The Symmetry and Asymmetry of Bidder and Target Termination Fees in Acquisitions

2.1 Legal Background

Termination provisions outline both parties’ rights and obligations involved in a merger or acquisition if the conditions necessary for closing the deal cannot be met or if the agreement is outright terminated by one side (Afsharipour, 2010). Such provisions regularly contain statements regarding termination or break-up fees. Termination fees serve as compensation for one side in case of premature cancellation of the agreement by the opposing side (Sneirson, 2002). They can be employed on both the bidder and the target side. The target pays target termination fees (alternatively: “termination fees”) to the bidder if the target evokes the reasons for the termination of the agreement, while bidder termination fees (or: “reverse termination fees”) oblige the bidder to pay a specified amount to the target if the bidder causes the termination of the deal.

Similar to a material adverse event or change clause that grant the right to terminate the deal in case of specified events (Denis & Macias, 2013), termination fees are generally employed in the period between the signing of the agreement and the closure of the deal (Afsharipour, 2010). Their primary motivation is to deter third-party bids by making alternative proposals costlier and the target firm less financially enticing (Sneirson, 2002). As the deterrence of third-party bidders may hinder the target from receiving more advantageous bids, the inclusion of such deal protection devices has been criticized and increasingly scrutinized in court as a potential violation of fiduciary duty (Choi, 2020).

Contractually, deal protection devices can be split into performance-related or damage-related provisions (Sneirson, 2002). In performance-related deal protection provisions, a party is required to engage or not engage in certain pre-specified actions. For example, during its acquisition by the Dutch-American Chicago Bridge & Iron Company (CB&I) the Shaw Group’s board agreed to use “reasonable best efforts” to obtain shareholder approval for the acquisition agreement (SEC, 2012). Another performance-related provision required the Shaw Group, a construction company for nuclear power plants, to actively assist CB&I in receiving the necessary regulatory approval (SEC, 2012). The Shaw Group furthermore agreed to refrain from specific actions. For example, the Shaw Group’s board was obligated by the agreement to “conduct business in the ordinary course of business with past practice” (SEC, 2012), thus refraining the board from breaking up the business organization or selling the assets of the firm. The agreement also included a so-called “no-shop” provision that legally bound the board of the Shaw Group to “immediately cease any solicitation, knowing encouragement, discussions or negotiations with any person that may be ongoing with respect to a takeover proposal” (SEC, 2012).^[7]

Deal protection devices that are damage-related effectively form contractual liquidated damage clauses (Sneirson, 2002). In mergers and acquisitions, liquidated damage provisions reimburse one party if the opposing party breaches an obligation in the agreement. In case of termination fees, this reimbursement constitutes a specified sum, and the breached duty are specific conditions that lead to premature termination of the deal.^[8] In the acquisition of the Shaw Group by CB&I, various conditions that could trigger the termination fees’ payment were outlined in the transaction agreement (SEC, 2012). For example, the Shaw Group had to reimburse CB&I if the target board withdrew its recommendation for shareholders to vote in favor of the proposed transaction, leading the bidder to terminate the transaction before the shareholder vote. Furthermore, CB&I was entitled to the termination fee if the Shaw Group terminated the transaction to solicit, negotiate, or accept alternative bids. Additionally, both CB&I and the Shaw Group had to reimburse the opposing side if approval from its respective shareholders could not be obtained, resulting in the non-consummation of the acquisition.

The example of the transaction agreement between CB&I and the Shaw Group shows how difficult it is to determine whether a given provision is performance-related or damage-related, as there may be considerable overlap. This distinction is nevertheless crucial because reimbursements for alleged non-performance of the opposing side are, in general, legally less enforceable than payments based on liquidated damages (Solórzano, 2009). Furthermore, there may be additional limitations concerning the size of termination fees. A high termination fee may, for instance, substantially limit the board’s ability to pursue outside opportunities, thus obstructing the director’s fiduciary duty to act in the shareholders’ best interest (Afsharipour, 2010). This is because the fee effectively reduces the target’s value and hence makes the potential acquisition by alternative bidders comparatively costlier.

While there is no universally defined acceptable size of termination fees, their levels have increased steadily over the past years as they have gained popularity as a useful deal protection device (Choi & Wickelgren, 2019; Restrepo & Subramanian, 2017). Nevertheless, numerous legal challenges have forced the courts to consider what constitutes a “reasonable” size for termination fees. For example, before 2011, when the UK banned the use of all deal protection devices, the size of termination fees was limited to 1% of the deal value (Restrepo & Subramanian, 2017). Similarly, the Australian Takeovers Panel provides guidance that sets the generally acceptable size of termination fees at not exceeding 1%. Crucially, these limitations stipulated by the UK and the Australian Takeover regulations generally only apply to target termination fees (Restrepo & Subramanian, 2017). Bidder termination fees are, therefore, not affected by these restrictions limiting the use or size.

In contrast, there is no regulation in the United States and Canada as to the upper limit of termination fee sizes. However, based on the prevailing legal opinion and prior legal challenges, the acceptable range for termination fees can be estimated to be around 1–5% of the deal value (Jeon & Ligon, 2011).^[9] Comparably, in Chinese takeover regulations, there is also no upper limit for target or bidder termination fees. The rather intransparent regulatory and governmental transaction approval process by Chinese authorities seems to induce some Chinese firms to offer termination fees that exceed 10% of the deal value (Klingsberg, 2016). As these regulatory uncertainties threaten the consummation of the transaction, firms apparently use substantial termination fees to reallocate deal risk.

