The Effect of Industry Restructuring on Peer Firms

We study the bond price reaction of a merged firms peers, in order to better understand how the market responds to a restructuring. We argue that a merger announcement may signal the possibility of a merger wave to the industry, and in doing so, increase the conditional probability that peer firms might themselves be acquired in the future. However, while peer firm equity holders expect a direct benefit from a potential acquisition---in the form of a price premium---peer firm bond holders can only expect an indirect benefit---in the form of a risk reduction. Consistent with these hypotheses, we show that price reactions are stronger for firms that have a higher unconditional probability of being acquired ex-ante. In addition, we document that, cross-sectionally, the abnormal returns we observe from peer bondholders are concentrated among firms that have the highest expected risk reduction benefit from a potential acquisition. In order to distinguish a potential reduction in risk as the explicit return driver, we show that abnormal bond returns within firm (between different bond issues) are also concentrated among issues that have the highest expected risk reduction benefit.

1 Introduction positive reaction from peer equity holders on average because equity holders stand to benefit 48 from a price premium paid for shares in any potential acquisition. A company's outstanding 49 bonds, on the other hand, are not acquired during a merger since the obligation is to the 50 1 Eckbo (1983,1985); Cai et al. (2011); Ahern and Harford (2014) etc. 2 We examine the reaction surrounding the announcement of the deal, but only for deals that were ultimately completed.
3 Specifically a peer is defined as a member of the same 2 digit SIC industry as the merged firm-which in 70% of the sample contains both the acquiring, and target, firm. In the remaining cases a peer is defined as a member of the acquiring firms 2 digit SIC industry. 4 (Song and Walkling, 2000) developed this hypothesis using peer firms equity price reactions.
firm and not it's ownership; thus, bondholders have no direct benefit from a price premium 2 Literature and Hypothesis Development purchased by a relatively less risky acquiror. 5 The target bondholders experience a positive wealth effect when the default risk of the combined firm falls, and their previously risky debt 177 becomes less so. However, this implies that for a peer bondholder to experience an increase 178 in wealth they would need to anticipate that their firm had the potential to benefit from 179 a risk reducing acquisition and that it would happen regardless of the potential acquiror.

180
Given that neither of these things can be known for certain, any reaction to a merger by 181 peer bondholders is likely to be concentrated in the holders of the riskiest debt, simply be-182 cause these debt holders have a larger potential set of value increasing acquisitions. These 183 conjectures lead to the following testable hypotheses:

184
Hypothesis 1: We expect a positive abnormal equity return for the average peer firm fol-185 lowing a merger or an acquisition, due to an increase in the average acquisition prob-186 ability of peer firms. 187 To the extent that Hypothesis 1 is correct, we should expect to see that the abnormal price 188 reaction is more pronounced for peer firms that have a higher unconditional probability of 189 being acquired. All else equal, we expect that peer firms with a higher probability of being 190 acquired, ex-ante, will be those that : operate in industries in which a follow-on acquisition is 191 more likely (i.e. highly competitive industry's, or industry's in which the merger market has 192 been dormant (Shleifer and Vishny, 2003)), and peer firms which are simply more attractive 193 targets (i.e. firms with low takeover defenses (Cain et al., 2017), or high profitability levels 194 (Ravenscraft and Scherer, 1989)). 195 Hypothesis 2: Peer firms that operate in industries in which acquisition is more likely, or 196 those which are more attractive targets, should experience the largest increase in their 197 acquisition probability. The APH (Hypothesis 1) suggests that these firms will be the 198 primary drivers of any abnormal price reaction.
199 5 More recent evidence comes from Chen et al. (2020) who make use of the much more complete TRACE data to confirm these earlier results.
However, while stockholders of peer firms stand to gain directly, and uniformly, from an 200 increase in their acquisition probability and a potential merger (via the premium they would 201 receive), bondholders with the highest expected gain from a potential merger are likely 202 to be those that are more likely to be acquired and hold the riskiest debt. Billett et al. 203 (2004) and Chen et al. (2020) document that bondholders of actual takeover targets which 204 had lower bond ratings, and higher leverage ratios, relative to their acquirors, experienced 205 the largest abnormal announcement returns. These characteristics should also predict the 206 peer bondholders that will experience a wealth gain, because the uncertainty surrounding a 207 potential merger wave and it's participants is so great that only the riskiest of debt holders 208 can have any reasonable expectation that they will experience a risk reduction in the event 209 of a possible acquisition. Thus, we expect that: 210 Hypothesis 3: Peer firm bondholders that are exposed to the greatest risk, ex-ante, will 211 experience positive abnormal bond returns following the announcement of a merger 212 in their industry, because they have the largest expected benefit from a potentially 213 risk-reducing future acquisition. 214 The APH suggests that we should only expect to see an abnormal return for peer firms 215 when those peers have a high probability of being acquired in a subsequent merger. However, 216 since bondholders can't benefit from a price premium in the same manner as equity holders, 217 they only benefit from a potential risk reduction. This means that we should only expect to 218 observe positive abnormal bond returns when peer firms experience an increase in acquisition 219 probability and when those same peers have a large expected benefit from a risk reduction. 220 This is not a trivial interaction to test, because the factors which influence a peers acquisition 221 probability may not be the same factors which will result in the peers' bondholders extracting 222 a risk reduction benefit. For instance, while firms with low ratings and high leverage enjoy the 223 largest expected benefit from a risk reducing acquisition, they may not also make attractive 224 takeover targets. To the extent that peer bondholders recognize these interactions it may 225 mean that we find no (or mitigated) results, and that it will be difficult to disentangle the two effects in what results we do observe. 227 It is also possible that a great deal of consolidation in the industry during a merger wave 228 could increase the recovery risk of the assets-in-place in the event of default. Research by 229 Nozawa (2017) and Zhdanov (2007) has shown that bond returns also reflect a significant 230 amount of recovery risk, and that the absence of competitors in the market place might 231 decrease the potential recovery in an asset sale, and thus exacerbate the recovery risk faced 232 by bondholders. It follows that there is the potential for a merger wave, or the anticipation 233 of a potential merger wave, to induce a somewhat competing effect to our prediction, or at 234 least to indicate an upper boundary for the risk reduction effect we are hypothesizing. While 235 we don't observe the negative abnormal returns that would suggest that this is the dominant 236 consideration for peer bondholders, it's certainly possible that this effect might mitigate our 237 results-particularly in the cross-sections where concentration is already severe, and further 238 consolidation might make recovery risk a significant factor.

