The Feedback Effect of Disclosure Externalities

Transcription

1 The Feedback Effect of Disclosure Externalities Jinhwan Kim MIT Sloan School of Management Rodrigo S. Verdi* MIT Sloan School of Management Benjamin P. Yost Boston College September 2017 Abstract We investigate whether managers internalize the spillover effects of their disclosure on peer firms, and strategically alter their disclosure decisions when doing so is beneficial. Using data on firminitiated disclosures around all-cash acquisitions, we find that acquirers originate more negative news articles during merger negotiations when there exist positive information spillovers between the acquirer and target firms (i.e., when bad news from the acquirer depress the stock price of the target). In contrast, acquirers originate more positive news articles during negotiations when there exist negative information spillovers between the two firms (again depressing the target s stock price). The strategy generates a short-lived walk-down in the target firm stock price during the period when the takeover price is determined, substantially reducing the acquisition cost for the acquirer. Our results are consistent with expected spillovers affecting the timing and content of firms disclosures. *Corresponding author contact information: 100 Main street, Cambridge, MA 02142; Phone (617) ; rverdi@mit.edu. We thank Mary Barth, Jacky Chau, Richard Crowley (discussant), Kurt Gee, Charles Lee, Devin Shanthikumar, Christina Zhu, seminar participants at the 2017 AAA annual meeting, the 2017 Deloitte Foundation/J. Michael Cook Doctoral Consortium, and seminar participants at MIT for helpful comments and suggestions. We also thank Shihui Chen for stellar research assistance. The authors gratefully acknowledge financial support from Boston College and the MIT Sloan School of Management.

2 1. Introduction We study firm disclosures and price formation in all-cash M&A transactions. Our goal is to study whether information spillovers, i.e., the extent to which information about one firm affects the stock price of another firm, influence firms disclosure decisions. Specifically, we study whether and to what extent managers recognize the potential for their firms disclosures to influence the asset prices of other firms, and strategically adjust their disclosures for gain. Our motivation is twofold: At the empirical level, prior research has documented that firms disclosures influence the stock prices of related firms (e.g., Firth (1976), Foster (1981), Freeman and Tse (1992)), i.e., that disclosure has spillovers. We extend this literature by asking the reverse question of whether spillovers influence the disclosure in the first place. Our second motivation comes from the theoretical disclosure literature which has provided many determinants of disclosure but, to the best of our knowledge, has not considered the effect of information spillover in equilibrium disclosure choices. Our central hypothesis is that acquirers will deliberately disclose information intended to affect the value of the target, when doing so is beneficial. A primary challenge to testing this prediction is finding a setting where the benefits of information spillover are strong enough to offset the other determinants of optimal disclosure choice. For instance, suppose that an acquirer s disclosure of negative news reduces the price of the acquirer as well as of the target. In this case, even if the acquiring firm benefits from the lower price of the target, the first order effect is the negative consequences of lowering the price of the firm s own stock. In such a setting we expect spillover effects to play a secondary role. Motivated by this challenge, we test our hypothesis using cash-based acquisitions and the disclosure choices made by acquiring firms. In this setting, the acquisition price is determined 1

3 during the merger negotiation period which precedes the acquisition announcement date. To the extent that the acquirer s disclosure can reduce the target s stock price, the acquirer benefits directly. 1 Further, because we focus on cash transactions the impact of the acquirer s disclosure on its own stock price is temporarily less important. 2 As a result, we expect that in such a setting information spillover can have a first order impact on the acquirer s disclosure choice. Our first hypothesis is that when the acquirer and target have positive information spillovers (i.e., when bad news from the acquirer depresses the stock price of the target), the acquiring firm is more likely to disclose negative news (e.g., by accelerating bad news or delaying good news) leading up to the M&A transaction in order to acquire the target at a lower price. Correspondingly, when the acquirer and target have negative spillovers (i.e., when good news from the acquirer depresses the stock price of the target), we predict the acquiring firm is more likely to disclose positive news (e.g., by accelerating good news or delaying bad news) during merger negotiations. Our hypothesis notwithstanding, there are reasons to expect acquirers not to strategically disclose information designed to affect the target stock price. As we elaborate on in Section 2, potential mitigating factors include both manager sensitivity to stock price movements, as well as the firm s risk of litigation. For example, if the acquiring firm manager has significant equity incentives then he may be more inclined to disclose positive news, but unwilling to disclose negative news that is likely to hurt the firm s own stock price. On the other hand, an acquiring firm 1 Implicit here is an assumption that the takeover price is a function of the target firm stock price but not a function of the acquirer s stock price. We are unaware of any study showing that the premium paid in cash deals is associated with the acquirer s stock price. 2 We argue that the acquirer s price is less relevant from the point of view of the takeover price given that the merger will be paid in cash. Of course, the manager and shareholders of the acquiring firms will certainly be concerned about the stock price because it is tied to stock ownership (which we explore in cross-sectional tests). However, since the negotiation window we investigate is relatively narrow (60 trading days) we expect that a temporary price decline over this window might also be of second order concern if it can be justified (ex post) by a better acquisition price. 2

4 subject to significant litigation risk may be more inclined to reveal negative news prior to the acquisition, but averse to initiating positive news that could leave the firm vulnerable to subsequent lawsuits. Ultimately we view acquirers strategic disclosure behavior around acquisitions as an empirical question. We base our empirical methodology on that used in Ahern and Sosyura (2014), in that our primary research design is to estimate difference-in-differences regressions using firm-deal fixed effects. 3 This design allows us to compare the changes in disclosures during versus before merger negotiations for acquirers with strong information spillovers (our treatment sample) relative to acquirers with little or no information spillovers (our control sample). Including firm-deal fixed effects controls for time-invariant differences both at the firm and deal levels, and thus ensures a design robust to a wide array of observable and unobservable factors (e.g., firm size, industry, past disclosures, historical performance, method of payment, etc.). Overall, this methodology identifies whether acquirers with strong information spillovers modify their disclosures relative to acquirers with no information spillovers during the short window when they stand to benefit from reductions in the target stock price. To implement our difference-in-differences regressions we create two treatment samples: (i) firms with positive spillovers in which good news for the acquirer is also good news for the target, and (ii) firms with negative spillovers in which good news for the acquirer is bad news for the target. These firms are benchmarked to a control sample of firms with little or no spillover in which news for the acquirer is irrelevant for the target. We then test whether the acquirer discloses more negative information during the deal negotiation period when the acquirer s disclosures have 3 While we build on their methodology, our research question differs from that in Ahern and Sosyura (2014). Using a setting of stock-for-stock acquisitions, they provide evidence of acquiring firms disclosing positive news designed to drive up the value of their own stock. In contrast, we aim to show that the disclosure decision is also influenced by information spillovers with the goal of driving down the value of the target. 3