A similar effect can be observed in cross-border deals in which national security concerns constitute an additional and often significant barrier for deal consummation. For example, Chinese investments in the United States have been increasingly scrutinized by CFIUS (Committee on Foreign Investments in the US) that evaluates the effects of selected transactions involving a foreign party on US national security. This, in turn, led some Chinese bidders for US targets to offer substantially higher reverse termination fees than usually observed to alleviate the risks stemming from these additional regulatory uncertainties (Reuters, 2016).

2.2 Termination Fee Sizes

Termination fees as a deal protection device have been given considerable attention in the academic literature (Jeon & Lee, 2014). Researchers consent that termination fees can be an efficient device to overcome contracting problems between the bidder and the target (see, e.g., Bates & Lemmon, 2003; Chapple, Christensen, & Clarkson, 2007; Officer, 2003). Officer (2003) states that termination fees are an efficient compensation device incentivizing the disclosure of private information such as post-acquisition plans in the acquisition negotiation. This is because the bidder knows that a termination fee will be granted if an offer by a third party, who may free-ride on the information contained in the initial bid, is accepted instead. As the ensuing stronger information sharing between the initial bidder and the target may help convince the target’s shareholders of the transaction’s benefits, deal completion becomes more likely if such a target termination fee is offered (Officer, 2003). Bates and Lemmon (2003) further support this efficient contracting hypothesis by showing that target termination fees are more frequently used in complex deals with a high degree of information asymmetry between the bidder and target. They conclude that target termination fees insure against higher negotiation costs and higher risk of bid failure.

On the contrary, Rosenkranz & Weitzel, 2013 describe theoretically that termination fees may also result from the parties’ relative bargaining positions in sequential negotiations with endogenous outside options. They explain that termination fees decrease in a party’s bargaining power as there is less need for this party to signal a commitment to the acquisition. Thus, this bargaining power hypothesis assigns a role to both the bidder and the target in offering termination fees, depending on which party has most to lose from the acquisition failing. In this respect, the argument is related to Chapple et al. (2007), who explain that a bidder needs a strong negotiation position to request a target termination fee inclusion in the first place.

Traditionally, scholars in the law-and-economics field aligned more strongly with the efficient contracting hypothesis. The predominant position in the field postulates that while the distribution of the price terms in a given contract is influenced by the respective bargaining position of buyer and seller, the non-price terms (i.e., the contract design) are efficiently determined by the involved parties independent of the relative bargaining position.^[10] In that view, one-sided non-price terms, e.g., termination fees, are not the result of inequality in bargaining power but indeed the “efficient” terms that both parties agreed to. This “irrelevance proposition”^[11] (Choi & Triantis, 2012) of bargaining power on contract design argues that the involved parties will agree to non-price terms that maximize the joint net payoff of a transaction and use the price term instead to distribute this payoff (or “surplus”) proportionally to the bargaining position of both sides. Following this rationale, one-sided non-price terms are efficient and agreed to by both sides if the benefit of the addition of these terms for one side is larger than the cost for the opposing side. According to the irrelevancy proposition, bargaining power only determines the partition of the joint payoff and not the structure of the efficient terms that both sides agree to.^[12]

However, during the last decade, the irrelevance proposition has come under increased scrutiny in the law-and-economics field. Practitioners have regularly noted the importance of bargaining power to explain variations in contract design and thus put the central assumption of the proposition in question. This practitioners’ perspective on the relationship between bargaining power and contract design is further corroborated by the model stock purchase agreement that the American Bar Association (ABA) published as a guideline for designing the first draft of a stock acquisition agreement. In its commentary, the ABA points out the substantial influence the bargaining power of the respective parties may wield on the specific contract terms. In detail, the ABA notes that with an increase in a target’s outside opportunities, the terms illustrated in the model agreement might be too bidder-friendly. In contrast, if the target’s bargaining position is weak, e.g., because of financial difficulties, the ABA acknowledges that the outlined terms could potentially shift even further in favor of the bidder in an actual acquisition agreement (American Bar Association, 2010). Consequently, this discrepancy between theory and practice has led to an emerging stream of theoretical work that modifies or outright challenges the irrelevancy proposition.

Choi and Triantis (2012) show theoretically how bargaining power can potentially influence non-price terms in various situations when the implicit assumptions of the irrelevancy proposition are loosened to better reflect real-world conditions. These assumptions of the irrelevancy proposition include no transaction costs and perfect information of all parties involved. In deals with information asymmetries, the unequal distribution of bargaining power incentivizes non-price terms as tools of screening and signaling to mitigate the potential for adverse selection and moral hazard instead of maximizing the joint payoff. Choi and Triantis (2012) demonstrate this via the example of adverse selection in the credit market. In a credit market where the bargaining power is skewed towards the lender (e.g., because of market conditions that reduce the availability of loans – like a credit crunch) the interest rate, i.e., the price term of a loan contract, is expected to rise. In turn, the higher interest rate shrinks the pool of low-risk borrowers and incentivizes the remaining (higher risk) borrowers to take more risks to compensate for the higher interest burden. As the lender cannot determine the risk-level of all remaining borrowers in the absence of perfect information, the lender can reduce the risk-level of the pool of borrowers by reducing interest rates and thus enticing low-risk borrowers back into the credit market. Alternatively, however, the lender can also utilize non-price terms, like collateral, warrants or covenants, as screening devices to filter out high-risk borrowers and reduce the risk-level of the pool of potential borrowers in the process. Following this rationale, the bargaining positions of both parties influence not just the price terms, but also the structure of the non-price terms to reduce the risk of adverse selection.

