Bankruptcy spillover effects on strategic alliance partners

Journal of Financial Economics 103 (2012) 551–569

Contents lists available at SciVerse ScienceDirect

Journal of Financial Economics journal homepage: www.elsevier.com/locate/jfec

Bankruptcy spillover effects on strategic alliance partners$ Audra L. Boone a,n, Vladimir I. Ivanov b a b

Mays Business School, Texas A&M University, College Station, TX 77843, United States U.S. Securities and Exchange Commission,1 Washington, DC, United States

a r t i c l e i n f o

abstract

Article history: Received 8 December 2010 Received in revised form 9 June 2011 Accepted 13 June 2011 Available online 25 October 2011

This paper examines whether a party to a strategic alliance or joint venture suffers from spillover effects when the other partner files for bankruptcy. We find that the nonbankrupt strategic alliance partners, on average, experience a negative stock price reaction around their partner firm’s bankruptcy filing announcement. This negative effect is strongest for longer partnerships and those with higher returns at the announcement of the initial alliance formation. Furthermore, horizontal alliance firms in declining industries have lower returns, indicating that industry conditions can exacerbate expected problems for the non-bankrupt firm. Non-bankrupt partners also experience drops in profit margins and investment levels in the subsequent two years with the worst performance concentrated among the longer-term agreements. There is very little impact on the returns or performance for joint venture partners, which suggests that these agreements are more insulating for the partner firm. & 2011 Elsevier B.V. All rights reserved.

JEL classification: G33 G34 L22 Keywords: Strategic alliance Joint venture Bankruptcy Spillover effects

1. Introduction Long-term collaborative partnerships between two firms, such as strategic alliances and joint ventures, provide alternative contracting arrangements to simple one-off market transactions or complete integration. A key feature of these contracts is the timing and nature of the termination of the agreement. Partnerships are

$ We are grateful for comments from an anonymous referee, Mike Hertzel, Scott Lee, Vahap Uysal and seminar participants at the 2010 University of Oklahoma Finance Conference and the 2010 Financial Management Association Meeting for comments on prior drafts. This paper was previously circulated under the working title, ‘‘The effect of bankruptcy on strategic alliance partners.’’ n Corresponding author. E-mail address: [email protected] (A.L. Boone). 1 The U.S. Securities and Exchange Commission, as a matter of policy, disclaims responsibility for any private publication or statement of any of its employees. The views expressed herein are those of the author and do not necessarily reflect the views of the Commission or of the author’s colleagues upon the staff of the Commission.

0304-405X/$ - see front matter & 2011 Elsevier B.V. All rights reserved. doi:10.1016/j.jfineco.2011.10.003

typically designed to last for a set number of years or until the parties reach a specified goal. Some contracts also contain early termination rights that allow one or both parties to walk away from the agreement in certain contingencies. These circumstances can be specific, such as change in control provisions, or vague, such as uncured breach (Robinson and Stuart, 2007). The intentionally incomplete nature of these early termination clauses provides a role for ex post litigation as a contracting tool that increases the overall surplus of the partnership (Scott and Triantis, 2006). These rights are valuable for the holder because they allow the firm to terminate the agreement when the perceived costs of continuing the partnership outweigh the gains. Bankruptcy, which can impair the filing party’s ability to maintain its end of the agreement, can also be considered grounds for possible termination. Regardless of whether the non-bankrupt partner leaves the relationship, the valuation and operational effects on that firm are not clear. The ability to terminate in this situation could be positive because it provides the opportunity to walk

552

A.L. Boone, V.I. Ivanov / Journal of Financial Economics 103 (2012) 551–569

away from a weakened partner and put resources toward a higher-valued use without necessarily harming its reputation from modifying the agreement. Alternatively, the partner’s bankruptcy could have a detrimental effect if it prematurely ceases the benefits that had been accruing from the agreement or harms the operations and growth of the non-bankrupt partner. Despite the increasing frequency and importance of alliances and joint ventures to corporate investment and growth, there is still little systematic evidence whether the distress of one party produces spillover effects for the other partner.2 We fill this gap by examining whether the bankruptcy filing of one party in a strategic alliance or joint venture affects the valuation and operating performance of the non-bankrupt partner. For the 1989–2007 time period, we present evidence that the bankruptcy filing of one party in a collaborative agreement is met with a negative stock price reaction for the counterparty. Furthermore, the non-bankrupt partners exhibit declines in profit margins and investment in the subsequent years after the partner firms file for bankruptcy. This evidence is consistent with the notion that bankruptcies have significant spillover effects on contracting parties. Thus, our work contributes to previous studies that have documented important contagion effects of financial distress and bankruptcy filings on customers, suppliers, and competitors (see Lang and Stulz, 1992; Hertzel, Li, Officer, and Rogers, 2008). We also investigate whether these negative effects are largely anticipated or unexpected by the market. Often bankruptcies are the culmination of a long slide into distress, and therefore, it is possible that the market has already incorporated the expected effects of the partner’s distress into the counterparty’s stock price. We find that the negative stock price effects are primarily confined to the bankruptcy filing announcement and that the full effects do not seem to be anticipated in the pre-filing period. To ascertain whether the spillover effects are stronger for certain types of partnerships, we study a number of key characteristics of the firms and partnerships. First, we analyze whether the bankruptcy effects differ between long-duration and short-duration partnerships. Longerduration agreements are likely to occur when the parties find the ongoing relationship to be particularly valuable. Hence, the occurrence of a bankruptcy filing would have the most detrimental impact for the counterparty in these situations. Robinson and Stuart (2007) also note that longer projects are more likely to have unanticipated events occur, so the bankruptcy would be less expected at the inception. Furthermore, if the parties anticipate distress, they might structure the partnership differently to mitigate any spillover effects. We find that the negative stock and operating performance is primarily found in the longer-term agreements, indicating that partner firms suffer more when the agreement is likely to have greater value.

2 Robinson (2008) notes that alliance arrangements have grown by 16% per year since 1985.

Next, we explore whether organizational structure— strategic alliance or joint venture—is an important determinant for the magnitude of the bankruptcy effects. Both types of collaborative arrangements involve significant negotiations to divide the income and intellectual/physical assets stemming from the venture, but while strategic alliances are cooperative arrangements between distinct firms set up to reach a common goal, joint ventures (JVs) create a new legal entity that operates separately from the contributing parties’ core operations. Consequently, each type of structure possesses some distinct contractual features that could either exacerbate or mitigate the impact of a counterparty bankruptcy. For example, strategic alliances are more fluid with greater ambiguity regarding specific goals, and consequently might be easier and quicker to unwind, but also more likely to involve the parties’ core business practices. In contrast, JVs have clear boundaries that could better insulate the parties’ remaining assets, but might involve more costs and rely heavily on the full participation of each party. We divide the sample by each type of arrangement and find that the negative stock price and operating performance effects are concentrated in the strategic alliance subsample, with little evidence of spillovers for JV partners. In further analysis, we find support for the notion that both the types of projects and firm characteristics at the time of the inception of the relationship affect the choice of strategic alliance versus JV. Consequently, these results suggest that JV agreements are more insulating to the partner firm, but that the partners may factor this consideration into their choice of organization form. Other aspects of the agreements, such as relationshipspecific investments and financial constraints, could affect the gains accruing from the relationship, and hence, the valuation effects and operational impacts from the bankruptcy filing. To explore this issue, we first study how the presence of equity stakes and board participation influence the bankruptcy’s impact and find significant negative wealth effects associated with the presence of those. In addition, we divide the data by low research and development (R&D) versus high R&D industries, horizontal versus vertical agreements, and financially constrained versus financially unconstrained firms. In all instances we continue to find that the negative wealth effects are the strongest for long-term alliance partnerships and alliances with high estimated benefits. Despite these declines, there is no significant change in bankruptcy probability for the non-bankrupt long-term, alliance partners. We also provide evidence that industry performance impacts the returns to the partner firms. In particular, partners in horizontal relationships with firms that go bankrupt, but are from better-performing industries, tend to fare better. Likewise, a declining industry results in worse returns. This finding is consistent with the notion that spillover effects are exacerbated when the company’s own prospects are weaker and might not have many alternatives for finding a replacement partner. The remainder of the paper is organized as follows. In Section 2, we discuss prior literature and findings on strategic alliance and joint venture arrangements. Section 3 describes the data sample process. In Section 4,

A.L. Boone, V.I. Ivanov / Journal of Financial Economics 103 (2012) 551–569

we provide results on stock return reactions to the nonbankrupt partner around the counterparty’s bankruptcy and distress dates in both the univariate and multivariate framework. Section 5 presents evidence on the operating performance of the non-bankrupt partner after the counterparty’s bankruptcy. Section 6 provides information on changes in bankruptcy probability and credit ratings for the non-bankrupt partner firms. Section 7 discusses the choices driving the organizational form of alliance versus joint venture. We conclude in Section 8. 2. Prior literature on strategic alliances and joint ventures 2.1. Formation and benefits Much of the prior research on strategic alliances and joint ventures emphasizes the value-enhancing aspects of these arrangements. The potential benefits include: learning and knowledge acquisition (Berg and Friedman, 1981), efficiency improvements, lower transaction costs and risk diversification (e.g., Stuckey, 1983; Hennart, 1988), access to financing (e.g., Berg and Friedman, 1981; Hennart, 1988), enhancement of competitive positions or market power (Kogut, 1988), and cooperation in the development of new technology (Gomes-Casseres, Hagedoorn, and Jaffe, 2006). Further work by Robinson (2008) provides theoretical and empirical support that these arrangements tend to occur when it is difficult to enforce contracts internally or when companies undertake diversifying operations. Given all the potential benefits, it is not surprising that the inception of strategic alliances and joint ventures are, on average, value-increasing. For example, McConnell and Nantell (1985) and Chan, Kensinger, Keown, and Martin (1997) find that participants experience positive stock price reactions around the announcement of the alliance or joint venture formation and observe improved operating performance in the following years. Allen and Phillips (2000) demonstrate that strategic alliances, joint ventures, and other product market relationships in conjunction with corporate block ownership lead to significant increases in target firm stock prices and improvements in their profitability and operating performance. Krishnaswami, Pablo, and Subramaniam (2004) show that strategic alliances coupled with equity stakes alleviate the capital constraints of smaller, high-growth firms and that these partnership announcements lead to significantly positive market reactions. Ivanov and Lewis (2009) find that Initial Public Offering (IPO) firms with alliances that commence before the offering tend to obtain higher IPO valuations, invest more, and grow faster than similar IPO firms without strategic alliances.

