts receivable for the firm year. For the step by step estimation procedure, refer to Anilowski et al. (2009).

26

untabulated results, similar to the prior literature, we find a positive association between earnings management and the probability of auctions. Additionally, we find that attempted auctions are also positively associated with earnings management compared to pure negotiations (our results are weakly significant with a p-value of 0.085). This result illustrates the similarity of auctions and attempted auctions in that both are associated with some (higher) earnings management when compared to negotiations. We do not find a significant difference in earnings management between failed and successful auctions. This finding suggests that action failure is not a result of an overly aggressive preauction earnings management by the target firm that was intending an auction but failed to secure more than one bidder. In general, it appears that the management of the firm was setting up for an auction from the initiation of the transaction. Table 7 Model 3 contains the results for the tests of differences in the magnitude of leakage for the three methods of sale examined. The premium obtained in mergers is known to directly relate to the leakage observed before an announcement. Keown and Pinkerton (1981) and Schwert (1996) document the presence of significant price run up (leakage) prior to M&A announcements. Mergers and acquisitions are not well-kept secrets, such that approximately 15% of the total target premium accrues prior to the announcement (shown in Table 3 Panel B). Negotiations should be associated with lower leakage around transaction announcements. Fewer informed parties in a negotiation should significantly reduce preannouncement informed trading. Our results are consistent with this expectation. In addition, transactions that occur during crisis are associated with higher levels of pre-announcement leakage. Deals where hitech targets exhibit lower levels of run-up. 8.

CONCLUSION By relaxing a key assumption proposed by Bulow and Klemperer (1996), we investigate a

separate method of sale, namely, attempted auctions. We identify firm- and deal-specific characteristics associated with attempted auctions. In contrast with the general auctions-versus-negotiations results, we find that the final premium is significantly lower in attempted auctions than in either successful auctions

27

or pure negotiations. Firms that attempt an auction but are unable to secure multiple bidders exhibit the lowest premiums. The final premium in an attempted auction is lower than that in a successful auction (negotiation). Furthermore, comparison of acquirer returns in attempted auctions with those in successful auctions and pure negotiations reveals a positive significant difference for attempted and successful auctions. We attribute these differences to the shift of negotiating power, from target to acquirer. Exploiting this negotiating power leads to significantly lower transaction premiums and better acquirer performance around the announcement. Attempted auctions thus constitute a subset of negotiations with a unique set of characteristics that differentiates them from both successful auctions and pure negotiations. We then investigate the differences in target, acquirer and deal characteristics between the methods of sale examined in the study. We find that target attractiveness, proxied by its liquidity and growth options does not affect the likelihood of auction failure. We document that targets in the attempted auctions are more likely to be purchased by low liquidity acquirers. This finding suggests that targets may underestimate the probability of a successful one-on-one negotiation with low liquidity potential acquirers prior to initiating an auction. Additionally, we examine the pre-announcement bidding process and document some significant differences in bidding practices employed in auctions, negotiations and attempted auctions. We show that attempted auctions are associated with aggressive (preemptive) bidding practices. Our results imply that, by placing high initial offers, acquirers are able to discourage the target of running an auction, which results in lower total premiums paid by the acquirer. Overall, we document a significant shift in wealth from target’s to acquirer’s shareholders that results from auction failure. The findings suggest that targets should assess the likelihood of a successful one-on-one negotiation (or auction failure) prior to initiating a multi-party auction. Furthermore, targets should be aware and cautious of preemptive bidding strategies employed by potential acquirers.

References Agrawal, A., J. Jaffe, and G. Mandelker, 1992, The post-merger performance of acquiring firms: a reexamination of an anomaly, Journal of Finance, 47, 1605-1621

28

Aktas, N., E. de Bodt, and R. Roll, 2010, Negotiations under a threat of an auction, Journal of Financial Economics, 98, 241-255 Ang, J., and N. Kohers, 2001, The take-over market for privately held companies: the US experience, Cambridge Journal of Economics 25, 723-748. Anilowski, C., Macias, A., and J.M. Sanchez, 2009, Target Firm Earnings Management and the Method of Sale: Evidence from Auctions and Negotiations, Working Paper. Asquith, P., 1983, Merger bids, uncertainty, and stockholder returns, Journal of Financial Economics, 11, 51-83. Bargeron, L., F. Schlingemann, R. Stulz, and C. Zutter, 2008, Why do private acquirers pay so little compared to public acquirers? Journal of Financial Economics, 89 (3):375-90. Boone, A. and H. Mulherin, 2007, How are firms sold? Journal of Finance, 62, 847-875. Boone, A. and H. Mulherin, 2008, Do auctions induce a winner's curse? New evidence from the corporate takeover market, Journal of Financial Economics, 89, 1-19 Boone, A. and H. Mulherin, 2009, Is there one best way to sell a company? Auctions versus negotiations and controlled sales, Journal of Applied Corporate Finance, 21, 28-37 Brigida, M and J. Madura, 2012, Sources of target stock price run-up prior to acquisitions. Journal of Economics and Business, 64 (2):185-98. Bulow, J. and P. Klemperer, 1996, Auctions versus negotiations, American Economic Review, 86, 180194 Bulow, J. and P. Klemperer, 2009, Why do sellers (usually) prefer auctions?, American Economic Review, 99, 1544-1575 Campa, J.M. and S. Kedia, 2002, Explaining the Diversification Discount, Journal of Finance, 57, 173162. Chevalier, J., 2004, What do we know about cross-subsidization? Evidence from merging firms. Advances in Economic Analysis and Policy, 4(1):1 – 27. Chira, I. and J. Madura, 2013, Impact of the Galleon case on informed trading before merger announcements. Journal of Financial Research, 36 (3):325-46. Devos, E., P.-R. Kadapakkam, and S. Krishnamurthy, 2009, How do mergers create value? A comparison of taxes, market power, and efficiency improvements as explanations for synergies. Review of Financial Studies, 22(3):1179 – 1211. Duchin, R. and B. Schmidt, 2013, Riding the merger wave: uncertainty, reduced monitoring, and bad acquisitions, Journal of Financial Economics, 107, 69-88 Eckbo, B., 2009, Bidding strategies and takeover premiums: A review, Journal of Corporate Finance, 15, 149-178