239
A unique benefit of studying bond pricing is that Hypothesis 3 provides a prediction about 240 both the firm level abnormal bond reaction, and the issue level abnormal bond reaction. If 241 this hypothesis is correct, we would expect to observe a difference between, and within, firms 242 according to the relative risk level of the portfolio, or issue, in question. Since by definition 243 the bondholders of a peer firms multiple issues will face the same acquisition probability-244 and experience the same change in the probability 6 -looking within firm provides a unique 245 test with which to distinguish these Hypotheses from each other, and from the plausible 246 alternatives that have been considered in previous literature.

247
6 Although potentially these bondholders may still have different expectations of the overall probability, and any change in that probability.

248
We also help add to a burgeoning literature that focuses on the pricing of corporate bonds 249 (Yin et al., 2018;Goldstein et al., 2019;Lin et al., 2020;Bai et al., 2019;Goldberg and 250 Nozawa, 2021), and on the effects of corporate events on bondholders (Fang-Klingler, 2019), 251 or on the joint reaction of all stakeholders (Back and Crotty, 2014;Kapadia and Pu, 2012).

252
This literature has expanded following the advent of the TRACE bond price reporting system  The APH conflicts with several alternative theories that could also potentially explain our 266 main results. The simplest alternative stems from neoclassical economics. Following a 267 merger, or a wave of mergers, the number of peer firms operating in a given industry is 268 reduced (Stigler, 1964;Bresnahan, 1989;Hackbarth and Miao, 2012). Industrial organization 269 theory would suggest that when the market is not perfectly competitive, a reduction in the 270 number of firms might increase the pricing power of the remaining firms. All else equal, this 271 should increase the profit margins for the remaining firms, which benefits all the stakeholders 272 of those firms.

273
A stronger version of this alternative posits that the peer firms which remain in operation 274 in the industry might orchestrate an increase in pricing power through explicit means. In 275 other words, less firms in the market should also make it easier for those that remain to 276 sustain, or establish, collusion (Eckbo, 1983(Eckbo, , 1985Bresnahan, 1987;Bresnahan and Reiss, 277 1991). Despite the seeming impracticality of collusion in modern times, there is evidence 278 that collusion remains an alluring option for many firms. In fact, specialty consulting firms 279 have even been formed in order to manage the web of incentives that underlie collusive 280 arrangements between firms. 7 Explicit price collusion should provide peer firms with a 281 larger increase in profit margins than implicit or simply mutually beneficial pricing. A rising 282 tide lifts all boats, and so rising profits should benefit all the stakeholders of peer firms.

283
These two, not entirely distinct, alternatives imply that positive abnormal returns for 284 peer firms' stakeholders occur because these stakeholders are anticipating some form of 285 increased pricing power, and thus increased profits, as a result of the merger. Neither 286 alternative specifically implies that peer firms' bondholders should have different reactions, 287 to implicit or explicit collusive pricing power, based on the relative risk of their bond holdings 288 before the merger. Although it's certainly reasonable to posit that improved margins may 289 provide a larger benefit to more risky firms. However, if that were the case we would expect 290 that the strongest support for either of these alternatives would be to observe the highest 291 abnormal bond returns for peer firms that operate in highly concentrated industries. The 292 more concentrated an industry, the more pricing power the firms operating in that industry 293 should have. Simultaneously, an increase in industry concentration should also make it 294 easier for firms to sustain collusion. Finally, these alternatives imply that abnormal returns 295 for peer firms should be positively correlated with the number of mergers that have occurred 296 7 In a Financial Times article from November 11, 2009, Nikki Tait writes: "A consultancy company has, for the first time, been slapped with a significant fine for cartel involvement by Europe's competition regulators -even though its alleged role was in organizing the market-rigging rather than participating directly. According to the European Commission, Zurich-based AC Treuhand prepared the operational framework for cartels involving "heat stabilizers", which are used as additives in the plastics industry. It then monitored implementation of the illegal agreements by nine companies -including Holland's Akzo, France's Elf-Aquitaine and Switzerland's Ciba." within a certain period. This should be the case because each additional merger will further 297 consolidate the industry, and greater concentration should imply greater pricing power.

298
A third alternative theory applies only in a more limited set of realizations, however it 299 also implies a positive abnormal stakeholder reaction for peer firms following a merger. If 300 the merger, or acquisition, is financed through a large amount of debt, then it might be the 301 case that the interest burden for the combined firm may constrain its cash flows. There is 302 ample evidence of leveraged buy-out deals in which the resulting firm struggles to service 303 its' debt, especially during the spate of massive LBO's in the latter part of the last century.

304
If the merged firms' peers are relatively less cash constrained, they may be able to engage 305 in price competition that drives the merged firm out of the market (Chevalier, 1995).

306
The potential for peer firms to be able to collectively force another peer from the market is 307 not necessarily valuable, to peer firm stakeholders, in and of itself. However, this ability feeds 308 into the two earlier alternatives because driving the merged firm out of the market results 309 in two less firms than existed before the merger and even greater concentration. Peer firm 310 stakeholders should benefit from this if greater concentration implies greater pricing power, 311 and higher profit margins. However, price competition is not a targeted attack on just the 312 merged firm, and it is likely to have a detrimental effect on any other peer firm with small 313 margins. Thus, at the most, we should only expect this third alternative to affect situations 314 in which the resulting leverage, for the combined firm, is greater than the industry average.