553

settings are generally incomplete, and this challenging contracting environment can lead to opportunistic behavior by at least one of the involved parties. For example, one partner can exploit the other by exerting insufficient effort, underinvesting, or expropriating a disproportionately large share of the joint surplus created by the strategic relationship (Lerner and Malmendier, 2010). The incomplete contracting literature (e.g., Klein, Crawford, and Alchian, 1978; Grossman and Hart, 1986; Hart, 1988, 2001; Aghion and Tirole, 1994) shows that equity ownership and corresponding control rights can mitigate potential hold-up problems between parties to a strategic alliance. Matthews (2006) demonstrates that ownership stakes by an established firm in an entrepreneurial alliance partner can serve to deter entry into the entrepreneur’s market. The trade-off is that the equity stake can decrease the incentive to produce entrepreneurial effort. Staged infusions of capital and detailed termination rights outlined in the contracts allow parties to leave the arrangement if one party fails to reach prescribed goals (Robinson and Stuart, 2007). 2.3. Termination and early exits from alliances and joint ventures Alliance and joint venture partnerships end for a variety of reasons. Participants typically structure these agreements to last for a particular length of time or until the arrangement achieves a predetermined goal. Walkaway provisions that provide early termination rights for one or both parties are common in long-term partnerships. For example, Lerner and Malmendier (2010) document that contracts often grant the financing partner in alliances the unilateral right to leave the agreement (with a fee payment), and even possibly award the intellectual property back to the financier partner as a way to ensure that the researcher does not reallocate the funds to other projects. Other work by Robinson and Stuart (2007) shows that termination rights may be granted for breach of contract, change of control, and loss of key employees. While the clauses can be very specific, other times the events granting the rights can be vague. Being intentionally imprecise has benefits, such as allowing for the role of litigation in the event of disputes (Scott and Triantis, 2006). These termination provisions serve to protect the parties, but also incentivize the other firm to act in the best of interest of the joint agreement by mitigating opportunistic behavior. Terminating the agreement would generally occur when the costs associated with continuing the arrangement outweigh the expected value from continuing. 2.4. Termination in bankruptcy and bankruptcy law

2.2. Risks and costs An emerging area of research discusses the potential risks and costs of intermediary forms of corporate arrangements. Alliances and joint ventures often suffer from a host of contracting problems, especially when they include very risky activities. Financial contracts in such

The bankruptcy of one party to a long-term contract could put a premature end to the sharing agreement even though the alliance or joint venture might still produce benefits for both parties. Some contracts even appear to anticipate the possibility of such an event and stipulate the rights of the other party under these circumstances.

554

A.L. Boone, V.I. Ivanov / Journal of Financial Economics 103 (2012) 551–569

For example, in an agreement reached between IMedicor and Direct Medical Solutions Corp. on November 5, 2009, the contract states that, ‘‘In the event that your business files for bankruptcy or ceases to operate for any reason, the revenue sharing will cease immediately.’’ Though the non-bankrupt partner is generally not responsible for the other party’s claims, it can still have a negative impact on its operations. U.S. bankruptcy law Section 362 provides for an automatic stay that prohibits creditors, including alliance partners, from collecting payments. So, a party to a strategic alliance or JV that had expected payments for its services would not be able to collect those funds if they were not considered part of normal operations and not approved by a bankruptcy judge. Furthermore, U.S. bankruptcy law Section 365 gives the bankrupt firm the right to accept or reject existing contracts and agreements for a period of time.3 Consequently, even if the bankrupt party survives and emerges from a Chapter 11 filing, the agreement may have been rejected or irreparably harmed during the process.

The Appendix provides examples of some of the strategic alliance and joint venture agreements that appear in our sample. It lists the type of partnership, the names of the parties, including the bankrupt firm (in bold), the initial announcement date of the agreement, and a brief description of the activities conducted as part of the transaction. 4. Wealth effects associated with partner firm bankruptcy filing In this section we estimate the wealth effects for the non-bankrupt sample firms around three dates: the initial announcement that the partner firms agreed to form a strategic alliance or joint venture, the filing date of the bankruptcy, and the pre-filing distress date. To do this, we construct cumulative abnormal returns (CARs) around each of these dates using the market-adjusted returns model (Brown and Warner, 1985) and the value-weighted market return from CRSP estimated over the days 46 to  226 relative to the announcement date.

3. Data and sample procedure Our alliance and JV partnership sample comes from the Securities Data Company (SDC) Platinum Joint Ventures/ Alliances database. We match the partnership sample with bankruptcy announcement dates from the SDC Platinum Bankruptcies database. We then select all strategic alliances and JVs for which a partner experienced bankruptcy during the period 1989–2007. In addition, we require that all partners are public companies with data in the Center for Research in Securities Prices (CRSP). Panel A of Table 1 presents the final breakdown of observations. There are 130 partners that file for bankruptcy who held a total of 366 partnerships at the time of the filing. Of these, 282 arrangements are strategic alliances and 84 are joint ventures. We analyze JVs separately because they result in the creation of a separate legal entity with assets contributed by the partners. As previously discussed, organizational structure could affect how much a firm’s bankruptcy affects its partner. There are 288 unique non-bankrupt partners, but some of these firms have more than one partner go bankrupt over the sample period. Consequently, there are 427 total observations where a non-bankrupt firm has a current partner file for bankruptcy. Panel B of Table 1 shows the annual distribution of bankruptcy announcements. The majority of these occur in 2001–2003, the period after the Internet bubble burst. As seen in Table 1, Panel C, almost half of the sample of the non-bankrupt alliance partners is concentrated in business services, telecommunications, electronic and electrical equipment, industrial machinery, and electric and gas services. Panel D shows that the same holds true for the bankrupt alliance partners. 3

Bankruptcy law (Section 365(n)) provides special protections for non-debtor firms involved in intellectual property licensing agreements by allowing a licensee to retain rights to the property so long as it keeps making royalty payments. If the bankruptcy firm is a licensee, it can still choose to reject or accept contracts.

4.1. Wealth effects around inception of alliance or joint venture formation Prior research has shown that strategic alliances and joint ventures generally result in positive wealth gains at their inception (e.g., McConnell and Nantell, 1985; Johnson and Houston, 2000; Chan, Kensinger, Keown, and Martin, 1997). To calibrate the results presented later in this paper, we first examine CARs for both a ( 1, þ1) and a (  2, þ2) window centered on the initial announcement of the formation of the alliance or joint venture. These results are presented in Table 2. The mean initial announcement returns are 1.03% for the full sample with alliances slightly lower at 0.98% and JVs slightly higher at 1.15%. These figures are roughly consistent with prior findings in the literature. We note that long-term alliances with eventual durations greater than 4 years have the lowest formation returns with a value of 0.54%.4 Since our sample consists only of agreements where one party ultimately files for bankruptcy, these announcement returns might not be representative of the general population of alliances and JVs over our sample period. Therefore, we also estimate announcement returns for a general sample of alliances and JVs over the 1989–2007 time period, regardless of whether one partner went bankrupt. For this general sample, we find that alliances average returns of 1.30% and JVs average returns of 1.05%. Comparing the initial announcement returns of bankrupt partners to the general population of alliances and JVs, we can see that the CARs for the JVs are comparable. However, strategic alliance firms who have a partner that eventually goes bankrupt have, on 4 There could be several explanations for the lower announcement returns for long-term alliances. First, those alliances could have been more difficult to price (more difficult for investors to assess the benefits). Alternatively, it may be that investors are able to anticipate from the beginning that there is some bankruptcy risk involved and priced this risk.

A.L. Boone, V.I. Ivanov / Journal of Financial Economics 103 (2012) 551–569

555

Table 1 Sample information. This table presents summary information for the sample of strategic alliance or joint venture partnerships where one party to the agreement files for bankruptcy over the 1989–2007 time period. Panel A breaks down the sample observations by bankrupt and non-bankrupt firms and by type of agreement—either strategic alliance or joint venture. Panel B shows the distribution of the sample by the year that one of the alliance or joint venture partners filed for bankruptcy. Panel C shows the industry distribution of the 288 non-bankrupt alliance partners. Panel D presents information on the industry affiliation of the 130 bankrupt firms. Industry affiliation is determined by two-digit SIC code. Panel A: Bankruptcies by type of alliance Number of bankrupt partners

130

Number of total partnerships held at time of bankruptcy Number of strategic alliances Number of joint ventures

366 282 84

Number of unique non-bankrupt partners (JVs and strategic alliances)

288

Number of instances where a non-bankrupt firm has a partner go bankrupt

427

Panel B: Annual distribution of sample bankruptcies Year

Number of bankruptcies

1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 Full sample

1 1 4 4 4 3 4 5 4 8 9 6 25 21 16 5 6 2 2 130

Panel C: Industry distribution of non-bankrupt alliance partners Two-digit SIC Business services Telecommunications Electronic & equipment Industrial machinery & equip. Electric, gas & sanitary services Transportation equipment Chemicals and allied products Food and kindred products Transportation by air Primary metal industries Printing and publishing Petroleum and coal products Instruments & related products General merchandise stores Holding & other invest. offices Wholesale trade—durable goods Food stores Oil & gas extraction Construction—special trade contractors Apparel & other textile products Depository institutions Non-depository credit institutions Rubber and miscellaneous products Motor freight transportation Transportation services Wholesale trade—nondurable goods Miscellaneous retail

73 48 36 35 49 37 28 20 45 33 27 29 38 53 67 50 54 13 17 23 60 61 30 42 47 51 59

Frequency

Percent

44 36 30 28 17 16 15 9 8 6 5 5 5 5 5 4 4 3 3 3 3 3 2 2 2 2 2

15.3 12.5 10.4 9.7 5.9 5.6 5.2 3.1 2.8 2.2 1.7 1.7 1.7 1.7 1.7 1.5 1.5 1.0 1.0 1.0 1.0 1.0 0.7 0.7 0.7 0.7 0.7

556

A.L. Boone, V.I. Ivanov / Journal of Financial Economics 103 (2012) 551–569

Table 1 (continued ) Panel C: Industry distribution of non-bankrupt alliance partners Two-digit SIC Insurance carriers Engineering & management services All other Total

63 87 n/a

Frequency

Percent

2 2 17 288

0.7 0.7 5.9 100.0

Panel D: Industry distribution of bankrupt alliance partners

Telecommunications Business services Industrial machinery & equip. Electronic & equipment Electric, gas & sanitary services Chemicals and allied products Transportation by air Primary metal industries Transportation equipment Instruments & related products Wholesale trade—durable goods General merchandise stores Heavy construction Apparel & other textile products Metal mining Transportation services Wholesale trade—nondurable goods Home furniture, furnishings, equip. stores Miscellaneous retail Engineering & management services All other Total

Two-digit SIC

Frequency

Percent

48 73 35 36 49 28 45 33 37 38 50 53 16 23 10 47 51 57 59 87 n/a

21 19 10 10 6 5 5 4 4 4 4 4 3 3 2 2 2 2 2 2 16 130

16.1 14.6 7.7 7.7 4.6 3.9 3.9 3.1 3.1 3.1 3.1 3.1 2.3 2.3 1.5 1.5 1.5 1.5 1.5 1.5 12.4 100.0

Table 2 Wealth effects around initial partnership formation announcement. This table presents mean and median cumulative abnormal returns for (  1, þ1) and ( 2,þ 2) windows at the initial partnership announcement for the non-bankrupt partner firms in the 427 instances when the counterparty to the alliance or joint venture files for bankruptcy. Day 0 represents the initial news announcement that the partner firms have agreed to collaborate on either a strategic alliance or joint venture. Abnormal returns are measured using a market model computed over a ( 46,  246) window relative to the formation announcement. Long-term (Short-term) partnerships have durations greater (less) than the sample median duration of 4 years. Returns for ( 1, þ1) window # Obs.