29

Faccio, M., J. McConnel, and D. Stolin, 2006, Returns to acquirers of listed and unlisted targets, Journal of Financial and Quantitative Analysis, 41 (1):197-220. Fidrmuc, J, 2014, Bidding competition and premiums in M&A, Working Paper, Warwick Business School. Fidrmuc, J., P. Roosenboom, R. Paap, and T. Teunissen, 2012, One size does not fit all: selling firms to private equity versus strategic acquirers, Journal of Corporate Finance,18 (4):828-48. Fishman, M., 1988, A theory of preemptive takeover bidding, RAND Journal of Economics, 19, 88-101. French, K. and R. McCormick, 1984, Sealed bids, sunk costs, and the process of competition, Journal of Business, 57, 417-441. Gentry, M. and C. Stroup, 2014, Entry and competition in takeover auctions, Working Paper. Gorbenko, A. and A. Malenko, 2014, Strategic and financial bidders in takeover auctions, Journal of Finance, forthcoming. Hansen, R. 2001, Auction of companies, Economic Inquiry, 39, 30-43. Heckman, J. J. (1979). Sample Selection Bias as a Specification Error, Econometrica, 47(1), 153–161. Kaplan, S. and L. Zingales, 1997, Do investment-cash flow sensitivities provide useful measures of financial constraints? Quantitative Journal of Economics, 112:169-215. Keown, A. and J. Pinkerton, 1981, Merger announcements and insider trading activity: An empirical investigation, Journal of Finance, 36, 855-869. Kirkegaard, R., 2006, A short proof of the Bulow-Klemperer auctions vs. negotiations result, Economic Theory, 28, 449–452. Kohers, N. and T. Kohers, 2001, Takeovers of technology firms: Expectations vs. reality, Financial Management, 30(3), 35-54. Loughran, T. and A. Vijh, 1997, Do long-term shareholders benefit from corporate acquisitions? Journal of Finance, 52, 1765-1790 Luo, Y. 2005. Do insiders learn from outsiders? Evidence from mergers and acquisitions, Journal of Finance 60, 1951-1982. Masulis, R. W. and S. A. Simsir, 2013, Deal initiation in mergers and acquisitions, ECGI Working Paper Series in Finance. Megginson, W., A. Morgan, and L. Nail, 2004, The determinants of positive long-term performance in strategic mergers: Corporate focus and cash, Journal of Banking and Finance, 28, 523-552 Miller, E. M. (1977). Risk, uncertainty, and divergence of opinion. The Journal of Finance, 32(4), 11511168.

30

Moeller, S., F. Schlingemann, and R. Stulz, 2004, Firm size and the gains from acquisitions, Journal of Financial Economics , 73 (2):201-28. Moeller, S., F. Schlingemann, and R. Stulz, 2005, Wealth destruction on a massive scale? A study of acquiring-firm returns in the recent merger wave, Journal of Finance, 60, 757-782 Pagano, M., F. Panetta, and L. Zingales, 1998, Why do companies go public? An empirical study. Journal of Finance, 53(1):27-64. Rau, R. and T. Vermaelen, 1998, Glamour, value and the post-acquisition performance of acquiring firms, Journal of Financial Economics, 49 (2):223-53. Schlingemann, F. and H. Wu, 2014, Determinants and shareholder wealth effects of the sales method in M&A, Working Paper. Schwert, W., 1996, Markup pricing in mergers and acquisitions, Journal of Financial Economics, 41, 153-192

Volkov, N. and G. Smith, 2015, Corporate diversification and firm value during economic downturns. The Quarterly Review of Economics and Finance 55, 160-175.