315
Further, we would expect that under this alternative, peer firms with slimmer margins should 316 be worse off than their more profitable peers.  information. Nearly all of the bond issues in TRACE are able to be matched into FISD and 332 thus enter the sample. Because TRACE reports every trade, it is necessary to engineer a 333 decision rule by which we obtain one daily price for each bond. Following Bessembinder et al.

334
(2006), we remove all trades of less than $100,000 as they tend to be non-institutional trades 335 and comprise less than 4% of the total sample. We then compute the trade-weighted average rating categories (Aaa-B) and four maturity classes (1 to 3 years, 3 to 5 years, 5 to 10 years, 347 and over 10 years). We then value-weight the matching portfolios as follows: (1) Where EBR i,t is the expected return for bond i on date t, and OBR g,t is the observed 349 return for bond g in the same rating/maturity category as bond i, which trades on the same 350 date t. M is the total number of bonds in the same rating/maturity category as bond i, 351 which trade on same day t, and w g,t is the value-weight of bond g relative to the total market 352 value of the rest of the bonds in the portfolio. We can now calculate the abnormal return 353 for bond i as follows: Where OBR i,t is the observed return for bond i, and EBR i,t is the expected return for 355 bond i calculated using the benchmark matching rating/maturity portfolios. Each firm's 356 abnormal bond return is the value-weighted average of the abnormal returns of its different 357 bond issues. Thus we calculate the abnormal return for firm k as: Where ABR i,t is the abnormal return for bond i of firm k's bond issues on date t. J is the total number of bonds outstanding for firm k, which trade on same day t, and w i,t is the market value weight of bond i relative to the total market value of the bonds outstanding 361 for firm k.

362
Two concerns remain especially relevant for this study. The first is that since corporate 363 debt still trades over the counter, a desired trade by a buyer or seller requires a party to take 364 the opposite position. The lack of a market maker or other facilitator can greatly increase the 365 time it takes to find a partner and complete a trade. In this situation, we believe that the use 366 of a larger window like [-7,7], will allow our tests to capture a turn-around transaction that is 367 a response to the merger, and yet still be small enough to mitigate the impact of any potential days before and seven days after the announcement, then a composite return allows for any 376 return windows that are a subset of [-7,7], which then includes [-7,4], [-7,5] For example any trade that occurs in one of the following windows is included in the [-7,7] return window: [-7,4], [-7,5], [-7,6], [-7,7], [-6,4], [-6,5], [-6,6], [-6,7], [-5,4], [-5,5], [-5,6], [-5,7], [-4,4], [-4,5], [-4,6], and [-4,7]. sorized at the 99.5% and 0.5% levels to mitigate the potential impact of outliers. 10 Further, because peer firm reactions to the same merger announcement may not be independent of 384 each other or of other mergers within their industry, we have clustered the standard errors 385 at the industry level in all the following analysis. In the baseline specification, we test the abnormal returns of peer firms 11 following the 388 announcement of a merger, or an acquisition, in their industry. Given this specification, we 389 show in Table 3 that following the announcement, peer firm bond prices exhibit a positive 390 abnormal return in the [-7,7] window of 4.89 basis points (bp) for all deals, which is sta-391 tistically significant at the 5% level. For deals in which the target and the acquiring firm 392 operated in the same industry prior to the announcement the abnormal bond return is 6.1 393 bp, which is also significant at the 5% level. 12

394
In order to demonstrate that the announcement of a merger has a symmetric impact on all 395 peer firm stakeholders, we also report the abnormal equity returns. However, not all firms 396 have public debt, and it may be the case that those firms which do, behave differently than 397 the universe of firms. Therefore, we restrict the sample to just those peer firms which have 398 public debt outstanding, and which also have a bond return in our window. We find that 399 following a merger announcement and in agreement with (Song and Walkling, 2000), peer 400 10 The OTC nature of the bond market means that trades must be manually recorded and reported rather than being automatically captured like stock trades. This leads to errors in bond pricing data being rather more common than in equity price data. Thus winsorizing significant outliers can help account for any errors in data recording.
11 Specifically a peer is defined as a member of the same 2 digit SIC industry as the merged firm-which in 70% of the sample contains both the acquiring, and target, firm. In the remaining cases a peer is defined as a member of the acquiring firms 2 digit SIC industry.
12 For simplicity, in the remaining analyses we will continue to use the sample of same industry deals as the main sample. However while the results are similar when using all deals, any inference drawn is less straightforward. For instance, in cross industry deals which set of peers might predict a merger wave following a merger -the acquiring firms peers, the target firms peers, or some combination of both? To avoid conflicting inferences we restrict our analysis to same industry deals. firm stock prices exhibit a positive cumulative abnormal return (CAR) in the [-7,7] window 13 401 of 0.60%, which is statistically significant at the 5% level. For deals in which the target and 402 the acquiring firm operated in the same industry prior to the announcement the CAR does 403 not change in size or significance.

404
These baseline results provide support for our first hypothesis, and despite seeming small is not surprising given our predictions, and is quite in line with Hypothesis 3. We should 422 only expect to find that the most risky firms and bondholders experience positive abnormal 423 returns and thus we expect that it is precisely those bondholder returns which are driving 424 13 Although a [-7,7] window is larger than what is typically used in equity event studies, we employ it here for consistency and comparison purposes. The results are similar for shorter and more typical windows. this overall average effect.

426
To show that peer firms experience a greater abnormal bond return if they have a larger 427 unconditional probability of being acquired we will rely on several identification strategies.

428
First, by identifying firms whose probability of being acquired is higher-whether because 429 the firms' characteristics make it a more attractive target, or because the firm operates in 430 an industry in which acquisitions are unconditionally more likely 14 . Second, by identifying 431 the mechanism through which value is created for the bondholders of peer firms. Finally, by 432 showing that this mechanism is a predictor of both firm level returns, and individual issue 433 level returns within firms.