Returns for (  2,þ 2) window

Mean (median)

t-Test for mean ¼ 0

Mean (median)

t-Test for mean ¼ 0

Full sample

427

1.03% (0.23%)

3.27

1.05% (0.32%)

3.33

Alliances

308

2.87

140

Short-term alliances

135

1.03% (0.24%) 0.38% (0.20%) 1.60% (0.14%)

2.23

Long-term alliances

0.98% (0.13%) 0.54% (0.13%) 1.24% (0.10%)

JVs

119

Long-term JVs

51

Short-term JVs

59

1.15% (0.26%) 2.21% (0.40%) 0.61% (0.62%)

average, 30% lower announcement returns compared to the general sample. These findings could indicate that the market anticipates potential problems for alliance

0.91 2.23 2.06 2.17 0.84

1.08% (0.33%) 2.31% (0.67%) 0.13% (0.02%)

0.45 2.35 1.33 2.39 0.11

partners—including effects from a bankruptcy. Work in subsequent sections will further investigate this potential explanation.

A.L. Boone, V.I. Ivanov / Journal of Financial Economics 103 (2012) 551–569

4.2. Wealth effects around bankruptcy and distress announcement dates As shown in Table 2, the formation of partnerships is generally considered to be value-enhancing by the market though the expected benefits appear to be less than the general population of all partnerships during the same time period. What are the value consequences when one party goes bankrupt? On the one hand, it is possible the bankruptcy will diminish, or in the worst case scenario completely eliminate, the synergies accruing from these partnerships, resulting in negative spillover effects on the non-bankrupt partner. If, however, the bankruptcy provides the non-bankrupt party with the ability to walk away from a weakened partner and redistribute its resources to a higher-valued venture, then it could be beneficial. Therefore, the effect of a bankruptcy filing on the other party is an empirical question. The CARs for the bankruptcy filing announcements are reported in Table 3. We focus on the narrower 3-day window, but also report results for the 5-day window for

557

robustness. We find that partners, on average, exhibit a negative stock price reaction to the bankruptcy announcement of the counterparty. The results are marginally significant for the (  1,þ1) window, but not for the (  2, þ2) window. Thus, the general impact of the bankruptcy filing is negative, but not strongly so for the nonbankrupt partners. Due to the differential nature of the protections and costs afforded between alliances and joint ventures, it is possible that the effects are largely confined to one type of agreement. Accordingly, we parse the sample based on organizational form. We find that the reaction is much stronger for strategic alliances than for joint ventures. The bankruptcy return for strategic alliances over the (  1, þ1) window is  0.6%, which is significant at the 5% level. The stock price decrease is also economically significant. Using the average market value of the nonbankrupt firms in the month prior to the bankruptcy announcement, the 0.6% drop translates into a $217 million dollar loss. For JVs, the CARs around the partner bankruptcy filing are insignificant. This difference

Table 3 Non-bankrupt partner wealth effects around partner firm bankruptcy announcement dates. The table summarizes the wealth effects for the non-bankrupt partner firms in the 427 instances when the counterparty to the alliance or joint venture files for bankruptcy from 1989 to 2007. Bankruptcy announcement return is the cumulative abnormal return centered on the bankruptcy announcement date for both 3-day ( 1, þ1) and 5-day (  2,þ 2) windows. Combined return is the weighted sum of the initial partnership return and the bankruptcy return over the relevant windows. It is computed as (Initial partnership formation announcement return from Table 2  (Pre-formation market value/ (Total market value))þ (Bankruptcy announcement return  (Pre-bankruptcy market value/Total market value)) where Total market value ¼ Preformation market value þpre-bankruptcy market value. Each market value is measured in the month prior to the relevant announcement. Total dollar value effect is the total gains/losses based over both the initial formation and bankruptcy announcement windows for each partner firm. It is calculated as (Initial partnership announcement return  Pre-formation market value) þ(Bankruptcy announcement return  Pre-bankruptcy market value). Longterm (Short-term) partnerships have durations greater (less) than the sample median duration of 4years. t-Values for returns ¼0 are given in parentheses. n nn nnn , , indicates significance at 10%, 5%, and 1% levels, respectively. Returns and values for (  1,þ 1) window

No. of obs. All agreements Alliances þ JV

427

Alliances

308

JVs

119

Long-term partnerships Alliances þ JV

191

Alliances

140

JVs

51

Short-term partnerships Alliances þ JV

194

Alliances

135

JVs

Long-term minus short-term Test of diff. in means Test of diff. in medians

59

Bankruptcy announcement return

 0.39% (  1.79)n  0.60% (  2.42)nn 0.16% (0.36)  0.94% (  2.86)nnn  1.12% (  3.24)nn  0.01% (  0.12) 0.12% (0.33)  0.01% (  0.02) 0.40% (0.55)

Combined return

0.21% (1.45) 0.16% (0.98) 0.34% (1.14)  0.11% ( 0.53)  0.28% (  1.11) 0.36% (0.85) 0.52% (2.34) 0.56% (2.14) 0.44% (1.02)

p-value 0.02 0.03

Total dollar value effect ($mil)

117.6 82.4 209.7

10.8  47.9 173.1

183.5 145.6 271.9

Returns and values for (  2, þ2) window Bankruptcy announcement return

Combined return

 0.44% (  1.61)  0.41% (  1.27)  0.54% (  0.99)

0.08% (0.47) 0.12 (0.64)  0.04% ( 0.11)

 0.93% (  2.66)nnn  0.83% (  1.97)n  1.20% (  1.93)n

 0.18% ( 0.77)  0.24% ( 0.90)  0.01% ( 0.03)

 0.11% (  0.24)  0.02% (  0.31) 0.003% (0.01) p-value 0.17 0.15

0.22% (0.87) 0.30% (1.04) 0.01% (0.03)

Total dollar value effect ($mil)

201.6 193.9 221.7

76.8 53.5 144.2

289.1 281.9 305.8

558

A.L. Boone, V.I. Ivanov / Journal of Financial Economics 103 (2012) 551–569

supports the notion that JVs, which involve the creation of a separate legal entity, tend to isolate the healthier parties from the negative impact of a partner’s bankruptcy. Next, we investigate whether the duration of the agreement plays a role in the stock price reaction of the non-bankrupt firms to a partner-firm bankruptcy announcement. We conjecture that partners in longerlived agreements suffer lower announcement returns for two reasons. First, partnerships are likely to last longer when the economic benefits still accruing from the alliance or JV continue to be positive and substantial. Second, the bankruptcy was less likely to be anticipated at the inception of partnership, resulting in fewer contractual protections against such an event. We segment the sample based on duration of the partnership and classify alliances and JVs as long-term when their duration is greater than the sample median duration of four years. The results show that the negative effect of the bankruptcy announcement is concentrated in the subsample of longer-duration strategic alliances. Notably, the 3-day bankruptcy return for the longer-duration subsample is  1.12% (t-stat¼3.24), while the wealth effects for shorter-duration alliances and all JVs are negligible. Both the mean and median price drop over this window are significantly larger for partners in long-term alliances. This evidence indicates that partners suffer more when the alliance agreement has more value or the partner was more exposed to the counterparty’s distress. To put these results into perspective, we compare them to the partnership formation announcement returns presented in Table 2 and compute a combined return that represents a weighted average of returns by the nonbankrupt firm’s market value one month prior to each announcement. Combined return¼(Initial partnership formation announcement return  (Pre-formation market value/Total market value))þ (Bankruptcy announcement return  (Pre-bankruptcy market value/Total market value)) where Total market value ¼Pre-formation market value þPre-bankruptcy market value. The results, presented in Table 3, show that the combined return for the full sample is positive at 0.21%, which indicates that partner firms fare little cumulative wealth loss overall. Long-term alliances are the only group to exhibit a negative combined return at  0.28%, but this result is not significant at the 10% level. To examine the economic impact, we compute the Total dollar value effect, which represents the dollar value gains/losses based over both the initial formation and bankruptcy announcement windows for each partner firm. It is calculated as (Initial partnership announcement return  Pre-formation market value) þ(Bankruptcy announcement return  Prebankruptcy market value we compute). The losses for the long-term alliances total $47.9 million. It is possible that the market anticipates the bankruptcy announcements, and thus, the returns centered on the actual bankruptcy filing from Table 3 could underestimate the wealth effects to the counterparty’s shareholders. Therefore, we also evaluate the effect of the initial distress event on the price of non-bankrupt partners. To identify pre-filing distress dates, we use the approach of Hertzel, Li, Officer, and Rogers (2008). In the

Table 4 Non-bankrupt partner wealth effects around pre-filing distress dates. The table contains wealth effects around the pre-bankruptcy distress date to the non-bankrupt partner firms in the 427 instances when the counterparty to the alliance or joint venture files for bankruptcy from 1989 to 2007. The mean cumulative abnormal returns are centered on the distress date for the partner that eventually files for bankruptcy. The distress dates are identified using the approach of Hertzel, Li, Officer, and Rogers (2008)—it is the day with the largest drop in abnormal returns in the year prior to the bankruptcy announcement. Long-term (Short-term) partnerships have durations greater (less) than the sample median duration of 4 years. t-Values that the mean return equals zero are presented in parentheses. n, nn, nnnIndicates significance at 10%, 5%, and 1% levels, respectively. CAR (  1,þ 1) CAR (  2,þ2) No. obs. All Agreements Full sample Alliances JVs Long-term partnerships Full sample Alliances JVs Short-term partnerships Full sample Alliances JVs Long-term minus short-term Test of diff. in means Test of diff. in medians

0.41% (0.66) 0.34% (0.41) 0.60% (1.33)

0.47% (0.69) 0.52% (0.57) 0.34% (0.53)

427

 0.67% ( 2.05)  0.73% ( 1.79)n  0.49% (  0.03)

 0.62% ( 1.44)  0.57% ( 1.09)  0.77% ( 1.05)

179

1.49% (1.10) 1.59% (0.85) 1.25% (1.56)

1.68% (1.16) 1.95% (0.98) 1.03% (0.88)

p-value

p-value

0.11 0.37

0.11 0.55

308 119

134 45

190 135 55

year prior to the bankruptcy filing, we identify the date with the largest negative abnormal return of the filing firm and use it as the distress date.5 Table 4 presents the results for the wealth effects around distress announcement dates. There is no significant effect for the full sample or the subsamples of strategic alliances and JVs. When we separate alliances into longer-term and shorter-term, we find that the former are associated with more negative effects than the latter, but only for the (  1, þ1) window. Long-term alliance partners suffer, on average, a 0.73% decline in market value on the distress date, which is statistically significant at the 10% level. 4.3. Equity stakes and board memberships Next we examine whether the presence of equity stakes and board memberships of strategic alliance 5 We also examine news announcements in the year prior to the bankruptcy announcements to identify the first time a bankruptcy possibility is mentioned. Using these dates, we repeat the wealth effect analysis, but do not find any significant results.

