!Корпоративное право 2023-2024 / Bargeron L._Do Shareholder Tender Agreements Inform or Expropriate Shareholders

for the other shareholders. In addition, if the sweetheart bidder is not the highest value bidder, then the total surplus is smaller to begin with. Second, agency conflicts within the target firm are positively related to the probability of an STA. A manager in a firm with greater agency conflicts is more likely to expropriate shareholders. Assuming that greater non-management ownership concentration increases shareholder monitoring, and thus decreases agency conflicts, the expropriation hypothesis implies a negative relation between non-management ownership concentration and the probability of an STA.8

Furthermore, assuming greater management ownership aligns incentives and reduces agency conflicts, the expropriation hypothesis implies a negative relation between management ownership and the probability of an STA.9

Table 1 summarizes the predictions of the three hypotheses. The bargaining hypothesis distinguishes itself from the other two hypotheses by predicting higher premiums among STA offers. The predictions of the certification and expropriation hypotheses combined with an appropriate choice of proxies enable us to distinguish between these two hypotheses as well.

4. Data and descriptive statistics

For offers announced as of 1995, tender offer documents are available online through the SEC’s Edgar database.10 I therefore begin with the set of all tender offers,

8See Shleifer and Vishny (1997), Admati, Pfleiderer, and Zechner (1994), Maug (1998), Bolton and von Thadden(1998), and Pagano and Roell (1998) for discussions on the relation between ownership concentration and monitoring.

9Jensen and Meckling (1976) demonstrate that increasing management ownership decreases the agency conflicts within a firm. Morck, Shleifer, and Vishny (1988) and McConnell and Servaes (1990) find evidence that the relation between firm value and management ownership is non-monotonic. However, Demsetz and Lehn (1985) suggest that in equilibrium, because firms optimize, ownership structure is not related to firm value.

10Although some offers made prior to 1995 are available online, many are not. To avoid potential selection bias in these early offers, I include only offers from 1995 onward.

9

excluding self-tenders, from Thomson Financial’s SDC database over the seven-year period from 1995 to 2001 for target firms traded on the NYSE, AMEX, or NASDAQ. To ensure the sample includes only initial offers, offers for firms that are targets of a tender offer or merger by another bidder in the preceding 12 months are excluded. This selection process yields a sample of 780 initial tender offers.

For each of these offers, I collect the following tender offer data from SEC documents: details of an STA, the initial and final offer price, the bidder toehold, and the number of shares bid for, tendered, and accepted. Missing SEC documents result in the exclusion of 53 offers. Of the remaining 727 offers, 31 require that the tendering shareholders accept at least partial payment in the form of bidder stock. Because of documented differences between cash and stock offers, to ensure consistency within the data I exclude this small subsample of stock deals.11 In addition, 29 offers are lost due to insufficient data to calculate the target firm’s β. I also exclude offers that do not seek control, offers in which the target price prior to the announcement is less than $2 per share, and offers in which the bidder owns a toehold of over 40%.12 The resulting data set contains 538 cash tender offers.

As Fig. 1 documents, the percentage of offers including an STA is fairly stable throughout the sample period, varying from 46% in 1997 to 67% in 2001. However, the total number of tender offers varies widely across years, with only 42 offers in 2001 but 132 in 1999. These numbers are consistent with the general merger trend reported in

11Several studies investigate differences in the method of payment. Examples include Carleton et al. (1983) and Amihud, Lev, and Travlos (1990).

12I define an offer as one that seeks control if the bid is for at least a majority of shares or a merger agreement has been proposed by the bidder. Excluding target firms trading at less than $2 attempts to filter out deals with financially distressed targets; “fire sales” by distressed firms may differ from non-financially distressed deals. Changing the cutoff to $1 or $5 does not significantly affect the results. Deals in which the bidder already owns a controlling interest in the target firm also may differ fundamentally from noncontrolling bidder deals. In order to filter out these differences, deals in which the bidder owns over 40% of the target firm are excluded. Changing this cutoff to 30% or 20% does not significantly affect the results.