434
This analysis is complicated by the interaction between the value creation mechanism 435 and the APH. The APH suggests that we should only expect to see an abnormal return 436 for peer firms when those peers have a high probability of being acquired in a subsequent 437 merger. However, since bondholders can't benefit from a price premium in the same manner 438 as equity holders, they only benefit from a potential risk reduction. This means that we 439 should only expect to observe positive abnormal bond returns when peer firms experience an 440 increase in acquisition probability and when those same peers have a large expected benefit 441 from a risk reduction. This is not a trivial interaction to test, because the factors which 442 influence a peers acquisition probability may not be the same factors which will result in the 443 peers' bondholders extracting a risk reduction benefit. In addition the null for these tests is 444 generally no reaction, rather than a negative reaction. To the extent that peer bondholders 445 recognize these interactions it may mean that we find no (or mitigated) results, and that it 446 14 There is some evidence that any expectation of a potential wave is not unfounded among peer firms. Within our sample more than 10% of the deals (166) were the peer of an earlier merger at some point. Table  3 in the Appendix documents that the peers who are eventually acquired in the future exhibit large and statistically significant abnormal bond returns. While there is little evidence to support the idea that these bondholders are more accurate predictors of their future acquisition than any other investor, this is evidence in support of the APH.
will be difficult to disentangle the two effects in what results we do observe.

448
Due to antitrust regulation and a generally higher level of regulatory scrutiny, the aver-  It's possible that firms in the most concentrated industries are also concerned that further 468 consolidation in the industry will exacerbate the recovery risk they face, and that this plays 469 a part in our results.

470
15 That is to say, highly competitive industries. Table 6 and Table 7 confirms the results of the univariate analy-

481
There is some evidence to suggest that more profitable firms would also make more attrac-482 tive acquisition targets (Ravenscraft and Scherer, 1989). There might be several reasons this 483 is the case. A profitable firm would be more able to cover interest payments if a large amount 484 of debt is used to finance the acquisition. Likewise a firm with high profitability might suf-485 fer from the excess cash flow problem as in Shleifer and Vishny (2003), which might make 486 them attractive "turnaround targets". Thus there is reason to expect that highly profitable 487 firms might experience an increase in their unconditional acquisition probability following 488 a merger. On the other hand, a peer firm with the excess cash flow problem might attract 489 interest from buyers who intend to use a great deal of leverage to finance the acquisition 490 under the presumption that high leverage can be sustained by the combined firm which has 491 high profit and excess cash. This would be unlikely to result in any risk reduction benefiting 492 bondholders. Thus, it is possible that we only observe an abnormal bond reaction when a 493 peer firms low profitability might make a highly levered transaction less likely. 494 We measure firm profitability as the industry adjusted profit margin (NI/Sales) using the 495 Compustat data. A univariate tercile sort in Table 5 does not exhibit a strong relationship 496 between peer firms' pre-merger profit margin and their abnormal bond returns surrounding 497 the merger. In multivariate regressions in Tables 6 and 7 the coefficient for industry adjusted 498 profit is generally negative and insignificant. In addition we observe in Columns IV-VI in 499 Table 7 that peer firms react positively to a merger announcement only when they have both 500 low rating and the lowest profit margins in the industry. This evidence suggests that while 501 a more profitable firm might indeed be a more attractive takeover target, it does not appear 502 that peer bondholders expect a potential acquisition to be risk reducing (and thus beneficial 503 for them) when their firm is highly profitable.

504
Despite that conclusion, there is some evidence that in certain cases peer bondholders 505 react positively to a merger announcement in their industry when they are highly profitable.

506
For instance, the interaction term of Rating x Profit x Initial Deal, is positive and significant 507 in Table 6 Column V and Table 7 Columns VI and IX. It's possible that an initial deal signals 508 the potential for merger wave, and that bondholders anticipate that any acquisitions which 509 occur during such a wave are more likely to be risk reducing. Regardless, it appears that 510 while peer firm profitability is likely to be a significant factor in its unconditional probability 511 of being acquired in the future, it is simultaneously a much more nuanced and uncertain 512 predictor of benefit for its bondholders.

514
A good ex-ante measure of a peer firms acquisition probability is likely to include whether 515 or not the firm operates in a state with extensive anti-takeover laws, and/or whether the firm 516 itself has incorporated any anti-takeover provisions in it's bylaws. A firm with a great deal of 517 takeover protection will be very unlikely to have a set of stakeholders with high expectations 518 for a future acquisition, and we would not expect those expectations to be greatly affected by 519 a merger in it's industry. However, takeover protections (including the decision to incorporate 520 in a anti-takeover state) are not exogenously determined, and it is certainly likely that the 521 factors which affect acquisition probability, or the potential for risk reduction, are the same In addition the inclusion of the hostile takeover measure drastically reduces overall sample size by nearly two thirds.
the probability of even more deals occurring during the wave increases but at a decreasing rate. However, as a wave progresses, the probability of subsequent mergers decreases, and 545 likely decreases at an increasing rate after each deal. Thus it should be the case that 546 early mergers, and particularly the first merger in a wave carry the strongest signal about 547 the probability of a wave and subsequent deals in the industry. Consistent with this, we 548 expect that following an initial deal, peer firms will experience a significant increase in their 549 acquisition probability. 550 We classify initial deals as those that occur after a 150 day period in which no merger 551 activity has occurred in the industry. In the dataset we identify 139 initial deals that affect 552 1042 peer firms. In keeping with Hypothesis 2, Table 4 exhibits evidence that initial deals 553 provide the industry with a strong signal about the potential for a merger wave. In the 554 [-7,7] window surrounding an initial merger peer firms experience an abnormal bond return 555 of 24.31 bp (significant at the 1% level). Following a non-initial merger peer firms experience 556 an abnormal bond return of 4.25 bp, which is still significant at the 10% level. These returns 557 are economically, and statistically, significant. Significant to the extent that given the perfect 558 merger prediction model, a representative investor could earn a profit by trading peer firm 559 debt surrounding an initial merger -even after accounting for transactions costs.