A.L. Boone, V.I. Ivanov / Journal of Financial Economics 103 (2012) 551–569

partners affect the size of the wealth effects. Ownership stakes and board presence by participating partners can be a response to contractual and market frictions. In particular, they tend to exist when partners might be concerned about potential hold-up problems surrounding large, relationship-specific investment (Fee, Hadlock, and Thomas, 2006). Thus, we expect that partners with equity stakes or representation on the board of other alliance members may suffer more if those members go bankrupt. We collect information on equity holdings and board membership from annual proxy statements. Consistent with Fee, Hadlock, and Thomas (2006), we find equity ownership and/or board membership is not that common in alliances. Only 30 observations of our sample alliances include equity and/or board presence by both the nonbankrupt party in the bankrupt party, and vice versa. The wealth effects associated with this subsample are presented in Table 5. Caution is due in interpreting some of the results, however, because of the small sample size. Nevertheless, we find that alliances with ownership and board membership have statistically negative and significant CARs on bankruptcy announcement dates. JVs, on the other hand, tend to experience positive CARs, although those are not significant. We further divide the subsample according to the partnership duration, but do not find any significant results, potentially due to the small sample size. Thus, the results in Table 5 provide some evidence that partners holding equity stakes and sitting on the Table 5 Wealth effects around bankruptcy announcement dates–equity ownership and board membership. This table presents the bankruptcy announcement wealth effects for the subsample of 30 non-bankrupt partner firms when the counterparty to the alliance or joint venture files for bankruptcy from 1989 to 2007 where the partnership involved equity stakes and/or board memberships between the two firms. The cumulative abnormal returns are centered on the bankruptcy announcement date. Long-term (Shortterm) partnerships have durations greater (less) than the sample median duration of 4 years. Those partnerships with durations of exactly four years are not included in the breakdown of long-term and short-term agreements. t-Values that the mean returns equals zero are presented in parentheses. n, nn, nnnIndicates significance at 10%, 5%, and 1% levels, respectively. CAR (  1,þ 1)

CAR (  2,þ 2)

 0.91% ( 1.30)  1.55% ( 1.82)n 1.18% (1.66)

0.12% (0.15)  0.29% ( 0.28) 1.48% (1.65)

Long-term partnerships Full sample  0.86% (  0.89) Alliances  1.93% ( 1.57) JVs 1.50% (1.55)

0.46% (0.53)  0.29% ( 0.27) 2.11% (1.84)

16

Short-term partnerships Full sample  2.22% ( 1.62) Alliances  2.54 ( 1.61)

 1.68% ( 0.72)  1.90% ( 0.69)

7

All agreements Full sample Alliances JVs

Num. obs.

30 23 7

11

559

boards of their alliance counterparts suffer wealth losses when a bankruptcy is announced. 4.4. Industry relationships—horizontal versus vertical agreements The type of industry relationship between the partner firms is another important characteristic that could significantly impact the wealth effects around the bankruptcy announcement. Generally, alliances and joint ventures could be classified as horizontal or vertical. Horizontal arrangements are formed with partners that operate in the same lines of business. The firms engaged in horizontal agreements benefit from economies of scale in production and distribution and from increased market power due to declining competition. Vertical alliances are formed with partners from different stages of the production chain. Partners in vertical agreements typically enjoy cost savings from transportation, inventory, etc. Prior studies have shown that horizontal JVs create value while vertical JVs do not (Johnson and Houston, 2000; Slovin, Sushka, and Mantecon, 2007). Furthermore, bankruptcies related to broader-based industry problems could have a differential impact on horizontal versus vertical partners. While counterparties to both types of relationships might have difficulty finding replacement partners when the industry is in trouble, horizontal firms could face the added problem that the declining industry conditions propagate more readily to their core business. These issues lead us to conjecture that partners in horizontal agreements are likely to suffer more serious consequences. To test this notion, we divide the sample into horizontal and vertical partnerships based on the two-digit Standard Industrial Classification (SIC) codes of the participating firms. If two partners have the same (different) two-digit SIC, then we consider it a horizontal (vertical) agreement. Panel A of Table 6 presents the results. We find that horizontal alliances suffer large, negative wealth effects from the bankruptcy of an alliance partner. The (  1, þ1) bankruptcy return of horizontal alliances is negative and significant, but insignificant for vertical alliances.6 For the subsample of long-term agreements, we find that partners from both horizontal and vertical alliances suffer significant negative returns upon the announcement of a bankruptcy, though the difference between these two groups is not statistically different. Overall, these results continue to support our earlier finding that long-term alliance partners generally tend to suffer lower stock price returns than short-term alliance partners. Following Hertzel, Li, Officer, and Rogers (2008), we also examine whether our results are simply driven by customer–supplier relationships. For example, a firm

5

6

6 We find similar results if we use four-digit SIC code to determine horizontal alliances. The CAR(  1,þ 1) for the 42 horizontal alliance partners is  1.51%, significant at the 5% level. This value is almost three times the average four-digit SIC rival reaction (  0.49%) found by Hertzel, Li, Officer, and Rogers (2008). The long-term horizontal alliance partners experience the worst performance with a CAR(  1,þ1) of  2.34%.

560

A.L. Boone, V.I. Ivanov / Journal of Financial Economics 103 (2012) 551–569

Table 6 Wealth effects around bankruptcy announcement dates: vertical vs. horizontal agreements, high vs. low R&D, and liquid vs. financially constrained. The table shows the mean cumulative abnormal returns broken down by firm and partnership characteristics for non-bankrupt partners in the 427 instances when the counterparty to the alliance or joint venture files for bankruptcy from 1989 to 2007. The return windows are centered on the bankruptcy announcement date. Duration is measured as the number of years that the two firms have been involved in an agreement. Panel A examines horizontal and vertical agreements. Horizontal agreements are defined as the partner firms operating in the same two-digit SIC code and Vertical agreements are classified as when the firms do not operate in the same two-digit SIC code. Long-term (Short-term) partnerships have durations greater (less) than the sample median duration of 4 years. Panel B examines high R&D versus low R&D firms. High R&D firms have a R&D/Total assets ratio that is higher than the median R&D/Total assets ratio across industries for the year of bankruptcy announcement. Panel C contains analysis on liquid versus constrained firms. Financially constrained firms are those with cash and after-tax operating cash flows are smaller than their investments in the year prior to the bankruptcy announcement. t-Values are presented in parentheses. n, nn, nnn indicates significance at 10%, 5%, and 1% levels, respectively.

Panel A: Horizontal vs. vertical agreements Horizontal agreements Full sample

No. of obs.

CAR (  1,þ 1)

CAR ( 2, þ2)

119

 0.54% ( 1.15)  1.10% ( 2.15)nn 1.37% (1.25)

 0.001% ( 0.01)  0.10% ( 0.14) 0.32% (0.24)

 1.30% ( 2.12)nn  0.47% (  0.46)

 0.16% ( 0.24) 0.025% (0.16)

 0.33% ( 1.38)  0.39% ( 1.39)  0.19% (  0.42)

 0.61% (  2.11)nn  0.54% (  1.62)  0.79% (  1.35)

 1.10% ( 2.54)nn 0.17% (0.42)

 1.10% (  2.05)nn  0.32% ( 0.72)

p-value

p-value

0.38 0.35

0.83 0.75

Alliances

92

JVs

27

Long-term partnerships Short-term partnerships Vertical agreements Full sample Alliances JVs

37 37

308 216 92

Long-term partnerships

103

Short-term partnerships

98

Long-term horizontal minus long-term vertical Test of diff. in means Test of diff. in medians Panel B: High R&D vs. low R&D firms High R&D alliance partners Full sample Alliances

226 172

JVs

54

Long-term partnerships

90

Short-term partnerships

65

Low R&D alliance partners Full sample

201

Alliances

136

JVs

65

Long-term partnerships

50

Short-term partnerships

70

Long-term high R&D minus long-term low R&D Test of diff. in means Test of diff. in medians

 0.34% ( 1.13)  0.52% ( 1.52) 0.24% (0.41)

 0.43% (  1.18)  0.39% ( 0.90)  0.54% ( 0.83)

 0.95% ( 2.34)nn  0.04% (0.06)

 0.45% ( 0.94)  0.46% ( 0.51)

 0.45% ( 1.41)  0.71% ( 1.96)n  0.009% (  0.14)  1.43% ( 2.23)nn 0.02% (0.05)

 0.46% (  1.09)  0.43% ( 0.89)  0.53% ( 0.63)  1.51% (  1.88)n 0.11% (0.17)

p-value

p-value

0.74 0.83

0.89 0.55

A.L. Boone, V.I. Ivanov / Journal of Financial Economics 103 (2012) 551–569

561

Table 6 (continued )

Panel C: Liquid vs. financially constrained firms Constrained firms Full sample

No. of obs.

CAR (  1,þ1)

CAR ( 2,þ 2)

114

 0.73% ( 1.37)  0.93% ( 1.78)n  0.13% ( 0.09)

 0.57% (  0.77)  0.27% (  0.33)  1.46% (  0.84)

 1.99% ( 2.58)nn  0.01% ( 0.02)

 0.60% (  0.61)  0.17% (  0.31)

 0.27% ( 1.19)  0.48% ( 1.70)n 0.25% (0.71)

 0.40% (  1.51)  0.46% (  1.44)  0.24% (  0.53)

 0.89% ( 2.32)nn  0.04% ( 0.09)

 0.88% (  1.91)n  0.29% (  0.59)

Alliances

85

JVs

29

Long-term partnerships

29

Short-term partnerships

135

Liquid firms Full sample

313

Alliances

223

JVs

90

Long-term partnerships

111

Short-term partnerships

92

Long-term constrained minus long-term liquid Test of diff. in means Test of diff. in medians

p-value

p-value

0.10 0.26

0.61 0.85

making a product could lose access to a critical component, leading to increased production costs or lost sales. Another possibility is that the bankruptcy of the seller of goods could signal declines in demand for the other firm’s products and lower future revenue. Using information from SDC’s Joint Venture database, we identify 11 instances of strategic alliances involving a supplier and a customer. The CARs( 1,þ 1) for the partners are positive but insignificant. We further divide the sample into customers (five cases), which have a CAR(1,þ1) of  2.03% and suppliers (six cases) which have a CAR(1,þ1) of þ3.44%. While we have relatively few observations, it appears that customers of the bankrupt firms suffer large wealth declines, while suppliers have positive returns.

higher than the median R&D/Total assets ratio across all industries for the year of bankruptcy announcement. The results from the analysis are presented in Panel B of Table 6. Both high and low R&D firms experience negative returns, but the wealth effects do not appear to be stronger for the former compared to the latter. In fact, the returns around the (  1,þ1) window for high R&D alliances partners is 0.52% which is not statistically significant, while the (  1,1) window for low R&D alliance partners is  0.71%, which is significant at the 10% level. As in the previous subsections, we find that the effect of an alliance partner’s bankruptcy announcement is especially strong for long-term alliances. Similarly, we find no effect for JVs.