5 a positive spillover effect on the target firm (relative to control firms). Alternatively, we test whether the acquirer discloses more positive information during negotiations when the acquirer s disclosures have a negative spillover effect on the target firm (again relative to control firms). Our primary sample consists of 464 cash-based acquisitions announced during the period 2000 to We measure information spillover between each acquirer-target pair using the pairwise correlation of daily stock returns during the 120 trading days preceding the start of the pre-negotiation period. We classify disclosure into good and bad news based on the tone of press releases initiated by the firm using the classification of positive and negative words from Loughran and McDonald (2011). In tests of our first hypothesis, we find that relative to deals with low information spillovers, acquirers with positive information spillovers disclose more negative information during merger negotiations by initiating 10.7% more bad news press releases and 6.4% fewer good news press releases. In contrast, acquirers with negative information spillovers do the opposite: they initiate 25.4% fewer bad news press releases and 15.7% more good news press releases during negotiations. Both results are consistent with our hypothesis that acquirers alter their disclosure to attempt to drive down target firm value during the time period when the firms are negotiating the takeover price. Interestingly, when we test for the difference in the effect for firms with positive and negative spillovers we find that the effect is stronger for negative spillover. This suggest that spillover effects are stronger when they do not come at the expense of a lower valuation of the disclosing firm. Next, we conduct a cross-sectional test to examine our prediction that potential mitigating factors such as acquiring firm CEO sensitivity to stock price movements affects the willingness to exploit information spillovers. Specifically, we examine whether the change in acquirer 4

6 disclosures during merger negotiations is affected by CEO equity incentives, measured as the dollar change in the CEO s equity portfolio value for a one percent change in the firm s stock price (i.e., portfolio delta). We find that CEOs with higher equity incentives are less willing to generate negative disclosures during merger negotiations than CEOs with lower equity incentives, but more willing to generate positive disclosures to exploit information spillovers. When then test whether the changes in disclosure during the negotiation period persist or reverse during the post-period to test whether our findings are driven by temporary disclosure incentives during the negotiation period or some permanent confounding factor. We find evidence of strong reversals in disclosure tone for acquirers with both positive and negative information spillovers after the negotiation period (i.e., after the M&A price is set). Further, the magnitudes of the disclosure reversals are comparable to the magnitudes of the abnormal negotiation period activity. The disclosure reversals highlight the temporary deviations in disclosure behavior during the negotiation period and also helps mitigate concerns that our findings capture some unobserved factor (say deteriorating performance) around the negotiation date. We next turn our attention to the valuation implications of the acquirer s disclosure choices. We show that strategic disclosures by acquirers with strong spillovers appear to depress the target s cumulative returns in the run-up period to the merger announcement. Compared to the total acquisition premium paid in which there is zero information spillover between the acquirer and target firms, bidders with strong (negative or positive) information spillover appear to purchase the target at a % lower premium (depending on the specification). In order words, while the average takeover premium in our sample equals 43.3%, the average premium for firms with negative (positive) spillover equals 38.4% (33.6%). For the average deal in our sample this 5

7 translates into dollar savings of approximately $20 million ($59 million) in the acquisition price, or 0.6% (2.3%) of the average acquirer s market equity. To better attribute our findings to the acquirer s spillover incentives, we perform three additional tests. First, we examine the target s (as opposed to the acquirer s) disclosures during the negotiation period. In this case, while the transaction is the same for the acquirer and the target, the nature of the spillover incentives are quite different. Specifically, in a cash deal the target s incentives is to maximize its stock price, independent of the acquirer s stock price. In other words, the target s incentive is to disclose more favorable information regardless of the nature of the spillover between the two firms. Consistent with this argument, we find that during merger negotiations, target firms uniformly disclose more positive news. Further, this behavior is independent of the nature of information spillovers with the acquiring firm. Second, we examine changes in the tone of news initiated by third parties (as opposed to news initiated by the acquiring firm) during negotiation periods. To the extent that firms have relatively less control over news they do not initiate, and that third party news providers (e.g., newspapers) do not benefit from the information spillover on the target s stock price, we do not expect to see a pattern of strategic behavior designed to influence the value of the target firm. On the other hand, if third party providers simply disseminate information provided by the firm (as Ahern and Sosyura (2014) show) then we would still find evidence of strategic behavior in news disseminated by third parties. Consistent with the former argument, we find no evidence of strategic changes in disclosure tone in media-initiated articles around merger negotiations, nor an effect of information spillovers. Finally, in a third falsification test, we examine acquirer disclosures in stock-for-stock transactions. Building from the results in Ahern and Sosyura (2014), we predict that when 6

8 acquirers use their own stock for payment, the incentive to inflate their own stock price will dominate the incentive to reduce the target s price by exploiting information spillovers. As expected, we find that acquirers in stock-based acquisitions generate more positive news during merger negotiations, and seem relatively insensitive to their spillover incentives. Overall, these three tests provide additional assurance that our primary results do not simply reflect omitted factors driving the tone of news in the deals we investigate. Our finding that information externalities influence a firm s disclosure decision contributes to several streams of literature. First, we extend the literature on information spillovers. Studies in this area have focused primarily on the economic impact of disclosure on peer firms. These studies typically take disclosure and its associated externalities as exogenous and investigate relevant market and corporate outcomes of affected peers. 4 In contrast, we reverse the analysis and study the extent to which spillovers affect the disclosure. Our study views managers as strategic players who incorporate the expected spillover effects when deciding the optimal disclosure choices around merger deals. Our study also contributes to the large body of work that investigates the incentives that shape and drive voluntary disclosure, in particular the disclosure of negative news. For example, prior research has shown that firms have incentives to disclose negative news when such disclosure reduces litigation costs (Skinner, 1994, 1997) and when firms have temporary incentives to lower their stock prices (Aboody and Kasznik, 2000). In contrast, we investigate a setting where information spillovers can provide a rationale for firms to strategically disclose negative news because such disclosure, for firms with positive spillovers, can benefit the disclosing firms (in this case the acquiring firm) by lowering the target s stock price. 4 Empirical evidence shows that disclosure externalities have liquidity (Bushee and Leuz (2005)), investment decision (Badertscher et al. (2013), Chen et al. (2013), Shroff et al. (2014)), and cost of capital (Shroff et al. (2017)) effects. 7