560
Multivariate regression analysis in Table 6 provides robust confirmation for the univariate 561 test in Table 4. The dummy variable for initial deal is positive and significant in specifications 562 I, II, and III, which are the most general specifications that lack any interaction terms.

563
The interaction of the initial deal dummy with other factors that may influence acquisition 564 probability (profit or competition) have the expected sign and significance. Highly profitable 565 peers of initial deals have positive and significant abnormal returns (Table 6 specification 566 IV), while peers that operate in competitive industry's which experience an initial deal also 567 have positive and significant returns (Table 7 specification I).

568
Broadly speaking, whether a deal was an initial deal appears to be the most significant 569 factor driving a shift in peer firms expectations about their acquisition probability. Whereas 570 competition is more a factor in the ex-ante level of a peer firms acquisition probability, given 571 that in the most competitive industries (which are where merger waves are most likely) one 572 deal is not likely to change the relative competition to a large degree. Finally profitability 573 appears to be largely a second order factor in peer expectations about their acquisition 574 probability, it may be significant when it interacts with other factors, but not unilaterally. largest expected benefit from a risk reducing acquisition, they may not also make attractive 593 takeover targets. This makes our hypotheses difficult to test for using normal interaction 594 terms within a standard regression analysis. We overcome this complexity by utilizing a 595 unique feature of the bond market-most firms have multiple bond issues-and the fact that 596 our third hypothesis should also predict within firm abnormal bond return differences based 597 on risk. One of the ways in which target firms can benefit from risk reduction is if they have a 600 lower bond rating than their acquiror. In this case, the target firm adopts the higher credit 601 rating, and has its risk effectively insured by the better financial position of the acquiror.

602
Ceteris paribus, this is good news for the targets' bondholders, and the target experiences 603 a positive abnormal bond price reaction. Therefore, we expect that peer firm bondholders 604 should have a bond price reaction that is inversely related to their current bond rating, 605 because firms with lower credit ratings should have higher expected gains from a potential 606 risk reducing acquisition. While it is possible that we observe some average effect according 607 to the peer firms' rating, we expect that any abnormal returns are concentrated among those 608 firms which have low ratings and have a high acquisition likelihood. Thus we expect that 609 impact of rating is much more significant when interacted with variables which help explain 610 a high acquisition probability. 611 We use Moody's credit ratings to determine the rating for peer firms. In order to focus 612 on bonds that are not in or near default we have excluded anything below B from our 613 sample, and so we only retain firms within the 6 major rating categories 18 . Univariate sorts 614 according to rating are excluded from the main text for succinctness however we observe 615 that significantly positive abnormal bond returns are concentrated among peer firms with non-investment grade bond ratings, while peers with investment grade ratings do not have a significant abnormal return 19 . A finer sort along the 6 major rating categories (Aaa-B) shows 618 that abnormal returns are largely increasing monotonically in both size and significance for 619 peer firms that have successively lower bond ratings. 620 Table 6 shows multivariate regressions on peer firms abnormal bond returns. We assign 621 an ordinal variable in place of the bond rating, so that firms rated Aaa are assigned a 1 and 622 firms rated B are assigned a 16. Column I shows the coefficient for rating is positive and 623 significant at the 10% level, which is further evidence that peer firms with poor credit ratings, 624 experience greater abnormal returns following a merger. The relative insignificance of rating 625 on it's own is not too surprising given that rating is likely to be negatively correlated with 626 peer firm acquisition probability. If this is the case, then we should expect to find stronger 627 results when interacting rating with factors affecting acquisition probability (i.e. Rating 628 x Profit x Initial Deal) and the coefficient of this interaction is significant in Column V.

629
Which is further evidence that peer firms with more to gain from a risk reducing acquisition, 630 experience greater abnormal returns following a merger in their industry. 631 Table 7 provides additional evidence for this interaction. Columns I-III show that peers 632 have a significant abnormal reaction to a merger only when they operate in highly competitive 633 industry's-which implies that their acquisition probability is high-and when their credit 634 rating is low-so that they have high expected benefit. Columns VII-IX give a sense of the 635 reaction of peer firms that should be the most likely to react to a merger. The interaction 636 for Rating x Profit x Initial Deal is highly significant and positive when the probability of a 637 hostile takeover is highest. These are the peers for which the acquisition probability should 638 be highest, and they only have an abnormal bond return when they also have a low rating 639 and their expected benefit is highest. Overall, although the interactions can be somewhat 640 convoluted, the analysis provides support for Hypothesis 3 : peer firms must be both likely 641 19 Univariate results for rating category are included in the Appendix.
to become an acquisition target, and have some expected risk reduction benefit, in order to 642 react to a merger in their industry. Another way that target firms benefit from risk reduction is if they have a higher leverage 645 ratio than their acquiror. The combined firm will emerge with a lower leverage ratio overall 646 than the target had initially, and other things equal this decreases the risk of the firm. 20

647
The targets' bondholders benefit from the reduced risk of the firm, and experience a positive 648 abnormal bond price reaction. Once again, we expect that the bondholders of peer firms 649 should have a greater reaction to an acquisition in their industry, if they have a higher 650 expected benefit through acquisition. The peer firms which are most likely to benefit from 651 this, may be those which have a high leverage ratio at the time of the announcement. 652 We calculate the industry adjusted book leverage for all peer firms using data from Com-653 pustat, and sort firms into terciles based on their leverage ratios. In Table 5, we document  However, a peer firms' leverage ratio is even more likely than their rating to negatively 660 affect their acquisition probability. Firms with very high leverage are likely to make unattrac-661 tive takeover targets, even though bondholders of such firms would have the most to gain.