4.5. Low R&D firms versus high R&D firms

4.6. Liquid versus financially constrained firms

We next turn our attention to companies from areas that are likely to have specialized assets and potential contractual problems and ask the question: do such companies suffer more when an alliance partner goes bankrupt? Allen and Phillips (2000) show that companies operating in high R&D industries benefit from alliances with other corporations through increased investment and earnings. Therefore, high R&D firms could experience larger negative effects than other firms when an alliance partner files for bankruptcy. These firms might also face higher distress costs because of their relative lack of tangible assets. To test for such effects, we divide the sample into high R&D and low R&D firms. We define a high R&D firm as one whose R&D/Total assets ratio is

Lastly, we examine the bankruptcy announcement effects for liquid versus constrained partners. One of the main benefits of strategic alliances and JVs is that they alleviate financial constraints of participating firms. Thus, losing a partner due to bankruptcy might have a more negative effect on a financially constrained partner than a liquid one. Following Allen and Phillips (2000), we define financially constrained firms as those whose cash and aftertax operating cash flows are smaller than their investments in the year prior to the bankruptcy announcement. Panel C of Table 6 presents the returns for constrained and liquid firms. We find that both constrained and liquid firms experience a significant negative wealth effect at the announcement of a bankruptcy of an alliance partner.

562

A.L. Boone, V.I. Ivanov / Journal of Financial Economics 103 (2012) 551–569

However, the announcement effect appears to be much stronger for constrained alliance partners. The size of constrained firms’ returns is almost twice the size of the returns of the liquid firms, although tests for equality of medians show no statistically significant differences. We again show that long-term alliance partners exhibit the largest negative wealth effects. As before, we find no effect for JVs. 4.7. Multivariate regressions of bankruptcy announcement effects on partner firms So far our analysis has focused on univariate tests. It is possible, however, that the differences we document in the previous subsections are driven by firm- and partnership-specific characteristics for which currently we do not control. In this subsection we implement Ordinary Least Squares (OLS) regressions of bankruptcy announcement abnormal returns on these characteristics along with several control variables. Because the bankruptcies in our sample are concentrated in certain years and industries, we cluster the OLS standard errors by bankrupt firm industry and time. 4.7.1. Firm and partnership characteristics Firm characteristics include whether the non-bankrupt partner was a high R&D firm and whether it was liquidity constrained. In addition to the basic partnership characteristics such as whether the agreement was structured as a strategic alliance (SA) or whether the partners were horizontally related, we include several other important factors. First, we include Long-term SA and Long-term JV dummy variables that are equal to one for partnerships with durations above the sample median of 4 years. Second, we also control for the relative size of the alliance partners. It is possible that only relatively large bankrupt alliance partners generate a significant impact on the non-bankrupt partners. Towards that end, we construct a dummy variable that is equal to one if the size of the bankrupt partner (measured by book value of assets in the year prior to bankruptcy) is larger than that of the nonbankrupt partner. Lastly, we compute a dummy variable equal to one if the partner firms share a common venture capitalist (VC). Lindsey (2008) finds that one of the important value-added contributions of VCs is facilitating strategic alliances among portfolio firms. We want to control for the fact that alliances initiated by a common VC (usually early on in a company’s life cycle) can be very important for the partners and thus entail stronger wealth effects at a bankruptcy announcement. Table 7 presents the coefficient estimates from the OLS regression of the 3-day cumulative abnormal returns for non-bankrupt firms around their partners’ filing date. Model 1 includes all firm and partnership characteristics. Consistent with the univariate results, the wealth impact of the bankruptcy is significantly negative and stronger for long-term strategic alliance partnerships. The bankruptcy announcement effects are also stronger when the bankrupt partner is larger than the non-bankrupt firm as indicated by the significantly negative coefficient on the Relative size dummy, which suggests that smaller partners

suffer relatively more losses.7 Though the Strategic alliance and Long-term JV dummies are both negative, they are statistically insignificant. Model 2 includes a measure to quantify the expected size of the benefits for the partners in the alliance. To accomplish this goal, we first compute the 3-day cumulative stock price reaction of the non-bankrupt partners around the initial alliance or joint venture formation announcement as we did in Table 2. We then multiply this return by the non-bankrupt firm’s market value at the end of the month immediately preceding the announcement date to produce the dollar value of the initial market reaction. This variable is labeled Initial market reaction. The coefficient on Initial market reaction is positive and significant at the 10% significance level. This finding suggests that firms in alliances that were considered beneficial by the market at their formation experience better (less negative) wealth effects at the partner’s bankruptcy announcement. However, when we interact Initial market reaction with Long-term SA, the coefficient on the interaction term is negative and significant at the 1% level (see Model 3). Thus, the coupling of a long-term strategic alliance with a higher initial assessed value results in the partner firm losing significantly more wealth on the bankruptcy announcement date. This finding confirms the notion that non-bankrupt partners suffer the most when the arrangement is longer lived and was expected to have a high value. We also note that the coefficients of both the Longterm SA and Relative size dummies are economically significant. For example, if an agreement switches from a short-term to long-term partnership, then its stock price reaction on the bankruptcy announcement day experiences an additional drop of 0.9%. Given that the average market value of non-bankrupt firms in the month prior to the announcement is $36.2 billion, this effect translates into a loss of $326 million. The median values are lower with a market value of $3.6 billion for a median loss of $32.4 million. We also estimate the regression with a measure of the alliance duration in years and find that an increase in the duration of the alliance of 2 years causes a decline in stock price reaction by 0.6%. The effect is stronger for the Relative size dummy. If the bankrupt firm is larger than the non-bankrupt firm, then the latter suffers an additional 1.2% stock price decrease. 4.7.2. Measures of expected bankruptcy One possible explanation of the results is that the bankruptcy announcement is more unexpected for particular types of partnerships. To examine the viability of this alternative explanation, we follow the approach of Gande and Lewis (2009) and include a bankruptcy likelihood measure for the bankrupt partner, computed in the year prior to the bankruptcy announcement, as an additional explanatory variable in the regression model. Each 7 For robustness, we refined the relative size measure to have separate dummies for big (a size ratio greater than 1.2), peer-to-peer (size ratio between 0.8 and 1.2), and small (size ratio less than 0.8). The results hold and indicate that the bigger the bankrupt firm relative to the non-bankrupt partner, the lower is the estimated bankruptcy return.

A.L. Boone, V.I. Ivanov / Journal of Financial Economics 103 (2012) 551–569

563

Table 7 Regression of bankruptcy announcement returns on various control variables. OLS regressions of the announcement returns for the non-bankrupt partner in the 427 instances when the counterparty to the alliance or joint venture files for bankruptcy from 1989 to 2007 on a set of control variables. Cumulative abnormal returns are measured over a 3 day window centered on the bankruptcy announcement date of the partner. Strategic alliance is a dummy variable equal to one for strategic alliances and zero for joint ventures. Longterm SA is a dummy variable equal to one if the strategic alliance duration is greater than 4 years and zero otherwise. Long-term JV is a dummy variable equal to one if the JV duration is greater than 4 years and zero otherwise. Horizontal partnership occurs when the two-digit SIC code of the alliance partner and the bankrupt partner are the same. High R&D firm is a variable equal to one if the firm’s R&D/Total assets ratio is higher than the median R&D/Total assets ratio across all industries for the year of bankruptcy announcement. Liquid firm is a dummy equal one for firms whose cash and after-tax operating cash flows are larger than their investments in the year prior to the bankruptcy announcement. Common VC is a dummy equal to one if the alliance partner and the bankrupt partner had the same VC backer. Initial market reaction is (abnormal return on alliance announcement (  1,þ 1))*(company market value at the end of the month prior to the alliance or joint venture announcement). Bankruptcy propensity  1 is the predicted probability of bankruptcy in the year prior to the bankruptcy announcement derived from a probit model of bankruptcy. Number of industry bankruptcies is the number of bankruptcies that occurred during the last year in the four-digit SIC industry of the respective bankrupt partner. Industry-adjusted return is the marketadjusted return on a value-weighed portfolio of companies within the same four-digit SIC industry over the past year. Idiosyncratic bankruptcy is a dummy variable equal to one if the industry performs better than the market and there are no bankruptcies during the prior year. Relative size is a dummy variable equal to one if the ratio of total assets of bankrupt firm  1/total assets of alliance partner  1 is greater than one and zero otherwise. t-Satistics based on two-way clustered (time and industry) standard errors are reported in parentheses. *, **, *** indicates significance at 10%, 5% and 1% levels, respectively.

Constant Strategic alliance Long-term SA Long-term JV Relative size Horizontal partnership High R&D firm Liquid firm Common VC

(1)

(2)

(3)

(4)

(5)

(6)

(7)

(8)

0.0057 (0.81)  0.0055 (  1.07)  0.0099 (  2.78)***  0.0051 (  0.68)  0.0120 (  3.06)***  0.0014 (  0.26) 0.0029 (0.60) 0.0001 (0.78)  0.0023 (  0.30)

0.0056 (0.78)  0.0057 ( 1.08)  0.0098 ( 3.01)***  0.0052 (  0.71)  0.0118 ( 2.95)***  0.0016 (  0.33) 0.0030 (0.59) 0.0001 (0.71)  0.0014 (  0.17) 0.00001 (1.91)*

0.0055 (0.77)  0.0057 ( 1.06)  0.0094 ( 2.55)**  0.0053 ( 0.77)  0.0119 ( 3.03)***  0.0020 ( 0.38) 0.0031 (0.63) 0.0001 (0.72)  0.0010 ( 0.12) 0.00002 (4.43)***  0.00003 ( 4.70)***

0.0071 (0.89)  0.0052 ( 1.00)  0.0088 ( 2.55)**  0.0051 ( 0.71)  0.0119 ( 3.15)***  0.0016 ( 0.31) 0.0028  0.56 0.0001  0.79  0.0004 ( 0.05) 0.00002 (4.52)***  0.00003 ( 5.44)***  0.1413 ( 1.12)

0.0084 (0.98)  0.0055 (  1.03)  0. 0088 (  2.49)**  0.0049 ( 0.67)  0. 0123 (  3.31)***  0.0012 ( 0.23) 0.0030 (0.55) 0.0001 (0.84) 0.0005 (0.10) 0.00002 (4.32)***  0.00002 (  5.40)***  0.1472 (  1.12) 0.0071 (1.19)  0.0018 ( 0.33)