31

Table 1. Sample description by type of takeover and year Panel A: Competitiveness by method of sale Number of contacted parties Number of interested parties Number of parties that bid Target initiated takeover N

Auctions 22.21 a 7.89 2.99 39.47% 343

Negotiations 3.51 1 1 16.00% 126

Pure Negotiations 1 1 1 11.90% 85

Attempted Auctions 10.39 1 1 24.39% 41

Panel B: Sample by year Year Auctions Pure Negotiations Attempted Auctions Total 2006 61 24 4 89 2007 69 19 8 96 2008 44 5 7 56 2009 21 8 4 33 2010 50 11 6 67 2011 49 9 6 64 2012 49 9 6 64 Total 343 85 41 469 a Approximately 10% of the targets in successful auction subsample contacted more than 40 parties, such that they approached virtually every possible company that might have been interested in acquiring them (including corporate and financial acquirers). Such targets skew the sample. Eliminating them from the sample of auctions reveals that targets in auctions approached 11.69 parties, which is statistically indistinguishable from the 10.39 parties approached in attempted auctions.

32

Table 2. Firm- and deal-specific characteristics This table reports the univariate statistics associated with the sample characteristics (independent variables). We split the sample two ways: (1) auction versus negotiation and (2) pure negotiation versus attempted auction. GOSHOP equals 1 if a go-shop provision is included in the merger documents and 0 otherwise; DEALVALUE is a measure of the size of the deal, equal to the natural logarithm (ln) of deal value obtained from SDC; RELATED equals 1 if the target and acquirer are in the same two-digit SIC code and 0 otherwise; TENDER equals 1 if the deal is a tender offer and 0 otherwise; CRISIS equals 1 if the deal was announced between December 2007 and June 2009 and 0 otherwise; POSTCRISIS equals 1 if the deal was announced after June 2009 and 0 otherwise; TTECH equals 1 if the target is in a high-tech SIC and 0 otherwise; EQUITY equals 1 if the deal was financed with 50% or more equity and 0 otherwise; ATYPE equals 1 if the acquirer is a corporate buyer and 0 if the acquirer is a financial buyer; TLIQ is the natural logarithm (ln) of the target’s cash and marketable securities scaled by total assets; TGROWTH is the target’s market-to-book ratio; ASTATUS equals 1 if the acquirer is public and 0 otherwise; TINIT equals 1 if the target initiated the sale and 0 otherwise; RELSIZE is the relative size of the bidder to the target, measured as ln (acquirer TA)/ ln(target TA); ; ALIQ is the natural logarithm (ln) of the target’s cash and marketable securities scaled by total assets ; the KZ index is the measure of the degree of financial constraint; Initiation to First Bid represents the days from the first mention of the possibility of a transaction until the first bid is received; First Bid to Announcement represents the days from the time the first bid was received until announcement date of the merger; and the HHI Index measures industry concentration. *, **, and *** indicate significance at 10%, 5%, and 1% levels, respectively. Successful Negotiations Pure Attempted Successful Auctions – Pure Negot. Succ Auctions Auctions Negotiations Auctions Negotiations (p-value) – Attempted – Attempted Auctions Auctions (p-value) (p-value) GOSHOP 6.12 3.97 2.35 7.31 2.15 -4.96* -1.19 (0.183) (0.090) (0.365) DEALVALUE 6.32 6.94 7.25 6.30 -0.62*** 0.95*** 0.02 (0.000) (0.000) (0.434) RELATED 40.52 32.54 25.88 46.34 7.98* -20.46** -5.82 (0.054) (0.018) (0.194) TENDER 32.36 38.10 36.47 41.46 -5.74 -9.98 -9.10* (0.128) (0.294) (0.096) CRISIS 21.28 22.22 18.82 29.27 -0.94 -10.45* -7.99 (0.414) (0.092) (0.102) POST CRISIS 42.27 36.50 32.94 43.90 5.77 -10.96 -1.63 (0.128) (0.116) (0.365) TTECH 34.98 30.95 23.53 46.34 3.99 -22.81*** -11.36* (0.204) (0.007) (0.058) EQUITY 4.37 6.34 5.88 7.32 -1.97 -1.44 -2.95 (0.189) (0.375) (0.187) ATYPE 90.67 94.44 95.29 92.68 -3.77* 2.61 -2.01 (0.095) (0.274) (0.353) TLIQ 28.23 26.19 24.27 30.19 2.04 -5.92 -1.96 (0.216) (0.101) (0.302) TGROWTH 3.56 2.24 1.94 2.89 1.32 -0.95 0.67 (0.138) (0.211) (0.142) TINIT 39.47 16.00 11.90 24.39 23.47*** -12.49** 15.08** (0.000) (0.036) (0.017)

33

ASTATUS

69.39

80.95

80.00

82.93

-11.56*** -2.93 -13.54** (0.004) (0.345) (0.043) RELSIZE 2.76 2.90 2.87 2.94 -0.14 -0.07 -0.18 (0.263) (0.422) (0.321) ALIQ 19.21 17.55 17.19 18.28 1.66 -1.09 0.93 (0.183) (0.368) (0.425) KZ index 0.08 0.06 0.07 0.07 0.02 -0.004 0.01 (0.308) (0.498) (0.229) Init. To First Bid 178 142 130 167 36* -37* 11 (days) (0.088) (0.077) (0.466) First Bid to 224 b 114 112 115 110 *** -3 109*** Ann. (days) (0.000) (0.910) (0.000) HHI Index 4.93 5.09 5.04 4.55 -0.16 0.49 0.38 (0.169) (0.106) (0.313) a The 175 days for Initial to First Bid includes both negotiations and attempted auctions subsamples. For pure negotiations, on average, it takes 143 days from the initial mention of the idea of pursuing a merger in the 14A and 14D filings until the receipt of the first bid, which is 78 days shorter than for successful auctions (difference is significant at 1% level). b We eliminated the number of days from the average that were >1,000; a few outlier deals were first mentioned up to five years before they were announced.