Additionally, the typical negotiation structure of acquisitions constitutes another potential avenue for bargaining power influencing non-price terms (Choi & Triantis, 2012). Corporate acquisitions frequently follow a multiple-stage model of negotiations where the price is determined first. At the same time, various non-price terms, including termination provisions, are negotiated in the second stage (Freund, 1975). As the price is already settled and thus unchangeable^[13] in the second stage, the negotiations regarding the non-price terms become distributional. This is because there is little incentive for either party to agree to the “efficient” terms, i.e., the terms that maximize the joint value of the transaction, when the primary way of obtaining this value, i.e., the price, is not adjustable anymore. Consequently, both parties may try to utilize their respective bargaining positions in this second stage to negotiate non-price terms that create the most value for themselves instead of agreeing to the terms that would maximize the joint surplus of the transaction (Williams, 2017).

In summary, the proposition that bargaining power only influences the price term of a contract but not the non-price terms appears dubious if information asymmetries and transaction costs are included in the theoretical considerations. Furthermore, the two-stage negotiation structure typical in corporate acquisitions constitutes another challenge for this “irrelevance proposition”. This is because it incentivizes the involved parties to utilize their bargaining position to negotiate the most one-sided non-price terms possible that potentially differ from the value-maximizing “efficient” terms.

Even after the theoretical challenges to the irrelevance proposition posed by Choi and Triantis (2012), surprisingly few studies tried to examine the validity of the proposition empirically with overall ambiguous results.^[14]Williams (2017) examines the influence of bargaining power on venture financing contracts and provides empirical evidence against the irrelevance proposition. Specifically, Williams (2017) finds that the bargaining position the respective parties possess, as proxied by the total supply of venture capital in the U.S., influences price and non-price terms of venture financing contracts. Badawi and de Fontenay (2019) investigate a possible first-drafter advantage on merger contract design. They follow the practitioners’ view that the party that provides the first draft of a merger contract possesses an advantageous bargaining position. By doing so, they find evidence that a first-draft advantage exists for difficult to monetize non-price terms, like go-shop provisions, while finding no such relationship for both types of termination fees.

We expand these previous studies to examine the influence of bargaining power on the symmetric and asymmetric structure of termination fees in majority acquisitions. Following Choi & Triantis, 2012; Rosenkranz & Weitzel, 2013, we assign the better bargaining position to the party that has less to lose from the acquisition failing based on related firm and situational characteristics. We do so by identifying bidder and target characteristics that likely relate to the specific bargaining position of the respective deal parties instead of using broader market trends. We use termination fees to test the influence of bargaining power on contract design as they provide a prime example of a non-price term that, due to its clearly monetizable payoffs, will likely be subject of negotiations by the parties themselves and not referred to the legal counsel (Badawi & de Fontenay 2019). Furthermore, termination fees can amount to a substantial percentage of the deal value^[15] and are thus likely considered an important contract term and an area of extensive negotiations in corporate acquisitions.

We examine target and bidder termination fees conjointly, i.e., the symmetric or asymmetric structure of both fees, because there are also negotiated conjointly. Thus, the size of bidder termination fees may influence the size of target termination fees (and vice versa). Trying to evaluate the determinants (e.g., bargaining power) of bidder and target termination fees in isolation ignores a potential critical influence on the size of said termination fees. Examining the asymmetric structure does not just implicitly include the size of each termination fee but also takes the reciprocal effect of termination fee negotiations into account. Consequently, this approach provides a better reflection of the bargaining process both parties are engaged in to determine the size of the respective termination fees.

Based on the theoretical derivation outlined previously and building on the existing empirical evidence, we formulate the following research question for our empirical analysis:

2.3 Bidder and Target Characteristics and their Association with Symmetric and Asymmetric Termination Fees

To answer our research question, we will derive immediately testable hypotheses that relate bidder and target characteristics and situational deal characteristics with the two parties’ relative bargaining position. We postulate that any characteristic that reduces a party’s bargaining position – i.e., that increases its commitment to the deal – raises the termination fee that this party offers in the negotiation process. Accordingly, if the bargaining power is evenly distributed, no side gains an advantage and termination provisions should be symmetric. These symmetric termination fees result either by both parties agreeing to termination fees of the same size or by no party being able to negotiate the inclusion of termination fees for the opposing side in the first place. By not including termination fees for the opposing side a deal party can disadvantageously affect its own position in a given deal. For example, the initial acquisition terms of the proposed takeover of Sun Microsystems by IBM in 2009 included no termination fees. This non-existence of bidder termination fees allowed IBM to terminate the deal after price renegotiations stalled without incurring a financial penalty. Just a month later, the withdrawal of IBM and, subsequently, the lack of competing bids prompted Sun Microsystems to accept a much lower takeover price offered by Oracle (Butler & Sauska, 2014). In this case, the inclusion of bidder termination fees could have increased the commitment of the initial bidder, IBM, to refrain from renegotiating price terms or, at the very least, compensated Sun Microsystems financially for the bidder induced termination of the deal. On the other hand, a lack of target termination fees in acquisitions can disadvantageously affect the bidder’s position as it incentivizes the target – due to a lack of financial repercussions and in hope of a better offer – to encourage and consider competing bids. In conclusion, the side with the greater relative bargaining power in a deal will therefore likely use their advantageous bargaining position to negotiate the inclusion of termination fees for the opposing side and thus a more favorable termination fee structure for itself. Consequently, the non-inclusion of termination fees for the opposing deal party by either side may hint at a balanced distribution of bargaining power as both parties agree to relinquish a potentially advantageous deal position by not including termination fees. From this bargaining power perspective, we thus consider the situation where no side is able to negotiate the inclusion of termination fees for the opposing side to be an implicit symmetrical termination fee structure. We identify the following three situational characteristics that we associate with an uneven (even) distribution of bargaining power increasing the likelihood of asymmetric (symmetric) termination fees.