9 Last, our paper relates to the large body of work that seeks to understand the wealth effects of mergers (see Eckbo (2014) for a review). Recent work in this area focuses on examining the determinants underlying the cross-deal variation of the relative wealth gains (or losses) that acquirers and targets realize (e.g., Savor and Lu (2009), Cai, Song, and Walkling (2011), Betton et al. (2014), Seru (2014), Wang (2016)). In addition, this study builds on Ahern and Sosyura (2014), who find that bidders in fixed exchange ratio stock mergers tend to strategically disclose positive news to inflate their own stock prices. Our paper complements this finding by demonstrating that bidders adjust their disclosure decisions to influence not only their own stock prices, but also the stock prices of their targets. By demonstrating that disclosure externalities enable acquirers to acquire targets at a discount, the findings of this paper reveal a previously unknown factor that influences the relative wealth gains during mergers. The rest of the paper proceeds as follows. Section 2 motivates our setting and develops our hypotheses. Section 3 discusses the data and outlines our empirical strategy, whereas Section 4 presents our main results. Section 5 concludes. 2. Research Setting and Hypothesis Development Our goal is to study whether information spillovers, (i.e., the extent to which information about one firm affects the stock price of another firm), influence firms disclosure decisions. A primary challenge, however, in empirically testing our prediction is that it is crucial to identify a setting where the benefits to a firm of exploiting those externalities outweigh the other determinants of optimal disclosure choice, such as maximizing its own short-run value. Motivated by this challenge, we examine cash-financed mergers and acquisitions as our setting to investigate the feedback effect of information externalities on a firm s own disclosures. 8

10 Cash-based M&A transactions offer several desirable features. First, negotiations regarding the takeover price resemble a zero-sum game in which the acquirer gains from a lower target stock price (i.e., the acquirer has a clear motivation to reduce the target stock price if possible). Second, using cash (as opposed to its own stock) to finance the transaction ensures that the acquirer remains relatively insensitive to any temporary decline in its own price. Third, the merger negotiation period is relatively short-lived (on average, beginning 60 trading days prior to the acquisition announcement), providing a limited window during which the acquirer must strategically disclose information. And fourth, market participants outside of the bidder and target firms are generally unaware of the merger negotiations while they are taking place, so the market likely perceives the disclosures as credible. In this setting we conjecture that the existence of information spillovers between the acquiring and target firms provides an incentive for acquirers to strategically disclose information to reduce the takeover price of the target. As a result, if the two firms have a positive information spillover (i.e., when bad news from the acquirer depresses the stock price of the target), the acquirer may seek to drive down the target s market value by disclosing negative news. Similarly, if the firms have a negative information spillover (i.e., when good news from the acquirer depresses the stock price of the target), the acquirer may seek to decrease the target s value by disclosing positive news. These arguments lead to our first set of hypotheses: H1a: Bidders in cash-financed acquisitions initiate more negatively-biased press release articles during merger negotiations when there exist positive information spillovers between the bidder and target firms. H1b: Bidders in cash-financed acquisitions initiate more positively-biased press release articles during merger negotiations when there exist negative information spillovers between the bidder and target firms. 9

11 Further, if acquirers are able to successfully exploit information spillovers with targets, then we should see an impact of the acquirer s disclosure on the target stock price during the merger negotiations. In other words, the presumed goal of bidder firms is to acquire the target firm at the lowest possible price, and H1a and H1b hypothesize that the acquirer would adjust the disclosure according to the nature of any information spillovers. Thus, although H1a and H1b make opposite predictions for disclosure tone among acquirers with positive and negative information spillovers, H2 makes the same prediction for all bidders which share information spillovers with target firms. In particular, our second hypothesis is as follows: H2: Targets in cash-financed acquisitions receive a lower offer premium in merger negotiations when there exist positive or negative information spillovers between the bidder and target firms. Notwithstanding the above discussion, there are several reasons why we might not observe our predicted outcomes. First, it is not clear ex ante whether any potential gains would outweigh the costs of intentionally guiding disclosures to influence the value of other firms. For instance, the potential economic gains of acquiring the target at the lowest possible price may be sufficient motivation for bidders to disclose good news (to exploit negative spillovers with the target) because it would have a positive influence on the acquirer s stock price, it is less clear whether bidders would disclose bad news (to exploit positive spillovers with the target) at the cost of reducing their own stock s value. Second, if the target firm s shareholders can unravel the strategic disclosures of the bidder, the target s stock prices will not react to bidder disclosures and will thus eliminate the incentive for bidders to disclose strategically. Third, if the target firm manager is able to perfectly observe the true value of his firm irrespective of market prices, bidders may be unable to negotiate a reduced takeover price and hence, have no incentive to disclose strategically. 10

12 Finally, if purchasing target firms at a discount increases litigation risk for acquirers ex post, this may be sufficient disincentive to prevent strategic disclosure by the acquirer. 3. Data and Methods 3.1 Background of the Merger Data To construct our sample, we start with completed mergers of U.S. publicly traded firms with deal values exceeding $10 million USD in the SDC database announced between January 1, 2000, and December 31, 2014, and we require the form of payment to be entirely cash. We begin the sample in January 2000, since Factiva s news coverage (used to measure disclosure tone as described below) is less comprehensive prior to this date. We exclude withdrawn merger bids because we require merger agreement documents to identify key dates in the negotiation process (i.e., the negotiation start dates). For a randomly selected subsample of deals, we retrieve information about the negotiation start dates of the merger process from SEC filings. 5 This information typically appears in the section entitled Background of the Merger in the merger agreement, which provides a narrative history of the merger process. In particular, we collect the date when the merging firms first discuss the potential merger and the period over which the target s stock price per share is determined. We find that 60 days is the average length of the time between the start of negotiations and the acquisition announcement date. Based on this hand-collected sample of negotiation start dates and the public announcement dates available on SDC, we define the following periods: (1) Prenegotiation Period: the 120 trading days immediately preceding the day on which the merger negotiations begin. 5 We search the following forms in the given order until we find the relevant data: 10-K, 8-K, DEFM14A, DEFA14A, DEF14A, PREM14A, PRER14A, S-4/A, S-4, 424B3, 424B2, F-4, 497, N-148C/A, SC13E3, SC13E3/A, and SCTO- T/A. Most deals negotiation dates are available in the 10-Ks. 11