10

Andrade, Mitchell, and Stafford (2001), Holmstrom and Kaplan (2001), and RhodesKropf, Robinson, and Viswanathan (2005), namely, a merger wave building throughout the 1990s and declining after 1999, and suggest that STAs are prevalent throughout the merger cycle.

Additional, untabulated results indicate that industry, state of incorporation, and investment banking advisor do not determine the use of STAs. With respect to industry, 13 two-digit SIC code industries contain more than 10 target firms in the sample. In these 13 industries, the percentage of offers that include an STA range from a low of 42% in the Transportation Equipment industry to a high of 86% in the Rubber & Plastics industry. With respect to state of incorporation, 11 states are the state of incorporation for eight or more target firms in the sample. Excepting Minnesota, at least 40% of the offers in each of these states include an STA. Finally, with respect to investment bank advisors, 10 different investment banks served as the primary advisor to more than ten target firms in the sample. The percentage of deals that include an STA for these 10 banks range from a low of 39% for Bear Stearns to a high of 64% for JP Morgan. Thus, STAs are a prevalent feature of tender offers regardless of industry, state of incorporation, or investment banking advisor.

Table 2 documents the type of shareholders that sign STAs. The first cut, presented in Panel A, divides the sample STA offers into three mutually exclusive groups: offers with only officers and directors signing the STA, offers with no officers or directors signing the STA, and offers in which at least one officer or director and one non-officer and non-director sign the STA. I find that 57% of the STAs are signed exclusively by officers and directors, whereas 15% do not include an officer or director

11

announcement, RUNUP, is also included as a control variable to take the effects of potential information leakage into account.13

An alternative, frequently used measure of the target shareholders’ premium is the cumulative return on the target firm’s equity during an announcement period window. In the context of this study, however, short-term announcement period returns are inappropriate measures for premiums. Whenever a pre-negotiated contract affects the probability of a takeover, a short-term announcement return is a misleading measure of the contract’s relation to the firm’s premium. The announcement period return incorporates the probability of the deal’s success. This obscures the correlation between the contract and the premium generated by a successful deal. For example, if a contract increases the probability of a successful value-increasing takeover and the contract is correlated with lower takeover premiums, then short-term announcement period returns erroneously dampen the measurement of this negative correlation.

To test the hypotheses I define proxies for asymmetric information, ownership concentration, and agency conflicts. Specifically, the number of years since the target firm entered the CRSP monthly index, AGE of TARGET, proxies for asymmetric information.14 Ownership concentration is given by the ownership percentage of the largest non-management shareholder (based on proxy statements filed with the SEC),

13 Firms’ market betas are calculated using the six-month period ending 126 days prior to the announcement. The results are robust to excluding the runup period prior to the announcement as well as using a runup period of 5 days, 20 days, or 126 days. In prior literature, the length of runup windows varies across studies. For example, Comment and Schwert (1995) initiate their return window 1 day prior to the announcement; Bradley, Desai, and Kim (1988) initiate the window 5 days prior to the announcement; Billet and Ryngaert (1997) initiate the return window 20 days prior to the announcement; and Betton and Eckbo (2000) initiate the window 60 days prior to the announcement. Andrade et al. (2001) consider several windows and note that precision decreases as the window lengthens.

14 Lang (1991) finds that, due to decreased uncertainty, earnings response coefficients decrease with firm age. Wu (2004) also uses age as a proxy for asymmetric information.

13

TOP NON-O&D OWNERSHIP.15 Within the context of the expropriation hypothesis, ownership concentration measures shareholder monitoring of the firm’s management, and thus is an inverse measure of the extent of agency conflict within a firm. Within the context of the certification hypothesis, ownership concentration represents the willingness of the large shareholder to incur the costs of an STA. Finally, O&D OWNERSHIP, the cumulative ownership percentage of the officers and directors of the target firm (also collected from proxy statements filed with the SEC) is an alternative proxy for agency conflicts.16 The appropriateness of these three proxies is discussed further in Section 5.3.