3.1 Data Source and Sample Selection

Our empirical analysis is based on a sample of secondary data from the Mergermarket database. The sample selection process is displayed in Table 1. The final sample consists of 25,026 majority acquisitions from 2012 to 2015.^[17] We choose the time period to avoid interference with any late effects of the financial crisis. Indeed, a sharp drop in M&A activity was observed in the years following the financial crisis 2007/2008 (Grave, Vardiabasis, & Yavas, 2012). From the initial dataset of 32,402 M&A deals, we exclude 4276 minority deals, 68 mergers, and four deals with implausible information. We follow Officer (2003) and consider only those majority acquisitions in which the bidder aims to control at least 50% of the target. Implausible deals include one acquisition that was reported twice and two deals whose termination fees were reported at implausibly high levels (1.575 and 55%). As a reference, Officer (2003) reports the average target termination fee to be 3.80% of the total deal value and Bates and Lemmon (2003) indicate it to be between 2.70% and 3.20%. Further 3028 deals are excluded due to missing data for our independent variables. We follow common research practice and include both completed and withdrawn deals in our sample (e.g., Bates & Lemmon, 2003; Boone & Mulherin, 2007; Officer, 2003).

3.2 Operationalization of Variables

The dependent variables in our analyses are based on the termination fee structure. We first create an indicator variable to distinguish between symmetric termination fees and asymmetric termination fees in a logistic regression analysis. We create a new dependent variable in a second step where we differentiate between cases where the target or bidder fee is higher in a multinomial regression approach. Accordingly, we assign a value of zero for a symmetric fee structure, i.e., if target and bidder termination fees are of equal size. For our logistic regression, a value of one in our dependent variable is assigned for all asymmetric termination fees (regardless of the target or the bidder fee being higher). For the multinomial logistic regression, a value of one characterizes an asymmetric fee structure where the target termination fee is higher, and a value of two is assigned for an asymmetric fee structure where the bidder termination fee is higher. Categories 1 and 2 hence refer to the two types of asymmetric termination fees.

To test hypothesis H1, we focus on the independent dichotomous variable INVESTOR, which takes the value one for all deals where the bidder is an institutional investor on a primary market and zero otherwise. To compute this variable, we build on the Mergermarket’s deal classification and identify all deals in which an institutional bidder either acquires a majority stake in the target company (Institutional Buy-In) or fully acquires the target (Institutional Buy-Out). H2 relates the termination fee structure to a target being insolvent. To compute the dichotomous variable INSOLVENCY, we identify all deals marked as a sale under insolvency by Mergermarket. H3 refers to buyouts in the secondary market. Based on the Mergermarket classification, we identify all secondary buy-out deals and code the dichotomous variable SECONDARY accordingly.

Our model includes several control variables: deal characteristics and target/bidder characteristics, time, and industry effects. First, we control for deal size, following prior studies that investigated the effect of either the target’s equity value (e.g., Bates & Lemmon, 2003; Officer, 2003) or the deal value (e.g., Boone & Mulherin, 2007; Coates & Subramanian, 2000) on termination fees. Except for Boone and Mulherin (2007), who find no relationship between deal size and termination fees, all other studies find a positive relationship (Boone & Mulherin, 2007). We define LN_DEAL VALUE as the acquired equity stake plus 100% of the target’s net debt. Including 100% of the net debt is reasonable as the sample includes only majority stake acquisitions.

Second, prior studies provide evidence that cross-border deals can affect the usage of termination fees. Very and Schweiger (2001) argue that cultural and language differences might place an additional burden on collecting the necessary information to evaluate the benefits and risks of acquisition in cross-border deals. Further, country-specific tax and governance systems can cause additional problems in an acquisition process (Erel, Liao, & Weisbach, 2012). Rossi and Volpin (2004) indicate that cross-border deals heighten information asymmetries and therefore influence the use of deal protection devices such as termination fees. Additionally, as abnormal returns are higher in cross-border than in domestic deals (Danbolt & Maciver, 2012), increasing the likelihood of deal closure via termination fees becomes even more critical. Consequently, we add CROSS BORDER equaling one for cross-border and zero for domestic deals.

Third, we include the dummy variable AUCTION as a proxy for takeover competition (yes = 1, no = 0) to control for potential effects on the termination fee structure. The variable AUCTION is restricted to public auctions, which describes an auction process triggered by the announcement of the seller or target that the firm is for sale. The following auction process generates the takeover competition, and the firm is usually sold to the highest bidder. Whereas Coates and Subramanian (2000) and Officer (2003) find a negative effect of takeover competition on the usage of termination fees, Boone and Mulherin (2007) find a positive effect.

Fourth, the dichotomous variable PUBLIC is included to control for pending shareholder approval that potentially reduces the bidder’s or target’s bargaining power and thus results in deals where asymmetric termination fees are more likely. The variable PUBLIC is coded as one if shareholder approval from any acquisition party is needed and zero otherwise.

Fifth, we control for management buyouts (MBO), where the acquiring management is likely to possess superior information compared to other bidders. To benefit from this, the management needs to make sure that the deal is successful. To protect its special bidder status, the management is likely to include deal protection devices such as termination fees in MBOs (Povel & Singh, 2006). We therefore add the dummy variable MBO to our list of controls.

Sixth, academic literature has shown that higher deal completion rates can be associated with termination fees (Bates & Lemmon, 2003). Officer (2003) finds that deals with target termination fees are 20% more likely to be completed. Therefore, we also control for the effect of terminated or withdrawn deals on symmetric and asymmetric termination fees. Consequently, we control for variable COMPLETE, which equals one for completed acquisitions and zero for withdrawn acquisitions.