13 (2) Negotiation Period: the 45 trading day window stretching from t=-60 to t=-16 days relative to the public announcement of the merger. (3) Post-Announcement Period: the 40 trading day window following the acquisition of the public announcement date. We use a window spanning t=+2 to t=+41 relative to the public announcement of the merger. We restrict the negotiation period to end 15 days before the public announcement to ensure that any firm-initiated press-releases do not contain information about the merger. In all cases, it is important to note that these dates are before there has been any public acknowledgement of the merger and are only realized ex post by reading the SEC filings. Further, we use a standard negotiation period length for all deals based on the average for the subsample of negotiation start dates we hand-collect. We opt to use an average negotiation period because in addition to being labor-intensive, we find that the process of determining negotiation start dates is subjective and somewhat noisy, as it is often unclear at what point serious talks begin between firm representatives. Figure 1 illustrates the sequence of time periods in a typical merger. 3.2 Firm-originated News Data Firm-originated news data is collected from the Factiva database. To study a firm s disclosure strategy around mergers, we collect all articles at the daily frequency from the prenegotiation period through the close of the merger from the top three newswire sources: Reuters News, Dow Jones News Services, and Business Wire. We use newswire articles to capture news that is firm-originated as these articles typically report firm press releases with no additional analyses. Moreover, press releases may be particularly suitable for strategic disclosure because they are typically less regulated than accounting statements, providing greater flexibility for firms to steer news content in a desired way (e.g., Dyck and Zingales 2003). 12

14 To identify strategic disclosure behavior, we categorize each of the words in the articles as either positive, negative, or neutral following Loughran and McDonald (2011). We then classify each day as a Good News Day, Bad News Day, or neither based on the overall tone of the day s news. To establish a benchmark news tone for the entire sample, we use the average gap between positive words and negative words in the day s disclosures. Then we compute the gap between positive and negative words on a daily basis, and label the day as a Good News Day (Bad News Day) if the positive minus negative words gap is above (below) the sample average. 6 Further details on the construction of key variables are provided in the Variable Appendix. 3.3 Identification Strategy The central premise of this paper is that acquirers make strategic disclosure choices to exploit gains from their disclosure externalities during mergers. To identify the causal relationship between making an acquisition and strategic disclosure behavior, we must address selection bias and endogeneity. First, the act of acquiring is not randomly assigned to a firm. Instead, there may be something characteristic about acquirers that leads them to either disclose bad or good news that is unrelated to the merger. For example, firms in industries that experience negative industry shocks may release more bad news and also conduct more mergers as potential targets become cheaper. In this case, an omitted firm-level characteristic, such as negative industry shocks, may cause a spurious correlation between the timing of a merger and the increased disclosure of bad news. More generally, it is likely that both observable and unobservable firm-level characteristics may simultaneously determine disclosure tone and merger outcomes. As a first step to mitigating 6 Our results are qualitatively similar if we define disclosure tone at the article level rather than at the daily level. For example, if we use dependent variables Good (Bad) News Articles instead of Good (Bad) News Days. 13

15 this concern, we use a firm-deal fixed effects difference-in-differences approach to control for any time-invariant firm characteristics as follows:,,,,, (1) where i and t index firms and time periods, respectively., is a measure of the overall sentiment, or tone, of disclosures related to firm i in period t., is an indicator variable for firm i during period t that takes the value of one if the observation occurs after merger negotiations have begun and zero if it occurs before negotiations have begun. ( ) is an indicator variable that equals one if the correlation between acquirer and target returns is significantly positive (negative) and zero otherwise (described below). The inclusion of both positive and negative spillover groups is motivated from our main prediction that strategic disclosure behavior should occur on both ends of the extreme spillover groups. Note that our empirical design compares the disclosure behavior of the extreme spillover groups against deals that are neither positive nor negative. We refer to this group as the zero spillover group, which serves as our control group throughout. All variables are described in detail in the Variable Appendix. is a firm-deal fixed effect, which captures any time-invariant characteristic of the acquirer or the merger, including observables such as historical disclosure tone, as well as any unobservable firm characteristics that are stable during the relatively short time of the merger negotiation. The coefficient on the interaction term, (i.e., captures the difference-in-difference estimate of the positive spillover acquirer s disclosure behavior relative to the zero spillover group during the negotiation period. Similarly, the 14

16 coefficient on, (i.e., measures the changes in disclosure behavior of the negative spillover acquirer s relative to the zero spillover group. To assess whether the acquirer and target returns demonstrate significant covariance (i.e., whether they are in the Positive Spilloveri or Negative Spilloveri groups), we obtain the correlation of the bidders and targets daily returns adjusted for the value-weighted market index over the 120 trading day window ending the day before the start of the pre-negotiation date. If the correlation is positive (negative) and more than one standard deviation above (below) zero, we classify the deal as being between firms with positive (negative) spillovers. Overall, the inclusion of the firm-deal fixed effect in Equation 1 alleviates much of the concern that omitted firm or deal characteristics are driving the acquirer s disclosure strategies, and isolates the effect of information spillovers on differences in disclosure behavior as the merger negotiations proceed. Another advantage of this design in terms of identification is that we make, as discussed in section 2, completely opposite directional predictions for the two extreme spillover groups (i.e., In other words, for potential endogenous factors to pose identification issues, (e.g., economy-wide shocks) they would have to affect the two extreme spillover groups in opposite directions. This nuanced prediction helps to reduce certain endogeneity concerns. Nonetheless, to further corroborate the inferences of our main findings, we perform a series of falsification tests to rule out other plausible interpretations of our results. We discuss the additional results in more detail in section 4. Last, restricting our attention to mergers that use only cash helps avoid endogeneity concerns related to the choice of financing, such as the acquirers decision to use cash versus stock (Andrade et al., 2001; Fuller et al., 2002; Moeller et al., 2007). Hence, exploiting variation in 15