662
Multivariate results in Tables 6 and 7 bear out this intuition. We observe that even though 663 20 We look at a potential leverage reduction separately from credit rating, as we would argue that it is not strictly equivalent to a potential ratings increase. That is to say, a leverage reduction does not strictly imply a ratings increase. Thus the benefit of reducing risk through a leverage reduction may be wholly separate from that achieved by a ratings reduction. the coefficients for leverage generally have a positive sign, they are almost never significant. 664 The evidence suggests that while peer bondholders of highly levered firms might benefit from 665 risk reduction if they are acquired, having high leverage probably makes them ex-ante less 666 likely to be acquired, and thus we observe much smaller abnormal bond reactions.

681
In Table 8, a firms' bond issue is denoted as its most risky if it has the longest current 682 maturity in the firms' portfolio, and it has the lowest seniority of the bonds in the firms' 683 portfolio, and it has the lowest rating of the bonds in the firms' portfolio. Conversely a 684 bond is denoted as a firms least risky if: it has the shortest current maturity in the firms' 685 portfolio, and it has the highest seniority of the bonds in the firms' portfolio, and it has 686 21 Although potentially these bondholders may still have different expectations of the overall probability, and any change in that probability. the highest rating of the bonds in the firms' portfolio 22 . We calculate the difference in the average abnormal bond return for all the firms' bond issues that qualify as most risky, and 688 the average abnormal return of all the bond issues that qualify as least risky. We report 689 that this difference is on average 11.1 basis points, and that it is significant at the 1% 690 level. This means that within the same peer firm, the bondholders of the most risky issues 691 have a stronger reaction to a merger announcement; which follows because they have the 692 largest expected gain from potential risk reduction through acquisition, while having the 693 same acquisition probability.

694
In Table 9, We regress individual issue abnormal returns against several common measures 695 of risk. We include four measures of risk in the regression : the issues' current maturity, 696 the issues' credit rating, the issues' seniority, and the issues' original offering amount, as 697 well as various interactions of those variables. So that we are only analyzing the within 698 firm differences, we also include a firm fixed effect in the regression. Our results are largely 699 consistent with the interpretation of the results in Table 8. In the base specification in 700 Column I the signs are generally consistent with our expectations, but because there is very 701 little variation within firm the results are not significant. However, in columns II, III, and V 702 we interact the various measures of risk in order to identify the issues that have higher risk 703 within a firms bond portfolio. In particular the interaction of rating, security level, and issue 704 size in Column III exhibits a significantly positive relationship with abnormal bond return 705 following a merger announcement.

706
These results strongly suggest that peer firm bondholders only expect to benefit from a 707 potential acquisition if they can benefit from a risk reduction in their risky bond holdings.

708
This distinction is observable even for bondholders that have the same probability of acqui-709 22 It should be noted that these are not the only proxies for issue risk, and that there are various other attributes that can affect the actual, or perceived, risk of an individual issue. However, these three measures are: readily available, easily quantifiable, and generally accepted measures of risk for bond issues. Creating a ranking of risk that incorporates all three should help to alleviate the concern that might arise from potentially omitting some alternative sources of risk. sition in the future, i.e. bondholders of an individual firms different issues. The expected 710 benefit from risk reduction appears to be larger for firms that also have a high acquisition 711 probability, and this relationship holds whether the peers' acquisition probability was high 712 ex-ante (because of firm or industry characteristics) or the peers acquisition probability rises 713 due to the nature of the industry merger itself. While we do observe positive abnormal bond and stock price reactions, we do not find any other evidence that might support this alternative. We show in Table 5, that peer firms 734 have the highest abnormal bond returns in the most competitive industries. However, under 735 imperfect competition pricing power is inversely related to industry concentration. Thus 736 we should expect to see the highest abnormal bond return in industries that are the most 737 concentrated (least competitive). In addition, we find in Table 8, that there is significant 738 variation across risk levels for individual bond issues within firm. If peer firms were reacting 739 to an expected increase in their pricing power, it is not clear why there should be differing 740 reactions for bondholders in the same firm with different risk levels. Finally we do not find 741 that peer firm abnormal returns get larger for successive mergers. In fact, in Table 5, we   742 show the opposite. The strongest abnormal bond reaction for peer firms occurs following an 743 initial merger, and the average reaction for subsequent mergers is much smaller in magnitude 744 and significance.

746
If the logic of imperfect competition is extended, then increased concentration in the in-747 dustry could also make it easier for the remaining firms to collude. Explicit price collusion 748 should provide peer firms with an even larger increase in profit margins, than that which 749 accompanies a simple increase in pricing power. However, with stronger assumptions come 750 stronger predictions. For instance, firms that are actively colluding are strictly better off 751 when there are less firms to collude with. Thus to the extent that a merger makes it easier 752 for the remaining firms to engage in collusive behavior, firms that are in already concen-753 trated markets should be strictly better off (Bresnahan and Reiss, 1991). Yet, we find no 754 support for this hypothesis in the data. Table 5 sorts firms by HHI, and shows that only 755 firms in highly competitive industries exhibit positive abnormal returns following a merger.

756
In fact, firms in highly concentrated industries, (in which sustaining collusion ought to be 757 easiest) experience negative, albeit insignificant, abnormal bond returns. Further, we show 758 in Table 8 and Table 9 that there are significant differences in the returns of the peer firms 759 individual bond issues. Finally, we do not find that peer firm abnormal returns get larger 760 for successive mergers. In fact, in Table 4, we show the opposite. The strongest abnormal 761 bond reaction for peer firms occurs following an initial merger, and the average reaction for 762 subsequent mergers is insignificant. These results, taken together, do not support the theory 763 that peer firms are reacting to a decrease in the cost of collusion, and given the regulatory 764 stance towards collusion in the U.S. this should not be surprising.