 0.0046 (  0.49)  0.0051 (  0.99)  0.0091 (  2.75)***  0.0044 (  0.60)  0. 0119 (  3.00)***  0.0025 (  0.50) 0.0031 (0.60) 0.0001 (0.91) 0.0001 (0.01) 0.00002 (6.01)***  0.00003 (  5.81)***  0.1462 (  1.10)

0.0090 (1.1)  0.0050 (  1.06)  0. 0100 (  2.71)  0.0054 (  0.75)  0.0113 (  2.44)  0.0006 (  0.14) 0.0023 (0.46) 0.0001 (0.62)  0.0048 (  1.07) 0.00002 (4.08)***  0.00003 (  6.29)***  0.1834 (  1.33) 0.0049 (1.38)

0.0104 (1.21)  0.0065 ( 1.33)  0. 0096 ( 2.66)  0.0058 (  0.79)  0.0099 ( 2.14)  0.0025 (  0.49) 0.0020 (0.39) 0.0001 (0.69)  0.0035 (  0.41) 0.00002 (3.86)***  0.00003 ( 6.28)***  0.1882 ( 1.32) 0.0097 (0.95)

Initial market reaction Long-term SA * Initial market reaction Bankruptcy propensity  1 Industry-adjusted return Idiosyncratic bankruptcy Number of industry bankruptcies

0.0011 (1.13)

Horizontal * Industry-adjusted return

0.0202 (2.13)**

Customer/supplier Number of observations Adjusted R2

0.0092 (1.30) 427 3.19%

427 3.52%

firm’s bankruptcy likelihood is estimated by running a probit regression with a panel data set that covers the period 1985–2007 and includes the bankrupt partners from our sample as well as matching companies with the same four-digit SIC code. The panel starts in 1985 to allow firms to have at least 4 years of data prior to the event. The dependent variable in the probit model equals one if a

427 3.83%

427 3.95%

427 4.22%

427 4.38%

427 4.77%

427 4.41%

given firm announced a bankruptcy, and zero otherwise. Explanatory variables are measured at the beginning of the year in which the bankruptcy is observed and include most of the factors employed by Shumway (2001): Working capital/Total assets, Retained earnings/ Total assets, Earnings before interest and taxes (EBIT)/ Total assets, Sales/Total assets, Net income/Total assets,

564

A.L. Boone, V.I. Ivanov / Journal of Financial Economics 103 (2012) 551–569

Total liabilities/Total assets, Current assets/Current liabilities, company’s excess return, and the idiosyncratic standard deviation of each firm’s stock returns. Each bankrupt partner is assigned a value called Bankruptcy propensity  1. This value is the predicted probability of bankruptcy in the year prior to the actual bankruptcy filing announcement derived from a probit model of bankruptcy. In Table 7, the coefficient on Bankruptcy propensity  1 is negative, which indicates that more expected events are associated with smaller price reaction at the actual filing, but the coefficient is not statistically significant. The coefficients on the other variables are similar to the results in the other models. 4.7.3. Idiosyncratic versus industry-wide declines and agreement type Another factor that could impact the size of the wealth effect at bankruptcy announcement is whether the bankruptcy is primarily associated with idiosyncratic factors or broader industry-wide declines. Though bankruptcies associated with industry-wide declines could be more readily anticipated, they also could present even more problems for the partner firm. Horizontal partners might find that their own prospects have eroded due to diminished industry revenues. Even vertical partners could suffer if they find it more difficult to find substitute partners following the bankruptcy. Either way, industry deterioration could exacerbate problems following a bankruptcy filing. To make a distinction between idiosyncratic bankruptcy risks and broader industry declines, we employ three empirical measures: (a) the performance of a portfolio of companies from the same industry of the bankrupt alliance partner in the year prior to the bankruptcy (Industry-adjusted return), (b) a dummy variable equal to one if there are no bankruptcies during the prior year (Idiosyncratic bankruptcy), and (c) the number of recent (within a year) bankruptcies that take place within that industry (Number of industry bankruptcies).8 We define an industry based on four-digit SIC codes as in Lang and Stulz (1992), but we also use two-digit SIC codes as a robustness check. Models 5 and 6 of Table 7 contain these variables. The coefficient on Industry-adjusted return is positive, which suggests that better (worse) industry performance mitigates (worsens) the magnitude of the negative stock price reaction, but the coefficient on Idiosyncratic bankruptcy, is negative, indicating that alliance partners’ stock price reaction is more negative if the bankruptcy is firmspecific. However, neither of these coefficients is statistically significant. Next, Number of industry bankruptcies has a positive coefficient, indicating that industry decline reduces the negative stock price reaction experienced by the non-bankrupt alliance partners, possibly because they convey information early on about potential likelihood of bankruptcy of the later-bankrupt alliance partner. Like the other industry variables, this coefficient is not statistically significant. 8 We also use a dummy variable equal to one if there have been any bankruptcies in the industry within the last year, and the results are similar to those when we use Number of industry bankruptcies.

We further explore how industry performance is linked to the type of partnership—horizontal or vertical—by interacting the Horizontal partnership dummy with the Industry-adjusted return variable. As reported in Model 7 of Table 7, Industry-adjusted return is still positive, but insignificant. The interaction term is also positive, but significant, indicating that better-performing industries help mitigate wealth deterioration from the bankruptcy for horizontal partners. Likewise, declining industry prospects would exacerbate the loss in value. As a related point, it is possible that industry consolidation drove the decision to form the partnership, which would affect the ability to compete or combine with other firms after the bankruptcy. In unreported results, we compute a consolidation variable that takes a value of one if the horizontal partners belonged to an industry in the top quartile of all merger and acquisition (M&A) deals over the 3 years prior to the bankruptcy and included this in the regressions. This variable was insignificant. In Model 8 of Table 7, we also included a dummy variable for customer–supplier agreements; this coefficient is positive, but not significant. For vertical partnerships, we examine the market for vertical agreements for firms from the same industry using data from SDC to gauge the competitiveness of the market for these types of relationships. We also examine the time to next alliance for the non-bankrupt partner. There was no evidence that these factors significantly impact the returns for vertical partnership firms. These results are unreported but available upon request. Overall, our results remain intact in the regression models. Non-bankrupt partners in long-term alliances with initially high assessed dollar value experience significantly negative wealth effects when a partner announces bankruptcy. Firms in horizontal relationships are further affected by industry conditions. Declining industry conditions make the wealth effects of a partner’s bankruptcy filing worse. These findings are consistent with the notion that, despite the unquestionable benefits, alliance partners can suffer spillover effects from the other partner’s bankruptcy filing. Furthermore, the bankruptcy effects on JV partners, even in the case of longterm JV agreements, appear to be negligible. 5. Effect of partner bankruptcy on the operating performance of non-bankrupt alliance partners To glean additional insight into why the market perceives a firm’s bankruptcy as negative for the other party, we investigate the changes in the operating performance and financial policies of non-bankrupt partners following the bankruptcy announcement. We are primarily interested in how their cost structure, sales growth, investments, and borrowings change as a result of a partner’s bankruptcy. We use information from Compustat to calculate sales growth (changes in item 12), profit margin (item 13/item 12), investments (item 46þitem 128/item 6), current ratio (data 4/data 5), quick ratio ((data 4  data 3)/data 5), and debt (data 9/data 6). We execute the operating performance analysis using a difference-in-differences approach. For each sample

A.L. Boone, V.I. Ivanov / Journal of Financial Economics 103 (2012) 551–569

firm, we select a matching firm that is from the same industry, is of similar size, and has similar profitability. Barber and Lyon (1996) and Lie (2001) show that it is important to control for past profitability in order to obtain well-specified test statistics in operating performance studies. Industry classification is based on SIC codes. We match on four-digit and three-digit SICs. If there are not enough matches, we rely on two-digit and one-digit SICs. Size is measured as the sales in the year prior to the bankruptcy announcement. We require that the size of each matching firm is within 50–150% of the size of our sample firm. We also require that there are at least five firms within that size group. Lastly, from the firms with similar size, we select the one with the closest past performance, where performance is measured by the firm’s return on assets (ROA) measured as operating income before depreciation divided by total assets. We follow the performance of our sample firms relative to that of their matched group in the 3 years after the bankruptcy announcement. Every year, the matched firmadjusted performance is compared to that in the year before the bankruptcy announcement. If a matching firm drops out, we splice the next closest matching firm. The results are presented in Table 8. We show that strategic alliance partners, but not JVs, experience lower profit margins and investment levels in the first and second year after the bankruptcy announcement. The effects are stronger for longer-term alliances. Though investments recover to their pre-bankruptcy levels by Year 3, profit margins tend to remain significantly lower in the 3 years following the bankruptcy announcement. In addition, by Year 3, strategic alliance partners have higher current and quick ratios than the pre-bankruptcy year. We do not detect any differences regarding sales growth and debt levels. 6. Changes in bankruptcy probabilities and credit ratings for non-bankrupt partner firms The results up to now suggest that the non-bankrupt party can face substantial wealth deterioration and operating difficulties. As a last way to investigate this issue, we attempt to capture the change in probability that the nonbankrupt partner will go bankrupt. We estimate the probability of the non-bankrupt alliance partners going bankrupt using a probit model as in Gande and Lewis (2009) and the specification in Shumway (2001). To estimate the model, we use all bankrupt and non-bankrupt firms with available CRSP and Compustat data in the period 1985–2007. We then use the coefficients from the model to calculate implied probability of bankruptcy for our sample of non-bankrupt firms. We follow the estimated bankruptcy probability for a 6-year period, from 2 years before the bankruptcy announcement until 3 years after it. The results are reported in Panel A of Table 9. We find that the implied bankruptcy probability of the non-bankrupt alliance partners increases slightly in the year prior to bankruptcy and then decreases slightly in Year 2, but most of the changes are not statistically significant. We also examine changes in credit ratings for sample firms