I incorporate several other control variables commonly used in the takeover literature. These include the standard deviation of the target firm’s returns over the oneyear period ending one month prior to the announcement, STDEV of TARGET RETURNS, the log of the market value of target equity, LNSIZE, as well as five indicator variables:17

TERMINATION FEE, which equals one if a termination fee, payable to the bidder, is included in the offer and zero otherwise; LOCKUP OPTION, which indicates whether a lockup option is included; DIVERSIFYING, which equals one if the target and the bidder do not share the same four-digit primary SIC code; TOEHOLD, which indicates whether the bidder owns shares of the target firm prior to the offer; and MERGER AGREEMENT, which equals one if a merger agreement was approved prior to the offer.18

15The results are robust to two alternative definitions of ownership concentration as well, namely, the cumulative ownership percentage of the largest three non-management shareholders and the cumulative ownership percentage of all non-management 5% blockholders.

16Proxy statements for 53 of the target firms are unavailable. Therefore, the number of observations drops from 538 to 485 whenever TOP NON-O&D OWNERSHIP and O&D OWNERSHIP are included in the analysis. Applying this smaller sample to all analyses does not significantly affect the results.

17I also consider the target firm’s market-to-book ratio and debt-to-equity ratio. However, both are insignificant in all tests.

18Lang and Stulz (1994) and Berger and Ofek (1995) find that diversified firms trade at a discount. As a result, diversification could affect takeover premiums. However, as Campa and Kedia (2002) and Graham, Lemmon, and Wolf (2002) suggest, self-selection can explain at least a portion of this discount. Shleifer

14

Finally, I include four variables that assess the results of the tender offer. In particular, DURATION is the number of days from the announcement of the offer until the completion or withdrawal of the offer; COMPETITION is an indicator variable equal to one if, based on SDC data, another bidder announced a merger proposal or tender offer in the 12 months after the announcement of the initial deal; SUCCESS is an indicator that equals one if the initial bidder accepted the tendered shares; and BIDDER CAR is the cumulative abnormal return for the bidder’s equity over the three-day window centered on the announcement day.

The first four columns of Table 3 present descriptive statistics for the above variables. Notably, 60% of the offers contain an STA, 75% contain a termination fee, 8% include a lockup option, and only 16% of the bidders own a toehold. Further, 94% of the offers succeed while 6% face competition.19 The typical target is around seven years old and is significantly smaller than the bidder. The average bidder announcement return is small but significantly positive.

The last five columns of Table 3 compare STA offers to non-STA offers. The average bid premium is lower in STA offers for both initial and final bids. This difference is statistically significant, at the 1% level when the premium is calculated using final bids. The age of the target firm, an inverse measure of information asymmetry, is significantly lower in STA offers. The ownership of the largest non-management shareholder is greater in the STA group. Officers and directors of the target company have a higher ownership stake in STA offers. All of these results are consistent with the predictions of the certification theory. However, the premium evidence contradicts the

and Vishny (1986) model the effect of toeholds on tender offers. Excluding MERGER AGREEMENT as an independent variable does not significantly affect the results of the tests.

19 97% (476 of 492) of the offers that do not face competition succeed while only 48% (16 of 33) of the offers that face competition succeed.

15

bargaining hypothesis and the ownership evidence contradicts the expropriation hypothesis.

Note that the results also indicate that STA offers differ significantly along dimensions not addressed by the three hypotheses analyzed in this paper. For example, fewer STA bidders own a toehold but more have termination fees, lockup options, and merger agreements. In addition, the targets of STA offers are significantly smaller and have a greater standard deviation of returns than non-STA targets.