Seventh, we control for differences between the Anglo-Saxon common law and continental European civil law system, which appear to be most relevant concerning aspects of investor protection (La Porta, Lopez-de-Silanes, Shleifer, & Vishny, 2000) and regarding other capital market characteristics (Cernat, 2004). Furthermore, La Porta, Lopez-de-Silanes, Shleifer, and Vishny (2000) show that cross-border acquisitions’ efficiency increases when the bidder is located in a common-law country. Accordingly, we use the variable EUROPE that indicates whether a bidder is located in Europe (1 = yes, 0 = no), where we consider all continental European nations: Austria, Belgium, Bulgaria, Croatia, Czech Republic, Denmark, Finland, France, Germany, Greece, Hungary, Italy, Latvia, Liechtenstein, Lithuania, Luxembourg, Malta, Monaco, Netherlands, Norway, Poland, Portugal, Romania, Serbia, Slovakia, Slovenia, Spain, Sweden, Switzerland, and Ukraine. The indicator variable ANGLO, in contrast, refers to a bidder location in Anglo-Saxon countries consisting of the USA, Canada, Great Britain, and Ireland (1 = yes, 0 = no). Here, we follow Gugler and Yurtoglu (2004) and include Great Britain and Ireland in the Anglo-Saxon dummy variable.

Eighth, we control for the payment method and deal premium paid by the bidder as these constitute important deal characteristics (Che & Lewis, 2007) that could potentially influence the structure of termination fees. The variable CASH equals one if the deal is financed entirely in cash (i.e., without an equity component) and zero otherwise. Correspondingly, the variable EQUITY equals one if the deal is financed entirely with equity (i.e., without a cash component) and zero otherwise. DEAL PREMIUM is calculated as a percentage difference between the price paid for the target and the target stock price one-day and one-month before the announcement.

Ninth, we include several deal characteristics that capture the deal’s competitive situation as third-party bidders offer outside opportunities for targets that in turn provide them with more leverage in the bargaining process. In detail, CONTESTED HOSTILE equals one if there is more than one bidder and the transaction bid is hostile and zero otherwise. CONTESTED RECOMMEND, on the other hand, equals one, if there is more than one bidder and the board of the target has recommended the transaction bid. Lastly, UNSOLICITED HOSTILE equals one if the board of the target has not recommended the deal at the announcement, and there is only one bidder and zero otherwise.

We also control for industry effects as the industry can influence the bargaining power if industries are dependent on each other (Ahern, 2012) and follow common practice by controlling for time effects (André et al., 2007; Bates & Lemmon, 2003).

Finally, to accommodate our international sample’s economic, regulatory, and institutional variety, we include numerous macroeconomic and governance-related variables to control for these differences between countries. A detailed list and description of these variables is provided in the Appendix.

3.3 Sample Description

Table 2 Panel A shows the annual distribution of the number of transactions and the deal values and the employment of target and bidder termination fees. The sample contains a total number of 1184 target termination fees and 734 bidder termination fees. On average, 4.73% of all transactions in our sample use a target termination fee, whereas 2.93% employ a bidder termination fee. Furthermore, 601 deals (or 2.40% of all transactions) employ both, target and bidder termination fees, at the same time. Similarly to Bates and Lemmon (2003), we find that the deal value of acquisitions with termination fees is higher than the value of those without. On average, 22.77% of the total deal value is tied to a target termination fee, while 15.59% of the deal value makes use of a bidder termination fee.^[18] In accordance with Officer (2003) and Bates and Lemmon (2003), target termination fees are hence employed more frequently and in larger deals than bidder termination fees in our sample. Interestingly, bidder termination fees appear to have become more common over time as the ratio of employment of bidder to target termination fees is 0.62 in our sample while it was 0.26 in the study by Officer (2003) which covered the earlier time period from 1988 to 2000.

Table 2 Panel B shows the annual distribution of symmetric and asymmetric termination fees in our sample. 96.37% (n = 25,026) of the acquisitions in our sample show implicit or explicit symmetric bidder and target termination fees.^[19] Symmetric termination fees account for only 80.60% of the total deal value, however. Asymmetric fee structures are hence predominantly employed in comparatively larger acquisitions. Deals with higher target than bidder termination fees are more than twice as often stipulated (2.52% of all deals, n = 631) than deals featuring higher bidder than target termination fees (1.11% of all deals, n = 277). Concerning transaction sizes, the former account for 11.40%, the latter for 8.00% of the total deal value in the sample. Though target termination fees hence still seem to dominate, our data show that asymmetric fee structures with higher bidder termination fees do not lag much behind in the larger acquisitions. This clearly corroborates the importance of studying the determinants of asymmetric termination fee structures.

Table 2 Panel C displays the distribution of termination fees by target industry. Symmetric termination fees are most common in the agriculture sector and the service industry. In contrast, asymmetric termination fees are most prevalent in the mining sector and tie to the high-tech sector’s largest deal values. More precisely, contracts with higher target termination fees account for 17.70% of the total deal value in the high-tech industry, whereas contracts with higher bidder termination fees account for 13.89% of the total deal value in this industry. Taken together, asymmetric termination fee structures appear to be important contracting devices in this industry.

Table 2 Panel D displays a breakdown of termination fee structures by the bidder’s country. Most bidders in our sample are located in the United States, followed by China, the United Kingdom, and Canada. Interestingly, 6.23% of deals with a US bidder possess asymmetric termination fees, while the same is only observed in 2.13% of deals with a Chinese bidder. There is a relatively even distribution of asymmetric termination fee structures for deals with a UK bidder. In contrast, the most common asymmetric termination fee structures in the United States feature higher target than bidder termination fees. A similar observation can be made for Canada and China.