34

Table 3. Univariate analysis Panel A: Final premium and acquirer returns: attempted auctions versus pure negotiations and successful auctions Final Premium is obtained from the SDC database (calculated from the target price four weeks before the M&A announcement), and acquirer return (A_Return) is calculated as the acquirer cumulative abnormal return over the periods (–1,+1) and (–1,+30). *, **, and *** indicate significance at 10%, 5%, and 1% levels, respectively.

Mean N Difference (Attempted Auct – Succ Auct/ Pure Neg) Difference (Succ Auct – Pure Neg)

Final Premium Attempted Auctions 35.61*** 41

Final Premium Successful Auctions

Final Premium Pure Negotiations

41.71*** 343 -6.10* (0.059)

43.42*** 85 -7.81** (0.048)

A_Return (–1,+1) Attempted Auctions 1.34** 33

A_Return (–1,+1) Successful Auctions -0.02 224 1.32* (0.067)

-1.71 (0.633)

A_Return (–1,+1) Pure Negotiations 1.59** 54 -0.25 (0.459)

A_Return (–1,+30) Attempted Auctions 2.16*** 33

A_Return (–1,+30) Successful Auctions -0.87* 224 3.03* (0.089)

1.61 (0.129)

A_Return (–1,+30) Pure Negotiations 1.28 54 0.88 (0.202)

2.15 (0.101)

Panel B: Initial offers, initial offer band, and leakage: attempted auctions versus pure negotiations and successful auctions The Initial (Low) Bid is the percentage premium offered to the target originally, relative to the price of the target one day before the first bid, and the band (Initial Offer Band) is the percentage difference between the initial low and initial high bound of the offers, relative to the target price one day before the first offer. Target leakage (T_Leakage) is the target’s cumulative abnormal return in the window (–30,–1). *, **, and *** indicate significance at 10%, 5%, and 1% levels, respectively.

Mean N Difference (Attempted Auct – Succ Auct/Pure Neg) Difference (Succ Auct – Pure Neg)

Initial (Low) Bid Attempted Auctions

Initial (Low) Bid Successful Auctions

Initial (Low) Bid Pure Negotiations

27.48 41

22.59 343 4.89** (0.033)

25.30 85 2.18 (0.220)

Initial Offer Band Attempted Auctions 5.82 41

-2.71 (0.355)

Initial Offer Band Successful Auctions 7.53 343 -1.70 (0.2102)

Initial Offer Band Pure Negotiations 3.11 85 2.71* (0.097)

4.42** (0.039)

35

T_Leakage (–30,–1) Attempted Auctions 3.08* 39

T_Leakage (–30,–1) Successful Auctions 6.85*** 318 -3.65 (0.120)

T_Leakage (–30,–1) Pure Negotiations 6.32*** 80 -3.37 (0.138)

0.53 (0.379)

Table 4. Premiums The dependent variable is Premium. Models 1 and 2 include all firms. Model 3 includes attempted auctions and successful auctions; Model 4 includes attempted auctions and pure negotiations. Premium is obtained from the SDC database (calculated from the target price four weeks before the M&A announcement). GOSHOP equals 1 if a go-shop provision is included in the merger documents and 0 otherwise; DEALVALUE is a measure of the size of the deal, equal to the natural logarithm (ln) of deal value obtained from SDC; RELATED equals 1 if the target and acquirer are in the same two-digit SIC code and 0 otherwise; TENDER equals 1 if the deal is a tender offer and 0 otherwise; CRISIS equals 1 if the deal was announced between December 2007 and June 2009 and 0 otherwise; POSTCRISIS equals 1 if the deal was announced after June 2009 and 0 otherwise; TTECH equals 1 if the target is in a high-tech SIC and 0 otherwise; EQUITY equals 1 if the deal was financed with 50% or more equity and 0 otherwise; ATYPE equals 1 if the acquirer is a corporate buyer and 0 if the acquirer is a financial buyer; TLIQ is the natural logarithm (ln) of the target’s cash and marketable securities scaled by total assets; TGROWTH is the target’s market-to-book ratio; ASTATUS equals 1 if the acquirer is public and 0 otherwise; TINIT equals 1 if the target initiated the sale and 0 otherwise; RELSIZE is the relative size of the bidder to the target, measured as ln (acquirer TA)/ ln(target TA); ; ALIQ is the natural logarithm (ln) of the target’s cash and marketable securities scaled by total assets. IMR is the Inverse Mills Ratio calculated using the probit model, which estimates the likelihood of the target initiating an auction using the target- and market-specific characteristics .All regressions are run with Newey-West correction for heteroscedasticity. We also included ALIQ and RELSIZE in Models 3 and 4 but found no difference in the results. The variance inflation factors ranged from 1.21 to 1.98. *, **, and *** indicate significance at 10%, 5%, and 1% levels, respectively. Variable