The outcome of the deals also differs significantly between STA and non-STA offers. STA offers face competing offers less frequently, conclude more quickly, and are more likely to succeed. Notably, however, the bidder’s cumulative abnormal return,

BIDDER CAR, is not significantly different across STA and non-STA offers.20

In summary, the univariate statistics suggest that STA offers differ significantly from non-STA offers in terms of target firm characteristics, the specifics of the offer itself, and the path the deal takes subsequent to the initial announcement, with the evidence supporting the empirical predictions of the certification theory.

5. Hypothesis tests

This section extends the investigation of STAs to a multivariate context. Two components of the tender offer decision are considered. The parties must decide if including an STA is optimal and they must determine the premium level to offer tendering shareholders. I model and estimate these decisions under two sets of assumptions. First, assuming the two decisions are made independently, I estimate the STA decision using a probit analysis and the premium decision using ordinary least

20 The bidder CAR evidence suggests that STAs are not simply a result of stronger bidders forcing an STA on targets. Unreported results from regression analysis of bidder CARs reinforce this evidence.

16

squares (OLS). Second, assuming the STA decision affects the premium decision, I estimate the system jointly using a Heckman two-step procedure. Appendix C presents details of the model.

5.1. OLS estimation of premium

The univariate analysis suggests that the average premium is lower in tender offers including an STA, supporting a key prediction of both the certification and the expropriation hypotheses but contradicting the bargaining hypothesis. Table 4 offers additional evidence in support of the negative relation between STAs and premiums. The table reports results from an OLS estimation of the premium decision. In models 1 and 2 the BID PREMIUM is calculated using the final bid price; models 3 and 4 use the initial bid price. To determine whether termination fees and lockup options are driving the results, I omit them from models 1 and 3 but include them in models 2 and 4.

The results are robust to the alternative definitions of the bid premium. All four specifications yield significantly negative coefficients for STA. The magnitude of the coefficients (5.9 to 7.5 percentage points) is economically significant as well, indicating that premiums in STA offers are approximately 20% lower than premiums in non-STA offers. In addition, runup is negatively related to target premiums. Approximately 25% of the runup is subsequently discounted from the bid premium.

The coefficients on termination fee and lockup option are positive, although termination fee is not statistically significant.21 For each measure of bid premium the inclusion of termination fees and lockup options increases the magnitude of the negative

21 The magnitude of the coefficient on lockup options is in line with the results in Burch (2001). The estimates for termination fees of 0.3% to 1.1% are lower than the 3% to 7% estimates in Officer (2003) and Bates and Lemmon (2003). This could result from the different samples studied, as their samples include both mergers and tender offers. In addition, the premiums in their larger sample studies rely on data from SDC while the premiums in this study rely on bid data collected directly from the SEC documents.

17

coefficient on STA. Given the positive correlation between STAs and both termination fees and lockup options, this result is not surprising. Prior studies find that target premiums are higher when either termination fees or lockup options are included in an offer. Including these contracts as control variables disentangles their positive effect on premiums from the negative relation between STAs and premiums, thus increasing the magnitude of the negative coefficient on STA.

5.2. STA estimation

Models 1 and 2 in Table 5 present the estimates from the probit analysis of the STA decision. The coefficient on AGE of TARGET is negative and significant in both specifications, indicating that the probability of an STA decreases as information asymmetry decreases. The effect is economically significant as well. A one-standard deviation increase in the age of the target firm is associated with in a nine-percentage point decrease in the probability of an STA. This result supports the prediction of the certification hypothesis.

The significantly positive coefficients on TOP NON-O&D OWNERSHIP and

O&D OWNERSHIP both imply that ownership concentration is positively related to the probability of an STA. The marginal effects suggest that a one-standard deviation increase in the non-O&D measure of ownership is associated with a 22-percentage point increase in the probability of an STA, while a one-standard deviation increase in O&D ownership yields a 33-percentage point increase in the probability of an STA. These results support the certification hypothesis. However, in conflict with the expropriation hypothesis, they indicate that decreased agency conflicts do not decrease the probability of an STA.

18