565

with available ratings data. We use the long-term credit rating from Compustat and again track the ratings and changes in ratings from 2 years before the bankruptcy announcement until 3 years after. We find evidence that credit ratings of our sample firms deteriorate in the years after the bankruptcy announcement. However, caution is needed in interpreting the credit rating results since these are only univariate results. In Panel B of Table 9, we also study the implied probability of bankruptcy of the subsample of long-term strategic alliances and obtain results that are similar to those for the whole sample. Due to small sample size for long-term agreements with available information, we do not perform a credit rating analysis. 7. The choice of organizational form The results in the prior sections suggest that parties to a strategic alliance suffer more than parties to a joint venture when their partners file for bankruptcy. If strategic alliance partners are more vulnerable than joint venture partners, then why do firms form strategic alliances? Earlier in the paper we discussed that strategic alliances can offer important benefits compared to joint ventures. Relatively higher startup and continuation costs for joint ventures could influence the decision to form an alliance as well. Despite the perils, strategic alliance partners can use contractual features to manage the potential risk of a partner bankruptcy. Unfortunately, we do not have access to the alliance and JV contracts for our sample firms to observe how those are structured at the inception of the partnership, or how they subsequently evolve in response to a changing financial environment. Therefore, in this section we model the choice of the organizational form as a function of firm characteristics and the primary purpose of the partnership. Our goal is to shed more light on whether the choice between strategic alliances and joint ventures depends on the future prospects of the involved firms. For this purpose we estimate a probit model using the sample of non-bankrupt strategic alliance/JV partners. The dependent variable in the model is equal to one if a firm selects a strategic alliance and zero if it selects a joint venture. The right-hand side of the model includes firm characteristics at the inception of the agreement, such as firm size and ROA, of both the bankrupt and non-bankrupt firms measured in the year prior to the announcement of the alliance or joint venture. These variables serve as proxies for the future prospects of the sample firms. We classify the strategic alliances and joint ventures into several categories based on information from the SDC Platinum Joint Venture database that describes the nature of the agreement (e.g., joint R&D development, marketing partnership, supplier–customer partnership, etc.). Table 10 presents the estimates of the probit model. There are several notable results. First, it is evident that certain types of projects are more likely to be executed using the strategic alliance organizational form. For example, firms engaging in licensing agreements and R&D agreements are more likely to form strategic alliances while those engaging in manufacturing and

566 Table 8 Operating performance following bankruptcy announcements of alliance partners—a difference-in-differences approach. The table presents the matching firm-adjusted measures of operating performance for the non-bankrupt partners with available data in the 349 instances when the counterparty to the alliance or joint venture files for bankruptcy from 1989 to 2007. We calculate the (Post–Pre) difference of each measure as the difference between the post-bankruptcy and pre-bankruptcy value of the measure. We peer-adjust the (Post–Pre) difference by subtracting the (Post–Pre) value of the corresponding matching firms. Matching firms are the closest firms to each sample firm by industry, size (50–150% of sales) and past performance (measured as ROA in the year prior to bankruptcy announcement). Year 0 is the year of bankruptcy announcement. Mean and Median denote the mean and median of this matching firm-adjusted difference. Long-Term (Short-Term) partnerships have durations greater (less) than the sample median duration of four years. We report in parentheses the p-values of t-tests for the means and Wilcoxon signed rank tests for the medians equaling zero. n, nn, nnnIndicates significance at 10%, 5%, and 1% levels, respectively. Sales growth

Profit margin

Investments

Current ratio

Quick ratio

Debt

Mean

Median

Mean

Median

Mean

Median

Mean

Median

Mean

Median

Mean

Median

Panel A: Year þ 1 Full sample

349

Strategic alliances only

245

0.0310 (0.39) 0.0315 (0.39) 0.0282 (0.65)  0.0118 (0.79) 0.0298 (0.72)

0.0011 (0.61) 0.0013 (0.40)  0.0003 (0.51)  0.0041 (0.74)  0.0155 (0.81)

 0.0093 (0.55)  0.0085 (0.52) 0.0006 (0.97)  0.0213 (0.35)  0.0113 (0.79)

 0.0056 (0.11)  0.0094 (0.02)nn  0.0158 (0.04)nn  0.0056 (0.18) 0.0002 (0.41)

 0.0028 (0.65)  0.0065 (0.34)  0.0163 (0.09)n 0.0033 (0.77) 0.0060 (0.65)

 0.0021 (0.08)n  0.0039 (0.03)nn  0.0036 (0.09)n  0.0040 (0.35) 0.0023 (0.55)

0.0113 (0.90) 0.0076 (0.95) 0.0796 (0.68)  0.0827 (0.64) 0.0211 (0.83)

0.0200 (0.30) 0.0369 (0.31) 0.0870 (0.11)  0.0169 (0.75)  0.0121 (0.76)

0.0136 (0.86)  0.0006 (0.99) 0.1048 (0.53)  0.1129 (0.45) 0.0518 (0.58)

0.0058 (0.42) 0.0084 (0.56) 0.0565 (0.14)  0.0398 (0.53)  0.0157 (0.53)

 0.0026 (0.72)  0.0057 (0.49) 0.0045 (0.70)  0.0089 (0.49) 0.0048 (0.75)

 0.0018 (0.53) 0.0000 (0.88) 0.0000 (0.84) 0.0006 (0.86) 0.0069 (0.20)

0.0663 (0.08)n 0.0559 (0.15) 0.0282 (0.65) 0.0699 (0.18) 0.0893 (0.28)

0.0212 (0.07)n 0.0207 (0.05)nn  0.0003 (0.51) 0.0335 (0.21) 0.0218 (0.78)

 0.1107 (0.39) 0.0248 (0.22) 0.0116 (0.66) 0.0394 (0.28)  0.4051 (0.32)

 0.0010 (0.54)  0.0057 (0.41)  0.0118 (0.06)n 0.0016 (0.37) 0.0028 (0.88)

 0.0051 (0.28)  0.0105 (0.08)n  0.0141 (0.18)  0.0074 (0.36) 0.0063 (0.42)

 0.00002 (0.22)  0.0029 (0.05)nn  0.0108 (0.04)nn 0.0030 (0.74) 0.0049 (0.42)

0.0599 (0.59) 0.0556 (0.70) 0.2781 (0.22)  0.1389 (0.55) 0.0704 (0.60)

0.0229 (0.37) 0.0323 (0.55) 0.0413 (0.26) 0.0028 (0.71)  0.0024 (0.52)

0.0564 (0.57) 0.0347 (0.79) 0.3045 (0.14)  0.2065 (0.30) 0.1109 (0.39)

0.0095 (0.56) 0.0191 (0.83) 0.0321 (0.49)  0.0198 (0.61)  0.0047 (0.54)

0.0084 (0.38) 0.0159 (0.19) 0.0296 (0.17)  0.0075 (0.61)  0.0078 (0.61)

0.0039 (0.21) 0.0015 (0.17) 0.0004 (0.52) 0.0021 (0.20) 0.0058 (0.76)

0.0153 (0.67)  0.0115 (0.78)  0.0106 (0.86)  0.0238 (0.70) 0.0771 (0.27)

 0.0070 (0.77)  0.0070 (0.55) 0.0077 (0.67)  0.0296 (0.18)  0.0232 (0.65)

 0.0021 (0.96)  0.0388 (0.37)  0.0175 (0.36) 0.0933 (0.30)  0.0922 (0.41)

0.0029 (0.46)  0.0018 (0.76)  0.0114 (0.02)nn 0.0109 (0.13) 0.0066 (0.08)n

 0.0075 (0.24)  0.0132 (0.13)  0.0216 (0.19)  0.0054 (0.60) 0.0049 (0.52)

0.0029 (0.94)  0.0011 (0.27)  0.0041 (0.27) 0.0029 (0.93) 0.0090 (0.11)

0.2927 (0.04)nn 0.3428 (0.06)n 0.3860 (0.10)n 0.3650 (0.25) 0.1653 (0.36)

0.0567 (0.09)n 0.0743 (0.12) 0.0634 (0.23) 0.0573 (0.41) 0.0308 (0.52)

0.2745 (0.04)nn 0.3100 (0.07)n 0.3769 (0.10)n 0.2959 (0.31) 0.1820 (0.28)

0.0501 (0.04)nn 0.0623 (0.06)n 0.1354 (0.22) 0.0453 (0.25) 0.0101 (0.51)

0.0023 (0.83) 0.0081 (0.53) 0.0210 (0.16) 0.0026 (0.91)  0.0108 (0.53)

Long-term partnership

100

Short-term partnership

115

Joint ventures only

104

Panel B: Year þ2 Full sample

317

Strategic alliances only

217

Long-term partnership

92

Long-term partnership

100

Joint ventures only

100

Panel C: Year þ3 Full sample

294

294

Strategic alliances only

203

203

Long-term partnership

87

87

Short-term partnership

97

97

91

91

Joint ventures only

A.L. Boone, V.I. Ivanov / Journal of Financial Economics 103 (2012) 551–569

#Obs.

A.L. Boone, V.I. Ivanov / Journal of Financial Economics 103 (2012) 551–569

567

Table 9 Change in bankruptcy probability and credit rating for non-bankrupt partner. We first estimate a probit model of bankruptcy that uses all bankrupt and non-bankrupt firms with available CRSP and Compustat data in the period 1985–2007 (see Gande and Lewis, 2009; Shumway, 2001). We then use the coefficients from the model to calculate implied probability of bankruptcy for the non-bankrupt partners in the 427 instances when the counterparty to the alliance or joint venture files for bankruptcy from 1989 to 2007. Year 0 is the year of the partner firm’s bankruptcy filing. Changes in credit ratings are computed for firms with available ratings on long-term debt from Compustat. Rating ‘AAA’ is assigned a value of 1, ‘AA þ’ a value of 2, and so on, with rating the lowest credit rating, ‘D,’ being assigned a value of 20. The t-test represents the test-statistic for a change in the bankruptcy probability or credit rating since the prior year. Panel A presents the evidence for the full sample of non-bankrupt partners and Panel B presents the evidence only for the long-term strategic alliance subsample.

Panel A: Full sample Bankruptcy probability t-Test for change ¼0 No. of obs. Credit rating t-Test for change ¼0 No. of obs.

Year  2

Year  1

Year 0

Year 1

Year 2

Year 3

0.070

0.074 2.23 427

0.073  1.01 427

0.072  0.70 427

0.068  2.83 427

0.070 1.46 427

8.88 5.13 66

8.96 2.29 78

9.37 3.05 71

9.29  1.74 55

9.77 2.20 49

0.068 1.76 140

0.067  1.48 140

0.068 0.88 140

0.064  2.12 140

0.064  0.54 140

427 8.45 89

Panel B: Long-term strategic alliance subsample Bankruptcy probability 0.066 t-Test for change ¼0 No. of obs. 140

Table 10 The choice of organizational form—strategic alliance vs. joint venture. The table presents the results from a probit model of the choice between a strategic alliance and a joint venture for each of the 427 partnerships in the sample. Only non-bankrupt firms are used in the estimation. Year  1 is the year prior to the announcement of the strategic alliance/joint venture. Each type of relationship is a dummy variable equal to one if the firms enter into the respective type of agreement and zero otherwise. Relative size is the ratio of Total assets of bankrupt firm  1/Total assets of non-bankrupt alliance partner  1. Horizontal partnership occurs when the two-digit SIC code of the alliance partner and the bankrupt partner are the same. ROA is the return on assets measured as operating income before depreciation divided by total assets.VC-backed is a dummy equal to one if the firm is backed by a VC and zero otherwise. t-Satistics based on robust standard errors are reported in parentheses. n, nn, nnn indicates significance at 10%, 5%, and 1% levels, respectively. Dependent variable: 1 if strategic alliance, 0 otherwise (1)

(2)

(3)