Table 2 Panel E shows the breakdown of symmetric and asymmetric termination fees by the target’s country. Most targets are located in the United States, followed by China, the United Kingdom, and Canada. There are relatively few deals with asymmetric termination fees with targets in the United Kingdom. This is not surprising as termination provisions have been prohibited as deal protection devices in the UK since 2011 (Restrepo & Subramanian, 2017). As might have been expected, we observe that deals involving US targets show higher target than bidder termination fees commonly, as most deals with a US target also involve a US bidder. The same can be said for Canada, while symmetric termination fees again dominate deals involving a Chinese target.

Table 3 gives a brief overview of our dataset. Panel A shows descriptive statistics for the independent variables to be used in our analyses. We observe that in 5.27% of the deals, the bidder is an institutional investor (INVESTOR) whereas sales under insolvency (INSOLVENCY) have a prevalence of 2.41%. Secondary buyouts (SECONDARY) amount to 2.64% of all transactions. We find the average logarithmic deal value (LN_DEAL VALUE) to be 3.78 million EUR regarding continuously distributed control variables.^[20] In 19.38% of all deals, shareholder approval (PUBLIC) was required from at least one deal party. The data shows furthermore that just under half (49.41%) of all bidders are located in Anglo-Saxon (ANGLO) countries, whereas only 23.35% of bidders come from Europe (EUROPE). Altogether 40.33% of the deals are cross-border (CROSS BORDER) transactions. With 96.75%, the clear majority of deals have been completed (COMPLETE), while 2.53% of the deals are Management Buyouts (MBO), and 3.34% of the targets are sold in a public auction (AUCTION).

Table 3 Panel B shows the pairwise correlation matrix for our independent variables for our full sample that includes deals without termination fees. The findings in the correlation matrix suggest that multicollinearity does not affect our multivariate analysis. We find only one case with a significant positive correlation higher than 0.4 as MBO and INSOLVENCY show a significant correlation of 0.409. This correlation appears reasonable as Peel (1995) indicates that MBOs can be an effective method to sustain insolvent targets that would otherwise face liquidation. Unsurprisingly, an additional negative correlation of −0.593 between both dummy variables that capture the payment method (CASH and EQUITY) is observable, pointing to the fact that a number of deals are financed entirely in cash or entirely in equity.

Furthermore, there is a positive correlation of 0.183 between EUROPE and CROSS BORDER and a weaker negative correlation of −0.102 between ANGLO and CROSS BORDER. Deals that include a bidder from continental Europe are thus more likely to be cross-border deals than deals where the bidder is set in the Anglosphere. This is likely explained by the strong economic integration of the countries that are part of the European Union which includes in most cases a common currency, making cross-border transactions less complex and therefore more appealing for both sides. Finally, there is a positive correlation between EQUITY and shareholder approval (PUBLIC) and a negative correlation between CASH and PUBLIC (0.305 and −0.331, respectively). This is reasonable as, for example, in the US, deals that are fully financed in cash are considered a board issue and thus generally do not require shareholder approval.

3.4 Research Methodology

We test our hypotheses in multivariate analyses where the dependent variable enters in categorical form. We estimate a logistic regression and a multinomial logit regression model to test our hypotheses. By doing so, we are largely in line with previous studies.^[21]Officer (2003), Bates and Lemmon (2003), Boone and Mulherin (2007), and Jeon and Ligon (2011) use logistic, multinomial logit, or the related probit model^[22] to examine the determinants of target and/or bidder termination fees. André et al. (2007) and Badawi and de Fontenay (2019) apply a multivariate linear regression using ordinary least squares (OLS) which would not be appropriate for our setup, as we a less interested in the absolute size of termination fees but the relative structure of such fees measured by our categorical dependent variable with a maximum of three outcomes. For the logit regression, our dependent variable can take on one of two categories: symmetric termination fees (SYM) and asymmetric termination fees (ASYM). The multinomial logit regression, in contrast, allows us to consider more than two categories (Wooldridge, 2010). As a reference category for each of the logits, we employ deals with a symmetric termination fee structure. We hence compare the reference category SYM to asymmetric deals with a higher target than bidder termination fee (ASYM1) and asymmetric deals with a higher bidder than target termination fee (ASYM2). The variable coefficients thus display the likelihood of both types of termination fees compared to our base outcome – i.e., symmetrically structured termination fees. We use the maximum likelihood procedure to estimate both the classical logit regression and the multinomial logit regression:

where (πiπSYM) is the coefficient of the ith category of the dependent variable in comparison to symmetric termination fees as the reference category.

4.1 Multivariate Results

The logistic regression results from a test of the likelihood of asymmetric versus symmetric termination fees are presented in Table 4 Panel A. Examining our control variables, we first find that higher deal values (LN_DEAL VALUE) are positively associated with asymmetric termination fees. Furthermore, the positive and highly significant coefficient of PUBLIC points to shareholder approval being an important positive factor of asymmetric termination fees. This is reasonable as pending shareholder approval puts the consummation of a transaction agreement in doubt and thus erodes the party’s bargaining position that still requires its shareholders’ approval to proceed with the deal. Finally, we also find regional determinants for the likelihood of symmetric and asymmetric termination fees. Deals that involve a bidder located on the European continent (EUROPE) are more likely to exhibit symmetric termination fees. In contrast, the structure of termination fees is more likely to be asymmetric in deals with a bidder from the United States, Canada, Great Britain, or Ireland (ANGLO).