17 information spillover within cash mergers provides an arguably clean setting to identify the feedback effect of disclosure externalities. 3.4 Summary Statistics Table 2 displays relevant firm and disclosure characteristics for firms with positive, negative, and zero spillovers. The partitioning strategy described above results in a zero spillover group containing 245 mergers, a positive spillover group containing 181 mergers, and a negative spillover group containing 38 mergers. The mean correlation of the daily returns for the acquirer and target firms is 0.01 for the zero spillover group, which is significantly different from that for both the positive spillover group (0.23) and the negative spillover group (-0.18). The average market capitalization of acquirers in the zero spillover group is $2.87 billion, which is not statistically different from the average acquirer size in either the positive or negative spillover group. The average market capitalization of target firms in the zero spillover group is $443 million, which is somewhat smaller than the target size in the positive spillover group ($613 million), but not significantly different from the targets in the negative spillover group ($411 million). Table 2 also shows that slightly more than half (54%) of the deals in the zero spillover group involve acquirers and targets in the same industry, where the industry is defined at the SIC 2-digit level. This compares with 67% (61%) of the positive (negative) spillover deals involving acquirers and targets in the same industry. 7 Table 2 also contains mean values for the disclosure variables created from Factiva press release data during the prenegotiation period. Good - Bad News Day Ratio represents the number 7 Although the target firms in the positive spillover group are relatively large and both the positive and negative spillover groups contain deals in which the firms are more likely to be in the same industry, we do not expect our results to be affected by this heterogeneity. The short time horizon of our analysis allows the firm-deal fixed effects to absorb most of the difference in target market capitalization and industry similarity. 16

18 of good news days minus the number of bad news days, divided by the total number of days in the period (e.g., pre-negotiation, negotiation, post-announcement). The zero spillover group has an average Good - Bad News Day Ratio of 0.109, which is statistically indistinguishable from the average values for the positive and negative spillover groups (0.104 and respectively). When partitioned into good news days and bad news days separately, we find that zero spillover group acquirers have a mean Good News Day Ratio of and a mean Bad News Day Ratio of To translate this into days, consider that the negotiation period is 45 days long, which suggests an average of nine days dominated by good news, and four days dominated by bad news. 4. Results In this section, we first investigate the change in the tone of acquirer press releases during negotiations for positive and negative spillover acquirer-target pairs. Second, we examine whether there is a reversal in the tone of acquirer press releases during the post-announcement period. Third, we investigate whether abnormal acquirer disclosures during the negotiation period appear to successfully decrease the value of target firms within the positive and negative spillover groups. And last, we perform falsification tests to further validate the inferences of our main findings. 4.1 Changes in acquirer disclosures during merger negotiations Hypothesis 1 predicts that acquirers will recognize the potential consequences of their disclosure externalities and strategically exploit them. Thus, we expect acquirers with positive information spillovers to strategically disclose negative news during merger negotiations to improve the terms of the merger by driving down the target s value (H1a). In contrast, we expect acquirers with negative information spillovers to disclose positive news during merger 17

19 negotiations (H1b), also with the aim of reducing the target s value and obtaining a more attractive takeover price. Figure 2 presents evidence consistent with these predictions. This figure presents a comparison for the three groups of firms (positive, negative, and zero spillover acquirers) of the cumulative ratio of good minus bad news days based on press articles issued. As described previously, the Good News Day and Bad News Day variables are constructed based on the relative count of positive and negative words in a firm s disclosures during the day. To assist in comparing the time trends, we adjust each group s disclosure tone during the negotiation period for that group s average disclosure tone during the pre-negotiation period. The figure shows that the overall sentiment of press releases become markedly more positive for negative spillover bidders around the start of the negotiation period. In contrast, the tone becomes cumulatively more negative during the negotiation window for positive spillover bidders. Though the differences in press release tone by the extent of information spillovers revealed in Figure 2 are suggestive, they are not statistical tests. In Table 3, we present coefficient estimates from multivariate fixed effects difference-in-differences tests as in Equation (1) for the all-cash merger sample. For ease of interpretation, we collapse deal-day observations within each period (i.e., pre-negotiation and negotiation windows) into single observations. The first column of Table 3, Panel A displays the overall change in press release tone during the disclosure period for the entire sample, without considering partitions for positive and negative spillover groups. The coefficient for Negotiation Period indicates that on average, acquirer press release sentiment improves during negotiations, although the effect is insignificant (coef.=0.005; t-stat=1.24). 18

20 In the second column we examine the change in press release tone for positive spillover acquirers separately from the other sample deals. The coefficient on the interaction term Negotiation Period Positive Spillover is negative and significant (coef.=-0.022; t-stat=-3.26), indicating that positive spillover acquirers are more likely to disclose negatively-biased news than zero and negative spillover acquirers during merger negotiations. In the third column, the coefficient on Negotiation Period Negative Spillover (coef=0.045; t-stat=3.23) shows that the opposite result holds for negative spillover acquirers. In terms of economic magnitude, relative to the zero spillover group, Good - Bad News Day Ratio for positive spillover acquirers drops by 16.3% during merger negotiations. On the other hand, acquirers in the negative spillover group experience a 33.3% increase in Good - Bad News Day Ratio relative to the zero spillover group benchmark, during the negotiation period. The final column shows that these results remain similar when we examine press release sentiment for both the positive and negative spillover acquirers relative to the zero spillover group. Collectively, these results indicate that positive spillover acquirers initiate more negatively-biased news press releases during merger negotiations, whereas negative spillover acquirers initiate disproportionately more positively-biased news during negotiations. Before we proceed, we note that the change in disclosures during merger negotiations is substantially stronger for acquirers in the negative spillover group (i.e., those for whom disclosing positive news is likely to drive down the value of the target firm) than for those in the positive spillover group (p-value=0.06 for the difference in the coefficients for each group). While we did not have an ex ante prediction about these relative magnitudes, this result is consistent with acquirers having a greater willingness to exploit information spillovers when there is a lower cost to doing so. 19