766
The third alternative hypothesis is largely dependent on how the combined firms' eventual 767 leverage compares to the leverage of its' peers. If the combined firms leverage ratio exceeds 768 the industry average, then ceteris paribus, it faces greater financial constraints than the 769 majority of its peers. This is because the combined firm now has higher interest payments 770 to cover, which in turn can reduce its ability to respond to competitive pressure. Indeed,

771
Chevalier (1995) shows that peer supermarkets with relatively low leverage, reduce prices 772 in an attempt to prey on rivals that have recently been acquired in a leveraged buy-out.

773
These attempts are shown to be associated with the LBO firms exiting the market. If this 774 theory is governing firm behavior, then we expect that a merger will positively affect the 775 current value of the peer firms' future cash flows, because the combined firms' excess leverage 776 weakens its competitive position, and this essentially acts as a negative shock to competition.

777
This should be accompanied by increasing peer bond, and stock, prices following the merger.

778
In addition, this alternative should imply that higher rated peer firms should have higher 779 abnormal bond returns. In the event that peer firms engage in price competition to drive 780 the merged firm out of the market, lower rated firms (who are also likely to have higher debt 781 burdens) should also struggle. In addition, this theory implies a symmetric impact across 782 all of a peer firms individual issues, because price competition occurs at the firm level.

783
If this hypothesis is correct, then the combined firms' competitive position should be most 784 affected when its leverage ratio greatly exceeds the industry average. We show that this 785 is not the case in Table 5. We construct the combined firms industry adjusted leverage 786 using Compustat data from the year following the merger completion. 23 In Table 5, we 787 sort peer firms into terciles based on the combined firms eventual leverage. We find no 788 significant relationship between peer bond returns and the combined firms eventual leverage.

789
In addition, our results on the effect of a potential risk reduction in Table 6 show that In this paper we examine the effect of a merger, or an acquisition, on the merging firms' 796 peers. Through the lens of the peers' bond prices we provide evidence that a merger leads to 797 an increase in the average acquisition probability for its peer firms within industry; or at the 798 very least we argue that our result indicate that the market participants believe this is the 799 case. We document that this leads to a positive abnormal bond price return, and confirm a 800 positive abnormal stock price return from previous work. We provide evidence that strongly 801 suggests that peer firm bondholders only expect to benefit from a potential acquisition if 802 they can benefit from a risk reduction in their risky bond holdings. However, this is not 803 a trivial hypothesis to test because the firms that are likely to make tempting targets are 804 not necessarily the riskiest firms who would benefit the most from a risk reduction. Thus 805 in cross-sectional results we show that the expected benefit from a potentially risk reducing 806 23 There is some concern, that using future data like this in our regressions, constitutes some form of lookahead bias. While we agree that this is a potential concern, we would contend that this information is, in fact, contemporaneous. At the time of the announcement, peer firms should be able to, at least, approximate the eventual capital structure of the combined firm. Thus, assuming that the bondholders of peer firms can react, at the announcement date, to the eventual leverage of the combined firm should not be an issue. acquisition appears to be larger for firms that also have a high acquisition probability, and this relationship holds whether the peers' acquisition probability was high ex-ante (because 808 of firm or industry characteristics) or the peers acquisition probability rises due to the nature 809 of the industry merger itself (because the merger was the first merger following a dry period).

810
Finally, we do not find any evidence to support the other prominent theories about the market 811 impact of a merger, or an acquisition. In fact, we provide a novel test that demonstrates that 812 these theories do not match the existing empirical evidence, by showing that support for our  This table provides summary statistics about the proposed mergers in the sample. The merger data is pulled from the SDC Platinum Database and covers the years 2004-2014. I keep all mergers that satisfy the following conditions: (1) Both the target and the bidder are public firms, (2) the deal type is merger and acquisition, (3) the deal size is greater than $50 million dollars, (4) the deal type is either completed or uncompleted. In the table: Mean Deal Size, is the average deal size by year in millions of dollars, Median Deal size, is the median deal size by year in millions of dollars, and SIC2 is the number of 2 digit SIC industries in which at least one deal was announced in a given year.  This table provides summary statistics about the peer firms in the sample. A peer firm is defined as a member of the same 2 digit SIC industry as the merged firm -which in 70% of the sample contains both the acquiring, and target, firm. In the remaining cases a peer is defined as a member of the acquiring firms 2 digit industry. Peer firms are further identified as all firms with a bond return within the sample period. In the table: Log(Assets) is the log of peer firm assets, Sales/Assets is the sales of peer firms scaled by their assets, Book Leverage is calculated as the ratio of Debt to Assets, Rating is a numerical identifier for the Moody's Bond Rating that is =1 if the bond is rated Aaa and =16 if the bond is rated B3, and Profit Margin is calculated as NI/Sales for peer firms. We report the Mean, Median, and Standard Deviation (in parentheses) sorted by year for all peer firms.  This table reports the average abnormal bond, and stock, returns for peer firms following the announcement of a merger in their industry. Bond prices are taken from TRACE and bond characteristics from FISD. Stock prices are from CRSP. We report the average abnormal bond return in Basis Points, over a window covering 7 days before and 7 days after the deal is announced. Likewise, average abnormal stock returns are reported in percentage terms over a window covering 7 days before and 7 days after a deal is announced. Bond returns have been winsorized at the 0.5% and 99.5% levels to mitigate the potential impact of errors in data recording. The standard errors in all results have been adjusted to account for clustering at the deal level. All deals contains returns for all peer firms regardless of whether the target and acquiror operate in the same primary industry before the deal is announced. Same industry deals contains returns for all firms only if the target and acquiror operate in the same industry prior to the announcement.
Peer Firm Abnormal Bond Return [-7,7  This table reports the average abnormal bond returns for peer firms following the announcement of a merger in their industry. Bond prices are taken from TRACE and bond characteristics from FISD. Initial deals are identified as those deals which occur after a dormant period of 150 days or greater. Peers are sorted into bins based on whether they have a return in an industry following the announcement of one of these initial deals. We report the average abnormal bond return in Basis Points, over a window covering 7 days before and 7 days after the deal is announced. Bond returns have been winsorized at the 0.5% and 99.5% levels to mitigate the potential impact of errors in data recording. The standard errors in all results have been adjusted to account for clustering at the deal level. t statistics in parentheses * p < 0.1, * * p < 0.05, * * * p < 0.01 These tables report the average abnormal bond returns for peer firms following the announcement of a merger in their industry. Bond prices are taken from TRACE and bond characteristics from FISD. Reported is the average abnormal bond return in Basis Points, over a window covering 7 days before and 7 days after the deal is announced. Terciles are formed at the Compustat universe level so that industries are labeled relative to the true population and not just the sample. Bond returns have been winsorized at the 0.5% and 99.5% levels to mitigate the potential impact of errors in data recording. The standard errors in all results have been adjusted to account for clustering at the deal level.