1.067 (3.85)nnn  0.686 ( 3.53)nnn 0.259 (1.63) 0.824 (1.40) 0.429 (2.46)nn  1.605 ( 2.79)nnn

1.030 (3.69)nnn  0.759 (  3.83)nnn 0.251 (1.55) 0.755 (1.33) 0.392 (2.21)nn  1.735 (  2.98)nnn

0.924 (3.24)nnn  0.759 ( 3.75)nnn 0.257 (1.57) 0.771 (1.33) 0.420 (2.27)nn  1.619 ( 2.75)nnn

 0.080 (  2.39)nn 0.355 (0.64)  0.169 (  1.05)

 0.072 ( 1.91)n 0.596 (1.04)  0.161 (  0.99)

Type of relationship Licensing agreement Manufacturing agreement Marketing agreement Supplier–customer agreement R&D agreement Exploration agreement Firm characteristics—non-bankrupt Ln(Total assets  1) ROA  1 VC-backed Firm characteristics—bankrupt Horizontal partnership Relative size  1 ROA  1 Constant

0.454 (5.14)nnn

1.171 (3.69)nnn

Number of observations Pseudo-R-squared

427 10.7%

427 11.8%

0.167 (1.03) 0.002 (1.11)  0.665 ( 2.17)nn 0.991 (2.63)nn 427 13.2%

568

A.L. Boone, V.I. Ivanov / Journal of Financial Economics 103 (2012) 551–569

exploration agreements are more likely to form joint ventures. Second, larger non-bankrupt firms are more likely to be involved in JVs. Lastly, the better the performance of the potential (bankrupt) partner, the more likely the firms are to use JVs. Overall, the results in Table 10 suggest that both firm characteristics at the time of the inception of the relationship and the types of projects affect the choice of strategic alliance versus JV. 8. Conclusion Alliances and joint ventures provide many benefits, but also can suffer from contracting problems due to information asymmetries, verifiability of inputs and outputs, and potential expropriation, among other concerns. Some contracts contain termination rights that allow a party to terminate the agreement in certain circumstances, including bankruptcy. We examine the financial and operational performance consequences for one partner when the other one files for bankruptcy. While the bankruptcy of a weakened partner could provide the opportunity to exit the agreement and put resources to higher-valued uses, it also could have negative spillover effects due to the reduction or elimination of a oncevaluable sharing agreement. We find that the bankruptcy announcement is a significantly negative event for the non-bankrupt partner of strategic alliances, but there is not a significant wealth impact for joint venture partners. Using the length of the agreement as a proxy for the value of the relationship, further tests show that the negative effect is concentrated

primarily among the longer-lived strategic alliance agreements. Following the bankruptcy filing, we examine several dimensions of performance and investment for non-bankrupt partners. Partner firms tend to experience lower profit margins and investment in the years after the bankruptcy, which indicates that the agreement had a significant impact on operations. We do not find that the probability of bankruptcy for these firms increases following their partner’s filing. We show that industry performance of the bankrupt firm can further affect the wealth gains. Those non-bankrupt firms in horizontal agreements do better (worse) with higher (lower) industry performance. This finding is consistent with the notion that spillover effects are exacerbated when the company’s own prospects are weaker and they might not have many alternatives for finding a replacement partner. Overall, our results provide evidence that, while strategic alliances lead to benefits, the loss of those benefits when one party files for bankruptcy can lead to a reduction of shareholder wealth and worsening performance for the non-bankrupt party. The more formal structure of joint ventures appears to insulate the counterparty from the bankruptcy process since the joint venture assets are more isolated.

Appendix A. Examples of alliance and joint venture agreements See Table A1.

Table A1 Parties (bankrupt firm in bold) announcement date

Brief description

Joint venture þ vertical alliance

Palm Inc. Delphi Automotive Systems Inc. 12 October 2000

Palm Inc., Delphi Automotive, and the venture capital company Mayfield agree to create a joint venture called MobileAria. The new company will combine handheld computer and speech-recognition technologies to develop wireless services, bringing news, e-mail, and online shopping to automobiles.

Joint venture þ exploration agreement

Amax Gold Inc. Piedmont Mining Co. Inc. 19 February 1991

Amax Gold would make a cash payment to Piedmont, assume certain property and exploration obligations, and deliver to Piedmont a preliminary feasibility study by May 1, 1992. Amax Gold could then earn a 62.5% interest in Piedmont’s Haile property by paying $1,750,000 in cash and issuing one million shares of Amax Gold’s Common Stock to Piedmont. A joint venture would then be formed to complete a bankable feasibility study and, if warranted, proceed with the development of the property.

Horizontal alliance þR&D agreement

MedImmune, Inc. BioTransplant, Inc. 12 October 1995

A strategic alliance to develop products to treat and prevent organ rejection. The alliance will include BioTransplant’s anti-CD2 antibody BTI-322, MedImmune’s anti-T-cell receptor antibody MEDI-500 (T10B9), and future generations of products derived from these molecules. The companies will collaborate on research and development and MedImmune will fund and assume responsibility for clinical testing and commercialization of any resulting products. BioTransplant will receive license fees, research support, and milestones which could total up to $16 million, as well as royalties on any sales of BTI-322, MEDI-500, and future generation products.

Vertical allianceþ manufacturing agreement

Sprint Corp Recoton Corp 15 June 1998

Recoton Corp will create, manufacture, and market a line of telephone accessory products under the Sprint Corp brand name.

Type of alliance

A.L. Boone, V.I. Ivanov / Journal of Financial Economics 103 (2012) 551–569

569

Table A1 (continued ) Type of alliance

Parties (bankrupt firm in bold) announcement date

Brief description

Licensing þmarketing þmanufacturing agreement

SymmetriCom, Inc. IMP, Inc 21 July 1997

As part of the strategic alliance, SymmetriCom, which designs and manufactures a wide variety of next-generation portable power, desktop power, and data communications analog and mixed-signal integrated circuit (IC) solutions, will gain the use of IMP’s advanced process technology, manufacturing capacity, and design resources. Also as part of this strategic alliance, IMP, which specializes in high-volume manufacturing of analog and mixed-signal process technologies, has endorsed and will license many of SymmetriCom’s advanced linear and mixed-signal IC product designs. The specific products resulting from this agreement will be announced at a later date. Under this alliance, IMP and SymmetriCom will also share marketing knowledge, long-term product and process development plans, and back-end manufacturing expertise.

Supplier–customer agreement

Sun Microsystems Inc. MicroAge Inc. 25 July 1990

Under the agreement, MicroAge Inc will be authorized to distribute and support Sun SPARC-based workstations to its network of company-owned and franchised locations.

References Aghion, P., Tirole, J., 1994. On the management of innovation. Quarterly Journal of Economics 109, 1185–1207. Allen, J., Phillips, G., 2000. Corporate equity ownership, strategic alliances, and product market relationships. Journal of Finance 55, 2791–2815. Barber, B.M., Lyon, J.D., 1996. Detecting abnormal operating performance: the empirical power and specification of test statistics. Journal of Financial Economics 41, 359–399. Berg, S., Friedman, P., 1981. Impacts of domestic joint ventures on industrial rates of return: a pooled cross-section analysis: 1964–1975. Review of Economics and Statistics 63, 293–298. Brown, S., Warner, J., 1985. Using daily stock returns: the case of event studies. Journal of Financial Economics 14, 3–31. Chan, S.H., Kensinger, J., Keown, A., Martin, J., 1997. Do strategic alliances create value? Journal of Financial Economics 46, 199–221. Fee, C.E., Hadlock, C., Thomas, S., 2006. Corporate equity ownership and the governance of product market relationships. Journal of Finance 61, 1217–1251. Gande, A., Lewis, C., 2009. Shareholder initiated class action lawsuits: shareholder wealth effects and industry spillovers. Journal of Financial and Quantitative Analysis 44, 823–850. Gomes-Casseres, B., Hagedoorn, J., Jaffe, A., 2006. Do alliances promote knowledge flows? Journal of Financial Economics 80, 5–33. Grossman, S., Hart, O., 1986. The costs and benefits of ownership: a theory of vertical and lateral integration. Journal of Political Economy 94, 691–719. Hart, O., 1988. Incomplete contracts and the theory of the firm. Journal of Law, Economics, and Organization 4, 119–139. Hart, O., 2001. Financial contracting. Journal of Economic Literature 39, 1079–1100. Hennart, J.F., 1988. A transaction costs theory of equity joint ventures. Strategic Management Journal 9, 361–374. Hertzel, M., Li, Z., Officer, M., Rogers, K., 2008. Inter-firm linkages and the wealth effects of financial distress along the supply chain. Journal of Financial Economics 87, 374–387. Ivanov, V., Lewis, C., 2009. The Effect of Strategic Alliances on Initial Public Offerings. Unpublished working paper. U.S. Securities and Exchange Commission and Vanderbilt University.

Johnson, S., Houston, M., 2000. A reexamination of the motives and gains in joint ventures. Journal of Financial and Quantitative Analysis 35, 67–86. Klein, B., Crawford, R., Alchian, A., 1978. Vertical integration, appropriable rents, and the competitive contracting process. Journal of Law and Economics 21, 297–326. Kogut, B., 1988. Joint ventures: theoretical and empirical perspectives. Strategic Management Journal 9, 319–332. Lang, L., Stulz, R., 1992. Contagion and competitive intra-industry effects of bankruptcy announcements. Journal of Financial Economics 32, 45–60. Lerner, J., Malmendier, U., 2010. Contractibility and the design of research agreements. American Economic Review 100, 214–246. Lie, E., 2001. Detecting abnormal operating performance: revisited. Financial Management 30, 77–91. Lindsey, L., 2008. Blurring firm boundaries: the role of venture capital in strategic alliances. Journal of Finance 63, 1137–1138. Mathews, R., 2006. Strategic alliances, equity stakes, and entry deterrence. Journal of Financial Economics 80, 35–79. McConnell, J., Nantell, T., 1985. Corporate combinations and common stock returns: the case of joint ventures. Journal of Finance 40, 519–536. Krishnaswami, S., Pablo, E., Subramaniam, V., 2004. When is Equity More than Just a Financing? Evidence from Strategic Alliances. Unpublished working paper, Tulane University. Robinson, D., 2008. Strategic alliances and the boundaries of the firm. Review of Financial Studies 21, 649–681. Robinson, D., Stuart, T., 2007. Financial contracting in biotech strategic alliances. Journal of Law and Economics 50, 559–595. Scott, R., Triantis, G., 2006. Anticipating litigation in contract design. The Yale Law Journal 115, 814–879. Shumway, T., 2001. Forecasting bankruptcy more accurately: a simple hazard model. Journal of Business 74, 101–124. Slovin, M., Sushka, M., Mantecon, T., 2007. Analyzing joint ventures as corporate control activity. Journal of Banking and Finance 31, 2365–2382. Stuckey, J.A., 1983. Vertical Integration and Joint Ventures in the Aluminum Industry. Harvard University Press, Cambridge, MA.