21 In Panel B of Table 3, we assess whether the changes in overall press release sentiment discussed above are driven by changes in good news, bad news, or both. In the first column of Panel B, we find the coefficient for Negotiation Period Positive Spillover is significantly negative (coef.=-0.012; t-stat=-2.53), indicating that positive spillover acquirers issue significantly less good news during merger negotiations. The second column displays the reverse result for negative spillover acquirers (coef.=0.028; t-stat=2.94), indicating that these acquirers issue significantly more good news during negotiations. Using zero spillover acquirers as a benchmark group, positive spillover acquirers reduce their good news disclosure by 6.4% and increase their bad news disclosure by 10.7% during merger negotiations. Columns 3 and 4 shows that these signs flip with respect to bad news disclosures: positive spillover acquirers issue significantly more bad news during negotiations, whereas negative spillover acquirers issue significantly less bad news. In this case, negative spillover acquirers increase their good news disclosure by 15.7% and decrease their bad news disclosures by 25.4%. Overall, these results are consistent with hypotheses H1a and H1b, and suggest that acquirers strategically disclose both positive and negative news during merger negotiations to affect target firm value and thereby obtain better terms of trade. 4.2 Cross-sectional test of the relation between spillovers and acquirer disclosures A key feature of our all-cash acquisitions setting is that it provides a window during which we expect the impact of the acquirer s disclosure on its own stock price to be less important because the acquirer s stock is not used as currency to pay for the transaction. However, other incentives (such as compensation incentives) will still be present and might serve as a counterforce to the incentives to exploit information spillovers. To test this prediction, we conduct a cross-sectional test in which we vary the acquiring firm CEO s wealth sensitivity to movements in the firm s own stock price. Specifically, we explore 20

22 whether the strength of the acquirer s disclosure response depends on the acquiring firm CEO s equity incentives. Following prior studies (e.g., Core and Guay 1999; Erickson, Hanlon, and Maydew 2006; Armstrong, Jagolinzer, and Larcker 2010; Jayaraman and Milbourn 2015), we measure CEO equity incentives as the portfolio delta, defined as the dollar change in the CEO s equity portfolio value for a one percent change in the firm s stock price. The first column of Table 4 presents the results of estimating Equation 1 for all transactions in which we have data to compute CEOs equity incentives. Though the sample is reduced to 357 deals, the economic magnitudes of the disclosure responses are quite similar to those in Table 3. The second and third columns of Table 4 display the results of estimating Equation 1 after partitioning the sample into groups of acquiring firm CEOs with low (below-median) and high (above-median) equity incentives. The results indicate that CEOs with low equity incentives exploit information spillovers by issuing bad news when there is a positive spillover between the acquirer and the target. In contrast, CEOs with high equity incentives exploit information spillovers by issuing good news when there is a negative spillover between the acquirer and the target. In other words, CEOs are more willing to exploit the information spillover channel by releasing negative information when they are personally less sensitive to a decline in the firm s own stock price, and they are more inclined to exploit information spillovers by releasing positive information when they personally benefit more from stock price increases. 4.3 Disclosure reversals in the post-announcement period Though the results above are consistent with acquirers affecting the value of targets through strategic exploitation of disclosure externalities, concerns remain that the disclosures could be driven by some other unobserved factor, such as deteriorating performance. To help rule out the 21

23 possibility that the disclosures reflect a change in underlying fundamentals, we next investigate how they change in the period following the public announcement of the merger. Figure 3 displays the trend in acquirer disclosure tone in the period following the acquisition public announcement. Similar to Figure 2, we adjust each group s disclosure tone for that group s average tone during the negotiation period. The figure shows that in the postannouncement period, the trends in press release sentiment flips for positive and negative spillover bidders: sentiment declines for negative spillover bidders while trending positive for positive spillover bidders. This pattern indicates a reversal in acquirer press release sentiment compared to the trends during merger negotiations. In Table 5, we present results from the statistical analyses of post-announcement disclosure trends. Consistent with acquirers strategically steering the tone of their disclosures during the negotiation period, Table 5 Panel A provides evidence of a reversal in the disclosure tone for both positive and negative spillover acquirers. The coefficient for Post-Announcement Positive Spillover in column 2 is positive and significant (coef.=0.018; t-stat=2.02), which is the reverse of the effect we document during the negotiation period in Table 3. This indicates that following merger negotiations, acquirers in the positive spillover group increase their output of good news, relative to zero spillover acquirers. Similarly, the negative coefficient for Post-Announcement Negative Spillover in column 3 indicates that acquirers in the negative spillover group increase their output of bad news, once the merger talks are over which again is the reverse of the behavior we document in Table 3. It is worth noting that in terms of economic magnitude, the size of the disclosure tone reversals in Table 5 are comparable to the size of the negotiation period trends documented in Table 3. 22

24 Overall, the results in Table 5 indicate that acquirers abnormal disclosure behavior during merger negotiations largely reverses following the acquisition announcement. These findings help to mitigate concerns that our results in Table 3 are simply capturing some unobserved factor, such as deteriorating performance, around the negotiation date. 4.4 Target values during merger negotiations A positive (negative) relation between the disclosure content of the acquirer and the market value of the target is implicitly assumed in the development of H1a (H1b) which predicts that acquirers actively exploit disclosure externalities. If the acquirer s underlying intention for strategically altering disclosure tone during the negotiation period is to purchase the target at a discount, then we should expect to see target stock prices decline during the negotiation period in equilibrium (our H2). We test this hypothesis by first providing a graphical representation of the target s price movement leading up to the merger announcement for firms with negative, zero and positive information spillovers. Figure 4 shows that the value of targets in the zero spillover benchmark group experience a steady rise in value leading up to the announcement, consistent with the typical trend documented in prior literature (Eckbo (2014)). In contrast, for deals with either strong positive or negative spillovers the target firm values appear to be suppressed throughout the negotiation window. This provides initial evidence consistent with our H2. In Table 6, we provide statistical evidence consistent with Figure 4. Specifically, the difference-in-differences estimates indicate that acquirers in positive and negative spillover deals are able to largely prevent the typical run-up in targets value during the negotiation period. Whereas target firms in the zero spillover benchmark group experience average cumulative 23