Model 1: Premium

Model 2: Premium

Model 3 : Premium Successful and Attempted Auctions

Model 4: Premium Pure Negotiations and Attempted Auctions

Intercept Negotiation Attempted Auction GOSHOP DEALVALUE RELATED TENDER CRISIS POST CRISIS TTECH EQUITY ATYPE TLIQ TGROWTH TINIT ASTATUS RELSIZE (T/A) ALIQ

55.84 (0.000)*** -2.622 (0.936)

47.67 (0.003)*** -3.31 (0.933)

66.60 (0.000)***

52.81 (0.002)***

-3.27 (0.503) -0.51 (0.786) -7.79 (0.031)** -1.98 (0.538) 10.95 (0.014)** 7.41 (0.054)* -10.75 (0.014)** -8.49 (0.094)* -9.49 (0.143) 27.32 (0.003)*** 0.83 (0.175) -12.46 (0.013)** 2.32 (0.472)

-6.13 (0.363) -0.40 (0.881) -7.13 (0.135) -2.26 (0.580) 9.15 (0.026)** 4.99 (0.347) -15.76 (0.007)*** -8.20 (0.115) -12.31 (0.373) 20.21 (0.076)* 0.64 (0.103) -6.32 (0.062)*

-10.38 (0.038)** -3.16 (0.516) -1.63 (0.434) -6.87 (0.079)* -4.74 (0.198) 12.09 (0.016)** 6.58 (0.131) -10.12 (0.029)** -14.09 (0.003)*** -10.16 (0.135) 21.36 (0.028)** 1.26 (0.017)** -14.21 (0.009)***

-9.20 (0.022)** -7.28 (0.106) -1.20 (0.505) -3.33 (0.446) 4.50 (0.223) 10.20 (0.044)** 5.11 (0.672) -14.28 (0.080)* 3.20 (0.605) -5.33 (0.504) 30.10 (0.000)*** 1.52 (0.362) -10.24 (0.060)*

IMR

-2.73 (0.253) 0.000*** 0.1316 445

-1.90 (0.103) 23.24 (0.124) -3.95 (0.211) 0.003*** 0.1519 330

-1.83 (0.519) 0.000*** 0.1409 364

-3.20 (0.516) 0.008*** 0.1654 123

Prob >F score R-Squared N

36

Table 5. Acquirer returns Models 1 and 2 include all public acquirers, Models 3 and 4 include attempted auctions and successful auctions, and Models 5 and 6 include attempted auctions and pure negotiations. GOSHOP equals 1 if a go-shop provision is included in the merger documents and 0 otherwise; DEALVALUE is a measure of the size of the deal, equal to the natural logarithm (ln) of deal value obtained from SDC; RELATED equals 1 if the target and acquirer are in the same two-digit SIC code and 0 otherwise; TENDER equals 1 if the deal is a tender offer and 0 otherwise; CRISIS equals 1 if the deal was announced between December 2007 and June 2009 and 0 otherwise; POSTCRISIS equals 1 if the deal was announced after June 2009 and 0 otherwise; TTECH equals 1 if the target is in a high-tech SIC and 0 otherwise; EQUITY equals 1 if the deal was financed with 50% or more equity and 0 otherwise; ATYPE equals 1 if the acquirer is a corporate buyer and 0 if the acquirer is a financial buyer; TLIQ is natural logarithm (ln) of the target’s cash and marketable securities scaled by total assets ; TGROWTH is the target’s market-to-book ratio; TINIT equals 1 if the target initiated the sale and 0 otherwise; RELSIZE is the relative size of the bidder to the target, measured as ln (acquirer TA)/ ln(target TA);; ALIQ is the natural logarithm (ln) of the target’s cash and marketable securities scaled by total assets. IMR is the Inverse Mills Ratio calculated using the probit model, which estimates the likelihood of the target initiating an auction using the target- and market-specific characteristics . All regressions are run with Newey-West correction for heteroscedasticity. The variance inflation factors range from 1.04 to 1.29. *, **, and *** indicate significance at 10%, 5%, and 1% levels, respectively. Variable

Intercept Negotiation Attempted Auction GOSHOP DEALVALUE RELATED TENDER CRISIS POST CRISIS TTECH EQUITY ATYPE TLIQ TGROWTH TINIT RELSIZE (T/A) ALIQ IMR Prob >F R-Squared N

Model 1: Auctions vs. Negotiations (-1.+1) 0.20 (0.004)*** 0.03 (0.106)

Model 2: Auctions vs. Negotiations (-1,+30) 0.17 (0.061)* 0.03 (0.228)

0.04 (0.270) -0.88 (0.050)** 1.02 (0.319) -0.79 (0.221) -2.90 (0.000)*** 0.31 (0.730) -0.64 (0.412) -3.75 (0.208) -0.33 (0.367) 1.12 (0.074)* 0.04 (0.750) -0.40 (0.690) 0.30 (0.349) 3.21 (0.090)* 0.04 (0.225) 0.087* 0.1021 269

0.10 (0.179) -1.01 (0.046)** 1.14 (0.054)* -1.15 (0.026)** -0.50 (0.851) -1.03 (0.318) -0.25 (0.918) -3.07 (0.300) 0.50 (0.286) 3.44 (0.035)** 0.02 (0.227) -0.42 (0.365) -0.06 (0.363) 3.08 (0.071)* 0.06 (0.116) 0.044** 0.1049 269