Our theoretical outline postulates that differences in bargaining power influences contract design by make asymmetric termination fee structures in acquisitions more likely. According to our hypotheses, differences in bargaining power can arise from bidder characteristics, such as the bidder being an institutional investor (INVESTOR), from target characteristics, such as the target facing insolvency (INSOLVENCY), or from situational deal characteristics as in the case of secondary buyouts (SECONDARY). For INSOLVENCY, we find a highly significant positive coefficient, which supports H2: for deals that involve an insolvent target, the bargaining power is unevenly distributed, leading to asymmetrically structured termination fees. The opposite result can be observed for SECONDARY, namely a highly significant negative coefficient. Thus, secondary buyouts are more likely to employ symmetric termination fees supporting H3 that more evenly distributed bargaining power also increases the likelihood of evenly distributed termination fees. For our INVESTOR variable, we find a positive but not significant coefficient. Thus, we cannot claim that an institutional bidder makes symmetric or asymmetric termination fees more likely. This result may be caused by the fact that we combine both types of asymmetric termination fee structures, i.e., higher target than bidder termination fees and higher bidder than target termination fees, in the logit regression. If the bidder is an institutional investor, we would only expect a higher likelihood for higher bidder than target termination fees, as we outline in our hypothesis H1. To investigate this further, we employ a multinomial logistic regression next, which allows us to differentiate between the two asymmetric fee structures.

Table 4 Panel B presents our multinomial logistic model’s regression results as laid out in Equations (2) and (3). We again control for the full list of variables as before. Entirely consistent with H1, the highly significant (p < 0.01) and positive coefficient of INVESTOR in the regression on ASYM2 shows that deals where the bidder is an institutional investor indeed are more likely to use an asymmetric, rather than a symmetric, termination fee structure with a higher bidder than target fee. This supports our conjecture that an institutional investor has a weaker bargaining position in the acquisition and accepts a disadvantageous termination fee structure to signal deal commitment. Simultaneously, we also find a significant (p < 0.05) and negative coefficient for INVESTOR in our ASYM1 model. This finding could explain why we do not see a significant effect in our logistic regression. The likelihood for higher target than bidder termination fees is significantly lower than for our base outcome of symmetric termination fees. This further supports the notion that the bargaining position is skewed towards the target if the bidder is an institutional investor.

Regarding H2, we again find supportive evidence: deals where the target is in insolvency proceedings (INSOLVENCY) are highly significantly (p < 0.01) positively associated with asymmetric termination fees where the target fee is higher than the bidder fee. Simultaneously, we observe that insolvency is significantly negatively associated with an asymmetrically higher bidder than target termination fee compared to a symmetric fee structure. Interestingly, the negative coefficient for asymmetrically higher bidder termination fees is much larger in absolute size than the positive coefficient for asymmetrically higher target termination fees. This may suggest that targets under insolvency are under increased pressure and therefore have very little bargaining power to help them realize a more favorable termination fee structure than the bidder.

Our last hypothesis (H3) posits that acquisition partners in secondary buyouts meet on an equal footage and are therefore less likely to agree to asymmetric termination fees. The multivariate regression results again support this argument. We find negative coefficients for both equations, and the observed associations between secondary buyouts and both types of asymmetric termination fee structures are significant at the 5 and 10% level, respectively. Consequently, we observe that both parties are more likely to accept symmetric termination fees in secondary buyouts than both, higher target than bidder termination fees and higher bidder than target termination fees.

Our control variables are mostly in line with expected effects. Interestingly, we find that LN_DEAL VALUE and PUBLIC are significantly positively associated with both asymmetric higher bidder than target termination fees and higher target than bidder termination fees as compared to symmetric termination fees. This corroborates our findings from the logistic regression that deals of higher value are, in general, more likely to employ asymmetric termination fees of any kind. We again find for EUROPE highly significant negative coefficients and for ANGLO highly significant positive coefficients for both models. This confirms our previous results that deals with bidders from the European continent are less likely to employ asymmetric termination fees of any kind. In contrast, bidders from the United States, Canada, Great Britain, or Ireland are more likely to be involved in deals with asymmetric termination fees. Lastly, for deals that are financed with equity (EQUITY) and for deals that did not receive a target’s board recommendation (UNSOLICITED HOSTILE), we find a higher likelihood for symmetric termination fees compared to higher bidder than target termination fees. No significant coefficients are observed for the control variables CROSS BORDER, AUCTION, MBO, COMPLETE, CASH, DEAL PREMIUM, CONTESTED HOSTILE and CONTESTED RECOMMEND.

To further validate our results, we conduct several robustness tests. From a bargaining power perspective, we so far assumed that contracts without termination fees for either parties are implicitly the same as contracts with symmetric termination fees. Nevertheless, deals that include no termination fees may be more fundamentally different from deals that do include termination provisions.^[23] To test the robustness of our results against this concern, we reduce our sample by excluding all deals (23,709) where neither the target nor the bidder negotiated a termination fee and rerun our analyses again. From the logistic regression with the aforementioned criteria for our subsample, we find no significant results for our variables of interest (Table 5 Panel A in the Supplementary Material). For INVESTOR and INSOLVENCY this again is explained by the fact that the logistic regression includes both types of asymmetric termination fee structures when we indeed would only expect one type of asymmetric termination fee to be significant. Consequently, this is supported by our multinomial logit model that runs with the same subsample (Table 5 Panel B in the Supplementary Material). in which we find highly significant results for both variables which emphasizes the importance to differentiate between the two types of asymmetric termination fee structures. In detail, even though this approach reduces our sample by a substantial amount, we still find for our INVESTOR variable a highly significant (p < 0.01) positive coefficient for higher bidder than target termination fees. Similarly, our results for INSOLVENCY remain robust, as we find again a significantly (p < 0.05) positive coefficient in ASYM1 (higher target than bidder termination fees) and a highly significant (p < 0.01) negative coefficient in ASYM2 (higher bidder than target termination fees). However, we do not observe significant results for our variable capturing secondary buyouts (SECONDARY). This result may be explained by the fact that by excluding deals with no (i.e., implicit symmetric) termination fees, we lose tremendous explanatory power in determining the likelihood of symmetric termination fees.