25 abnormal returns (CAR) of 7.6% during this time, targets subject to acquirer spillovers experience a CAR of just 0.7% during the same time period. This reduced stock price run-up during the negotiation period appears to manifest in the total premium paid for the target as well. Although acquirers in the zero spillover group appear to pay an average total premium of 43.3% for the target, acquirers in the negative (positive) spillover groups pay an average total premium of 38.4% (33.6%), a discount of 4.9% (9.7%) relative to the zero spillover group. 8 Given that the average market capitalization of target firms is $493 million, the reduced premium paid by acquirer firms in negative (positive) spillover deals translates to a cost savings of approximately $20 million ($59 million), or 0.6% (2.3%) of the average acquirer market capitalization. Overall, the results in Table 6 suggest that acquirers in deals with strong information spillovers are able to successfully push down target valuations during merger negotiations, and benefit from this effort by completing the acquisition at a lower cost. 4.5 Additional tests In order to better attribute our primary findings to the acquirer s strategic disclosure incentives, we perform three additional tests. First, we investigate target firm disclosures during merger negotiations. Second, we perform a falsification test in which we examine press articles initiated by sources other than the firms themselves. Third, we repeat our original analysis of acquirer firm disclosures, but we do so for stock-for-stock mergers. 8 Note that although the 9.7% discount received by acquirers in the positive spillover group is significantly different from zero, the 4.9% discount received by acquirers in the negative spillover group falls short of statistical significance at conventional levels. 24

26 Before we proceed, we note that we perform these additional analyses using data from RavenPack because the disclosures from Factiva must be hand-collected. 9 Using RavenPack data, we construct a measure of disclosure tone to parallel our measure using Factiva data. In particular, we classify disclosures as being good news or bad news based on whether the primary sentiment variable, CSS, is above or below the average for all RavenPack disclosures within our sample. To ensure that the data in RavenPack is comparable to that in Factiva we repeat our analysis from Panel A of Table 3 using RavenPack data. We present the results of this analysis in Panel A of Table 8. Table 7 Panel A displays the results from repeating the analysis of acquirer disclosures during the negotiation period using data from RavenPack (i.e., the results shown in Table 3 Panel A). The sample is reduced from 464 deals to 325 deals because of the shorter coverage period in RavenPack PR Edition. Nevertheless, Panel A of Table 7 shows similar results with respect to acquirer disclosures as found using Factiva data. In particular, acquirers with positive (negative) information spillovers disclose more bad (good) news during merger negotiations Changes in target disclosures during merger negotiations In this section, we examine the target s (as opposed to the acquirer s) disclosures during the negotiation period. Although the transaction is the same for the acquirer and the target, the target s incentive in a cash deal is to maximize its own stock price, independent of the acquirer s stock price. In other words, the target s incentive is to disclose more favorable information regardless of the nature of the spillover between the two firms. 9 Limitations of using RavenPack data include that coverage in the PR Edition begins in 2004, reducing our sample period, and the primary sentiment variable, CSS, is proprietary in nature such that we do not know exactly how it is constructed. 25

27 We display the results of our analysis with respect to target disclosures in Panel B of Table 7. Consistent with our expectations, we find that targets disclose more positive news during merger negotiations, and this effect is independent of the nature of information spillovers between the acquirer and target firms. In addition to shedding light on targets disclosure behavior during merger negotiations, we view this test as helping to rule out alternative explanations for our primary findings regarding acquirers disclosures during negotiations Media-initiated articles The evidence presented suggests that acquiring firms internalize their disclosure externalities by strategically considering the content of their press releases during merger negotiations. In this section, we examine changes in the tone of news initiated by third parties (as opposed to news initiated by the acquiring firm) during negotiation periods. To the extent that firms have relatively less control over news they do not initiate, we do not expect to see a pattern of strategic behavior designed to influence the value of the target firm. On the other hand, if third party providers simply disseminate information provided by the firm then we would still find evidence of strategic behavior in news disseminated by third parties. Using non-press releases from the RavenPack Dow Jones dataset, we present the results of our analysis in Table 8. Consistent with acquirers having less control over news initiated by third parties, we find no significant change in media-initiated press article tone during merger negotiations, regardless of the nature of the information spillover between the acquirer and target firms. This result helps to further alleviate concerns that omitted factors are responsible for our primary findings that acquirers manipulate disclosures over which they have discretion to exploit information externalities during merger negotiations. 26

28 4.5.3 Stock-for-stock mergers As an additional falsification test, we repeat our original analysis of acquirer disclosures during merger negotiations, but instead of examining deals in which the consideration paid is 100% cash, we do so for deals in which the consideration is 100% stock. Our prediction is that when the acquirer is using its own stock as currency, the incentive to inflate the firm s own stock price will dominate the incentive to manipulate the target s stock price through information spillovers. Using press releases from RavenPack, we present the results of our analysis in Table 9. As expected, we find that acquirers using their own stock as consideration disclose unusually large amounts of positive news during merger negotiations designed to increase the value of their stock. This result appears to be consistent with Ahern and Sosyura (2014). Most important to the findings in our paper, the positive bias in acquirer s disclosures for stock for stock transaction is not a function of information spillover incentives. The implication is that spillover incentives do not materially influence acquirer s disclosures when their own stock price incentives are temporarily more important. 5. Conclusion Using data on firm-initiated news and merger negotiations, we study whether and to what extent managers recognize the potential for their firms disclosures to influence the asset prices of other firms, and strategically adjust their disclosures for gain. In doing so, we extend the disclosure literature by considering the effect of information spillovers in equilibrium disclosure choices. We show that acquirers originate substantially more negative news articles during merger negotiations when positive information spillovers exist between the two firms. In contrast, 27

29 acquirers originate more positive news articles during negotiations when negative information spillovers exist between the firms. The strategic disclosure behavior results in a short-lived walkdown in target firm value during the time when the takeover price is determined, which substantially reduces the cost of the investment for the acquirer. Our estimates indicate that this strategy reduces the cost of the takeover for acquirers with negative (positive) spillovers by approximately $20 million ($59 million), or 0.6% (2.3%) of the average acquirer s market equity. The results of this paper highlight an overlooked aspect of disclosure externalities. We show that disclosure externalities have feedback effects because firms strategically disclose information with the intention of influencing other firms when doing so is beneficial. In the context of mergers, this behavior has substantial impact on the relative wealth gains that bidders and targets realize, highlighting a new role of strategic disclosure on merger outcomes. The exploitation of disclosure externalities is likely to affect many corporate actions beyond mergers, such as product market competition, equity issuance, and executive compensation. 28