Model 3: Attempted vs. Successful Auctions (-1,+1) 0.61 (0.505)

Model 4: Attempted vs. Successful Auctions (-1,+30) 0.78 (0.343)

Model 5: Attempted Auctions vs. Pure Negotiations (-1,+1) 0.31 (0.000)***

Model 6: Attempted Auction vs. Pure Negotiations (-1,+30) 1.17 (0.200)

0.81 (0.035)** 0.56 (0.219) -1.07 (0.014)** 0.94 (0.437) -0.69 (0.460) -2.61 (0.038)** 0.68 (0.501) -0.40 (0.677) -4.17 (0.110) -0.53 (0.269) 3.14 (0.093)* -0.03 (0.728) -0.30 (0.773) 0.24 (0.387) 2.04 (0.091)* 0.02 (0.339) 0.0620* 0.1314 215

0.90 (0.028)** 0.57 (0.384) -1.18 (0.036)** 1.13 (0.060)* -1.14 (0.048)** 1.04 (0.666) -1.18 (0.021)** -0.71 (0.694) -3.05 (0.331) 0.44 (0.331) 3.60 (0.070)* -0.04 (0.075)* -0.30 (0.757) -0.03 (0.600) 2.88 (0.041)** 0.03 (0.220) 0.003*** 0.1722 215

0.53 (0.121) 0.71 (0.247) -1.02 (0.045)** 1.11 (0.060)* -1.02 (0.188) -0.03 (0.091)* -0.03 (0.216) -0.35 (0.165) -3.03 (0.503) -0.08 (0.095)* 3.22 (0.082)* 0.03 (0.347) -0.39 (0.645) -0.04 (0.235) 3.19 (0.008)*** 0.75 (0.088)* 0.000*** 0.1380 86

0.69 (0.055)* 0.59 (0.523) -1.49 (0.023)** 1.15 (0.039)** -1.05 (0.040)** 0.01 (0.703) -0.08 (0.830) 0.28 (0.455) -3.15 (0.129) 0.03 (0.459) 3.46 (0.051)* -0.03 (0.276) -0.71 (0.111) -0.23 (0.034)** 1.14 (0.126) 0.63 (0.104) 0.002*** 0.1612 86

37

Table 6. Determinants of auctions, negotiations, and attempted auctions Panel A. Binomial logit regressions21 This table presents the multivariate logit results identifying characteristics associated with negotiations versus auctions, attempted versus successful auctions, and attempted auctions versus pure negotiations for publically traded targets and acquirers. The Initial (Low) Bid is the percentage premium offered to the target originally, relative to the price of the target one day before the first bid; GOSHOP equals 1 if a go-shop provision is included in the merger documents and 0 otherwise; DEALVALUE is a measure of the size of the deal, equal to the natural logarithm (ln) of deal value obtained from SDC; ASTATUS equals 1 if the acquirer is public and 0 otherwise; RELATED equals 1 if the target and acquirer are in the same two-digit SIC code and 0 otherwise; TENDER equals 1 if the deal is a tender offer and 0 otherwise; CRISIS equals 1 if the deal was announced between December 2007 and June 2009 and 0 otherwise; POSTCRISIS equals 1 if the deal was announced after June 2009 and 0 otherwise; TTECH equals 1 if the target is in a high-tech SIC and 0 otherwise; EQUITY equals 1 if the deal was financed with 50% or more equity and 0 otherwise; ATYPE equals 1 if the acquirer is a corporate buyer and 0 if the acquirer is a financial buyer; TLIQ is the natural logarithm (ln) of the target’s cash and marketable securities scaled by total assets ; TGROWTH is the target’s market-to-book ratio; TINIT equals 1 if the target initiated the sale and 0 otherwise; RELSIZE is the relative size of the bidder to the target, measured as ln (acquirer TA)/ ln(target TA);; ALIQ is the natural logarithm (ln) of the target’s cash and marketable securities scaled by total assets . IMR is the Inverse Mills Ratio calculated using the probit model, which estimates the likelihood of the target initiating an auction using the target- and market-specific characteristics . The variance inflation factors range from 1.33 to 2.08. A The dependent variable is negotiation = 1 and auction = 0 in Model 1, attempted auctions = 1 and successful auctions = 0 in Model 2, and attempted auctions = 1 and pure negotiation = 0 in Model 3. Model 1 compares auctions to negotiations, Model 2 compares attempted auctions with successful auctions, and Model 3 compares attempted auctions with pure negotiations. *, **, and *** indicate significance at 10%, 5%, and 1% levels, respectively. Variable

Intercept Initial (Low) Bid GOSHOP DEALVALUE RELATED TENDER CRISIS POST CRISIS TTECH EQUITY ATYPE TLIQ TGROWTH TINIT RELSIZE (T/A) ALIQ Prob >F R-Squared N