As a second robustness test, we exclude deals that fall into the sectors Telecommunication & Media, Financial Industry, and Utilities. These are highly regulated industries that might provide an additional obstruction to consummate a transaction for bidders and targets. For example, in the utilities sector, the regulatory approval process often includes public interest considerations that introduce additional risks for both transaction partners. To rule out that these regulatory considerations drive our results, we rerun our full multinomial model (i.e., with our sample that includes deals with and without termination fees) but exclude firms from these three industries (Table 5 Panel C in the Supplementary Material).^[24] Even though we lose 7726 observations (n = 18,300), our results remain mostly intact with two exceptions. We lose significance on our SECONDARY variable in ASYM2 (p = 0.203) and marginally on our INVESTOR variable in ASYM1 (p = 0.105) while all other effects exhibit an even stronger significance than in our base model. Crucially, this includes INVESTOR in ASYM2 (p = 0.000), which means that we maintain support for H1, positing that institutional investors as bidders are more likely to exhibit higher bidder than target termination fees.

In an additional step, we again excluded all deals without the explicit inclusion of termination provisions and rerun the previous robustness test that only considers deals in highly regulated industries (Table 5 Panel D in the Supplementary Material). Even though we again lose substantial statistical power by reducing the number of observations (n = 841), the results are largely in line with the preceding analyses. In detail, we lose significance for our SECONDARY variable in ASYM1 and ASYM2 and for our INVESTOR variable in ASYM1, while again finding significant effects for INSOLVENCY in both models (ASYM1 and ASYM2) and for INVESTOR in ASYM2. We therefore conclude that asymmetric termination fees are not exclusively a consequence of heightened regulations in specific industries.

Additionally, even though we tried to control for country-specific differences in our sample through various macroeconomic and governance-related controls, we cannot completely rule out that our results are affected by economic, regulatory, or institutional differences between countries. To address such concerns, we rerun our model with only domestic deals from the United States, as US bidders and US targets constitute the most considerable portion in our sample (Table 5 Panel G in the Supplementary Material). By doing so, our sample size is reduced to 6623 observations, and we lose some significance for our SECONDARY variable in ASYM1 (p = 0.224) and ASMY2 (p = 0.204) while all other effects remain unchanged.^[25]

In the next step, we again employ our reduced sample that only consists of deals that explicitly include termination fee provisions for our robustness test that excludes all non-US deals (Table 5 Panel H in the Supplementary Material). By doing so, we lose more than half of our sample size (n = 641) and again significance for our SECONDARY variable in both model specifications (ASYM1 and ASYM2). while also losing some significance for INSOLVENCY in ASYM1. The effects of INVESTOR and INSOLVENCY are nevertheless still present in ASYM2 when applying the substantially reduced sample. For US deals that explicitly include termination fee provisions a bidder being an institutional investor is, thus, associated with a significantly higher likelihood of higher bidder than target termination fees, while the target being insolvent is associated with a significant lower likelihood for higher bidder than target termination fees.

Another aspect of country-specific differences is the cultural variation between countries that might influence the acceptance and propensity of termination provisions. Chakrabarti, Gupta-Mukherjee, & Jayaraman (2009), for instance, find that specific cultural traits and the cultural distance (or proximity) between two parties influence deal outcomes. Based on Hofstede’s (2001) original four dimensions, we therefore include the cultural distance between bidder and target in terms of power distance (PDI BID TAR), individualism/collectivism (IDV BID TAR), masculinity/femininity (MAS BID TAR), and uncertainty avoidance (UAI BID TAR) in our base model (Table 5 Panel I in the Supplementary Material). Interestingly, two of our cultural control variables show some explanatory power. We find a significantly higher likelihood for higher target than bidder termination fees and a lower likelihood for higher bidder than target termination fees compared to a symmetrical structure if the difference in power distance (i.e., inequality of power distribution) between the bidder and target country is more pronounced. On the other hand, if the cultural distance regarding the perceived value of individualism between bidder and target society is larger, we find a significantly higher likelihood for higher bidder than target termination fees. These results may be related to the larger literature regarding cross-border acquisitions and show that cultural distance may not just affect deal outcome and long-term performance (Chakrabarti, Gupta-Mukherjee, & Jayaraman, 2009) but also deal negotiations by influencing contract design. Nevertheless, our results for all three hypotheses remain unchanged.

In an additional analysis, we included the cultural control variables in the multinomial logit model with our subsample that excludes all deals with no termination fee on either side, resulting in 1302 observations (Table 5 Panel J in the Supplementary Material). Again, our results remain unchanged compared to our base model in Table 5 Panel B. Specifically, we find a significant higher likelihood for higher bidder than target termination fees for deals that involve an institutional bidder and a significant higher (lower) likelihood for higher target than bidder (higher bidder than target) termination fees for deals with insolvent targets.

In conclusion, our results are mainly robust, except for our variable that captures secondary buyouts. Concerning this specific variable, the effect from our main logistic regression loses some strength in our multinomial model and is only partially robust when applying our additional tests. This result could potentially point to the fact that the situational differences between two institutional investors that govern the resulting bargaining power in a deal supersede the presumed interest by both parties in quick negotiations and in an existence of an efficient secondary buy-out market.