Peer Firms' Abnormal Bond Returns sorted by Industry Competition
The Herfindahl-Hirschman Index (HHI) is computed from Compustat sales figures in the standard way. Peer firms fall into bins based on how their primary industry is classified using HHI.  This table reports a regression of peer firms' abnormal bond returns surrounding the announcement of a merger in their industry. Bond prices are taken from TRACE and bond characteristics from FISD. Moody's credit rating for each bond issue is among the bond characteristics taken from the FISD database. Moody's credit rating for each bond issue is among the bond characteristics taken from the FISD database. For the purposes of this regression each credit rating is given a numeric value starting with Aaa=1, and ending with B3=16. The Herfindahl-Hirschman Index (HHI) is computed from Compustat sales figures. Individual firm leverage is the book leverage, and is calculated as the ratio of Debt to Assets, which is then Industry adjusted at the 2 digit SIC industry level. Individual firm profit is the firms profit margin, which is then industry adjusted at the 2 digit SIC industry level. Initial Deal is a dummy=1 if the deal occurred after a dormant period of 150 days or greater. Deal Size is reported by SDC Platinum database in millions of dollars. Abnormal bond returns are calculated over a window covering 7 days before and 7 days after the deal is announced. Coefficients and Standard Errors are reported in Basis Points for ease of interpretation. Bond returns have been winsorized at the 0.5% and 99.5% levels to mitigate the potential impact of errors in data recording. The standard errors in all results have been adjusted to account for clustering at the deal level.  This table reports a regression of peer firms' abnormal bond returns surrounding the announcement of a merger in their industry. Bond prices are taken from TRACE and bond characteristics from FISD. Moody's credit rating for each bond issue is among the bond characteristics taken from the FISD database. Moody's credit rating for each bond issue is among the bond characteristics taken from the FISD database. For the purposes of this regression each credit rating is given a numeric value starting with Aaa=1, and ending with B3=16. The Herfindahl-Hirschman Index (HHI) is computed from Compustat sales figures. Individual firm leverage is the book leverage, and is calculated as the ratio of Debt to Assets, which is then Industry adjusted at the 2 digit SIC industry level. Individual firm profit is the firms profit margin, which is then industry adjusted at the 2 digit SIC industry level. Initial Deal is a dummy=1 if the deal occurred after a dormant period of 150 days or greater. Deal Size is reported by SDC Platinum database in millions of dollars. Abnormal bond returns are calculated over a window covering 7 days before and 7 days after the deal is announced. Coefficients and Standard Errors are reported in Basis Points for ease of interpretation. Bond returns have been winsorized at the 0.5% and 99.5% levels to mitigate the potential impact of errors in data recording.

Level of Competition
The standard errors in all results have been adjusted to account for clustering at the deal level.
Peer Firm Abnormal Bond Return [-7,7] (I)  This table reports the difference, in the abnormal bond returns, between a peer firms' most risky bond issue and a peer firms' least risky bond issue. A firms' bond issue is denoted as most risky if: it has the longest current maturity in the firms' portfolio, and it has the lowest seniority of the bonds in the firms' portfolio, and it has the lowest rating of the bonds in the firms' portfolio. Conversely a bond is denoted as the least risky if: it has the shortest current maturity in the firms' portfolio, and it has the highest seniority of the bonds in the firms' portfolio, and it has the highest rating of the bonds in the firms' portfolio. We calculate the difference in the average abnormal bond return for all the bonds in the firms' portfolio that qualify as most risky and the average abnormal return of all the bonds that qualify as least risky. We report the average abnormal bond return in Basis Points, over a window covering 7 days before and 7 days after the deal is announced. Bond returns have been winsorized at the 0.5% and 99.5% levels to mitigate the potential impact of errors in data recording. The standard errors in all results have been adjusted to account for clustering at the deal level. With Year Fixed Effects Total 5468 11.1 * * (.0005) 2.08 * p < 0.1, * * p < 0.05, * * * p < 0.01 This table reports a regression of peer firms' abnormal bond returns surrounding the announcement of a merger in their industry. The bond returns are at the individual issue level (which means there may be several for every peer firm) and the regression includes firm fixed effects. Bond prices are taken from TRACE and bond characteristics from FISD. Moody's credit rating for each bond issue is among the bond characteristics taken from the FISD database. For the purposes of this regression each credit rating is given a numeric value starting with Aaa=1, and ending with B3=16. Deal Size is reported by SDC Platinum database in millions of dollars. Current maturity is the length of time left until the issue matures. Security Level is a range from 1 if the issue is senior secured and =6 if the issue is subordinated. Log(Offer Amount) is the log of the original offer amount for the issue. Firm fixed effects have been added to this regression so that the remaining variation is within firm. Abnormal bond returns are calculated over a window covering 7 days before and 7 days after the deal is announced.Coefficients and Standard Errors are reported in Basis Points for ease of interpretation. Bond returns have been winsorized at the 0.5% and 99.5% levels to mitigate the potential impact of errors in data recording. The standard errors in all results have been adjusted to account for clustering at the deal level. t statistics in parentheses * p < 0.1, * * p < 0.05, * * * p < 0.01