Model 1: Negotiations vs. Auctions -3.43 (0.003)*** 0.14 (0.043)** 0.54 (0.344) 0.36 (0.000)*** 0.25 (0.042)** 0.29 (0.334) -0.26 (0.484) -0.74 (0.018)** -0.37 (0.233) 0.50 (0.357) -1.10 (0.100)* -1.09 (0.096)* -0.011 (0.017)** -1.24 (0.000)*** 0.19 (0.022)** 1.30 (0.184) 0.0003*** 0.1909 330

Model 2: Attempted Auctions vs. Successful Auctions -1.02 (0.466) 0.113(0.035)** 1.53 (0.051)* -0.09 (0.554) 0.09 (0.808) 0.503 (0.198) 0.28 (0.556) 0.15 (0.735) 0.55 (0.012)** 0.74 (0.301) 0.19 (0.757) -0.07 (0.841) -0.05 (0.384) 0.80 (0.016)** 0.07 (0.559) -1.67 (0.028 )** 0.0746* 0.0548 261

21

Model 3: Attempted Auctions vs. Pure Negotiations 4.40 (0.156) 0.05 (0.312) 0.49 (0.693) -0.97 (0.005)*** 0.53 (0.276) 0.10 (0.841) 0.92 (0.107) 1.17 (0.039)** 1.11 (0.028)** 0.14 (0.866) 0.15 (0.821) -0.56 (0.572) 0.09 (0.413) 0.75 (0.166) -0.46 (0.016)** -3.03 (0.014)** 0.0121** 0.1154 101

We also ran models that included both private and public acquirers (N = 448 for Model 1, N = 364 for Model 2, N = 123 for Model 3) and include year fixed effects but found no significant differences in the results.

38

Panel B: Multinomial Regression where Attempted Auctions=0 (baseline). Pure negotiations=1 and successful auctions=2. Variable

Opening Bid GOSHOP DEALVALUE RELATED TENDER CRISIS POST CRISIS TTECH EQUITY ATYPE TLIQ TGROWTH TINIT RELSIZE (T/A) ALIQ Prob >F R-Squared N

Negotiations vs. Attempted Auctions (Attempted=base) -0.36 (0.448) 0.67 (0.443) -0.70 (0.012)** 0.31 (0.684) 0.95 (0.215) 0.88 (0.118) 1.80 (0.014)** 0.79 (0.228) 0.39 (0.760) 0.13 (0.983) -1.12 (0.189) 0.09 (0.711) 1.10 (0.158) -0.46 (0.016)** -2.68 (0.054)* 0.0001*** 0.1425 330

Auctions vs. Attempted Auctions (Attempted=base) -0.20 (0.071)* 1.55 (0.066)* -0.24 (0.355) 0.05 (0.940) 1.14 (0.118) 0.48 (0.224) 1.73 (0.117) 0.40 (0.036)** 0.85 (0.485) 0.17 (0.985) -0.70 (0.632) 0.02 (0.901) 1.29 (0.021)** 0.07 (0.559) -1.62 (0.068)* 0.0001*** 0.1425 330

39

Table 7. Pre-announcement process: Auctions, Negotiations, and Attempted Auctions The dependent variable is the Initial (Low) Bid in Model 1 and the initial offer band width in Model 2. The Initial (Low) Bid is the percentage premium offered to the target originally, relative to the price of the target one day before the first bid, and the band is the percentage difference between the initial low and initial high bound of the offers, relative to the target price one day before the first offer. Specification A includes deal- and target-specific characteristics; specification B also includes acquirer characteristics. The Leakage dependent variable is calculated as the cumulative abnormal return over the period (–30,–1). The sample includes all firms, pure negotiations, attempted auctions, and successful auctions. GOSHOP equals 1 if a go-shop provision is included in the merger documents and 0 otherwise; DEALVALUE is a measure of the size of the deal, equal to the natural logarithm (ln) of deal value obtained from SDC; RELATED equals 1 if the target and acquirer are in the same two-digit SIC code and 0 otherwise; TENDER equals 1 if the deal is a tender offer and 0 otherwise; CRISIS equals 1 if the deal was announced between December 2007 and June 2009 and 0 otherwise; POSTCRISIS equals 1 if the deal was announced after June 2009 and 0 otherwise; TTECH equals 1 if the target is in a high-tech SIC and 0 otherwise; EQUITY equals 1 if the deal was financed with 50% or more equity and 0 otherwise; ATYPE equals 1 if the acquirer is a corporate buyer and 0 if the acquirer is a financial buyer; TLIQ is the natural logarithm (ln) of the target’s cash and marketable securities scaled by total assets ; TGROWTH is the target’s market-to-book ratio; ASTATUS equals 1 if the acquirer is public and 0 otherwise; TINIT equals 1 if the target initiated the sale and 0 otherwise; RELSIZE is the relative size of the bidder to the target, measured as ln (acquirer TA)/ ln(target TA); ALIQ is the natural logarithm (ln) of the target’s cash and marketable securities scaled by total assets . IMR is the Inverse Mills Ratio calculated using the probit model, which estimates the likelihood of the target initiating an auction using the target- and market-specific characteristics . All regressions are run with Newey-West correction for heteroscedasticity. The variance inflation factors range from 1.03 to 1.58. *, **, and *** indicate significance at 10%, 5%, and 1% levels, respectively. Variable Intercept Pure Negotiation Attempted Auction GOSHOP DEALVALUE RELATED TENDER CRISIS POST CRISIS TTECH EQUITY ATYPE TLIQ TGROWTH TINIT ASTATUS RELSIZE (T/A) ALIQ IMR Prob >F R-Squared N

Model 1A: Initial (Low) Bid 50.17 (0.188) 6.44 (0.060)* 12.27 (0.035)** 4.47 (0.822) -6.20 (0.049)** 1.01 (0.900) -2.60 (0.962) 33.52 (0.000)*** 5.02 (0.534) -1.46 (0.794) -1.40 (0.895) 5.04 (0.060)* 9.93 (0.522) -1.60 (0.070)* 6.49 (0.222) -6.10 (0.704)

-2.10 (0.825) 0.009*** 0.1141 445

Model 1B: Initial (Low) Bid 31.47 (0.157) 8.79 (0.031)** 18.07 (0.024)** 11.84 (0.295) -7.38 (0.026)** 6.77 (0.441) -3.14 (0.804) 26.60 (0.007)*** 4.97 (0.617) -8.09 (0.295) -12.12 (0.288) 19.98 (0.012)** 11.74 (0.488) -1.11 (0.349) 0.02 (0.558) -2.84 (90.095)* 42.94 (0.075)* -2.67 (0.721) 0.004*** 0.1728 330

Model 2A: Initial Offer Band 14.12 (0.001)*** -2.18 (0.013)** 1.84 (0.375) -4.02 (0.065)* -1.11 (0.907) -2.51 (0.124) 1.11 (0.329) -3.14 (0.552) -1.31 (0.454) -2.02 (0.76)* -3.79 (0.009)*** -2.57 (0.317) 3.98 (0.878) -1.24 (0.254) 1.44 (0.565) 0.21 (0.856)

-1.41 (0.216) 0.000*** 0.0899 445

40

Model 2B: Initial Offer Band 18.42 (0.001)*** -1.98 (0.060)* 2.09 (0.377) -5.63 (0.005)*** -1.20 (0.837) -2.93 (0.112) 1.52 (0.257) 0.38 (0.841) -1.74 (0.332) -1.49 (0.288) -2.06 (0.216) -4.98 (0.147) 3.03 (0.376) -1.90 (0.290) 0.73 (0.780) -1.14 (0.686) -9.99 (0.015)** -1.64 (0.144) 0.000*** 0.1363 330

Model 3: Leakage 20.87 (0.002)*** -2.80 (0.071)* -3.04 (0.105) 1.67 (0.505) -3.20 (0.063)* -1.80 (0.222) -5.15 (0.850) 6.21 (0.011)** -1.25 (0.358) -3.22 (0.044)** -3.88 (0.442) -0.98 (0.603) 3.18 (0.204) -1.55 (0.130) 2.58 (0.022)**

-1.80 (0.084)* 0.040*** 0.0722 415

Appendix A Similarly to Campa and Kedia (2002) and Volkov and Smith (2015), we model an endogenous self-selection model, applying Heckman’s correction to control for the self-selection bias. In the first stage, we conduct a probit analysis using variables that are likely to impact the target’s selection of the method of sale. Inverse Mills Ratio (IRM) is calculated using the probit model. The ratio estimates the likelihood of the target initiating an auction using the target- and market-specific characteristics. The estimates of the probit model (IMR) are included to control for endogeneity [of the decision to use auction or negotiation as a sale method in an M&A transaction] in stage 2, which is reported in Tables 4, 5 and 7. The IMR estimation results suggest that negotiations are a prevailing sale method in larger transactions and that firms in related industries are more likely to negotiate. The deals involving firms from the tech industry and target-initiated transactions are more likely to be conducted through auctions. Furthermore, the post-financial crisis period is also associated with a higher probability of auctions. Probit Regression - First Stage Results The dependent variable is pure negotiation = 1 and 0 otherwise. The first stage probit regression incorporates the variables that may influence the choice of the method of sale by the target from the outset of the transaction. DEALVALUE is a measure of the size of the deal, equal to the natural logarithm (ln) of deal value obtained from SDC; RELATED equals 1 if the target and acquirer are in the same two-digit SIC code and 0 otherwise; CRISIS equals 1 if the deal was announced between December 2007 and June 2009 and 0 otherwise; POSTCRISIS equals 1 if the deal was announced after June 2009 and 0 otherwise; TTECH equals 1 if the target is in a high-tech SIC and 0 otherwise; TLIQ is the natural logarithm (ln) of the target’s cash and marketable securities scaled by total assets ; TGROWTH is the target’s market-to-book ratio; TINIT equals 1 if the target initiated the sale and 0 otherwise; *, **, and *** indicate significance at 10%, 5%, and 1% levels, respectively. Variable

Probit Negotiations vs. Auctions -2.03 (0.000)*** 0.27 (0.000)*** 0.39 (0.017)** -0.29 (0.147) -0.49 (0.004)*** -0.34 (0.048)* 0.04 (0.899) -0.07 (0.274) -0.81 (0.000)*** 0.000*** 0.1582 445

Intercept DEALVALUE RELATED CRISIS POST CRISIS TTECH TLIQ TGROWTH TINIT Prob >F score R-Squared N

41