30 References Aboody, D., and Kasznik, R., CEO stock option awards and the timing of corporate voluntary disclosures, Journal of Accounting and Economics, 29(1), Ahern, K., Sosyura, D., Who writes the news? Corporate press releases during merger negotiations. Journal of Finance 69, Andrade, G., Mitchell, M., Stafford, E., New evidence and perspectives on mergers. Journal of Economic Perspectives 15, Armstrong, C., Jagolinzer, A., Larcker, D., Chief executive officer equity incentives and accounting irregularities. Journal of Accounting Research 48, Badertscher, B., Shroff, N., White, H., Externalities of public firm presence: Evidence from private firms investment decisions. Journal of Financial Economics 109, Betton, S., Eckbo, E., Thompson, R., Thorburn, K., Merger negotiations with stock market feedback. Journal of Finance 69, Bushee, B., Leuz, C., Economic consequences of SEC disclosure regulation: Evidence from the OTC bulletin board. Journal of Accounting and Economics 39, Cai, J., Song, M., Walkling, R., Anticipation, acquisitions, and bidder returns: Industry shocks and the transfer of information across rivals. Review of Financial Studies 27, Chen, C., Young, D., Zhuang, Z., Externalities of mandatory IFRS adoption: Evidence from cross-border spillover effects of financial information on investment efficiency. The Accounting Review 88, Core, J., Guay, W., The use of equity grants to manage optimal equity incentive levels. Journal of Accounting and Economics 28, Core, J., Guay, W., Estimating the value of employee stock option portfolios and their sensitivities to price and volatility. Journal of Accounting Research 40, Dyck, A., Zingales, L., The bubble and the media. Corporate Governance and Capital Flows in a Global Economy (Oxford University Press, Oxford), Eckbo, E., Corporate takeovers and economic efficiency. Annual Review of Financial Economics 6, Erickson, M., Hanlon, M., Maydew, E., Is there a link between executive equity incentives and accounting fraud? Journal of Accounting Research 44, Firth, M., The impact of earnings announcements on the share price behavior of similar type firms. Economic Journal 86, Foster, G., Intra-industry information transfers associated with earnings releases. Journal of Accounting and Economics 3, Freeman, R., Tse, S., An earnings prediction approach to examining intercompany information transfers. Journal of Accounting and Economics 15, Fuller, K., Netter, J., Stegemoller, M., What do returns to acquiring firms tell us? Evidence from firms that make many acquisitions. Journal of Finance 57, Jayaraman, S., Milbourn, T., CEO equity incentives and financial misreporting: The role of auditor expertise. The Accounting Review 90,

31 Loughran, T., and McDonald, B., 2011, When is a liability not a liability? Textual analysis, dictionaries, and 10-Ks. Journal of Finance, 66: Moeller, S., Schlingemann, F., Stulz, R., How do diversity of opinion and information asymmetry affect acquirer returns? Review of Financial Studies 20, Savor, P., Lu, Q., Do stock mergers create value for acquirers? Journal of Finance 64, Seru, A., Firm boundaries matter: Evidence from conglomerates and R&D activity. Journal of Financial Economics 111(2): Shroff, N., Verdi, R., Yost, B., When does the peer information environment matter? Journal of Accounting and Economics, forthcoming. Shroff, N., Verdi, R., Yu, G., Information environment and the investment decisions of multinational corporations. The Accounting Review 89, Skinner, D., Why firms voluntarily disclose bad news. Journal of Accounting Research, 32(1), Skinner, D., Earnings disclosures and stockholder lawsuits, Journal of Accounting and Economics, 23(3), Wang, W., Bid anticipation, information revelation, and merger gains. Working Paper. 30

32 Variable Appendix Detailed definitions of all variables used in our empirical analyses This table provides a detailed description of the procedures used to compute each variable used in our analyses. Our data are obtained either through Compustat, CRSP, SDC Platinum, RavenPack, Factiva, and firms financial filings available on EDGAR. All continuous variables are winsorized at 1% and 99% of the distribution. The variables are listed according to alphabetical order. Variable Acquirer Premium Acquirer Run-up Bad News Day Ratio Equity Incentives Good News Day Ratio Good - Bad News Day Ratio Negative Spillover Negotiation Period Positive Spillover Post-Announcement Definition The acquiring firm s cumulative daily abnormal return (CAR) starting with the negotiation period through the public acquisition announcement date (t=-60 to t=+2 days relative to the acquisition public announcement date). Abnormal returns are computed as the daily returns adjusted for the value-weighted index. The acquiring firm s cumulative daily abnormal return (CAR) starting with the negotiation period until 16 days prior to the public acquisition announcement date (t=-60 to t=-16 relative to the public announcement date). Abnormal returns are computed as the daily returns adjusted for the value-weighted index. The proportion of days in the period (e.g., prenegotiation, negotiation, or postannouncement) for which the positive/negative words gap is below the sample average. Words are categorized as positive or negative using the word list in the 2014 Master Dictionary available on Bill McDonald s website: (Loughran and McDonald, 2011). The portfolio delta of the CEO of the acquiring firm, measured as the dollar change in the CEO s equity portfolio value for a 1 percent change in the firm s stock price. We estimate delta based on the methodology in Core and Guay (1999, 2002). The proportion of days in the period (e.g., prenegotiation, negotiation, or postannouncement) for which the positive/negative words gap is above the sample average. Words are categorized as positive or negative using the word list in the 2014 Master Dictionary available on Bill McDonald s website: (Loughran and McDonald, 2011). The proportion of days in the period for which the positive/negative words gap is above the sample average, less the proportion of days in the period for which the positive/negative words gap is below the sample average. Words are categorized as positive or negative using the word list in the 2014 Master Dictionary available on Bill McDonald s website: (Loughran and McDonald, 2011). An indicator variable equal to one if the Spearman correlation of the acquirer and target firm stock returns is more than one standard deviation below zero, and equal to zero otherwise. The correlation is measured using daily returns adjusted for the value-weighted index from days t=-300 to t=-181 relative to the acquisition public announcement date. The window of time during which deal negotiations between the acquiring and target firms take place. We use a 45-day window stretching from t=-60 to t=-16 relative to the acquisition public announcement date. An indicator variable equal to one if the Spearman correlation of the acquirer and target firm stock returns is more than one standard deviation greater than zero, and equal to zero otherwise. The correlation is measured using daily returns adjusted for the value-weighted index from days t=-300 to t=-181 relative to the acquisition public announcement date. The window of time following the acquisition public announcement date. We use a window stretching from t=+2 relative to the acquisition public announcement date to 40 days after the acquisition effective date. 31

33 Prenegotiation Period Spillover Target Premium Target Run-up The window of time leading up to deal negotiations between the acquiring and target firms. We use a 120-day window stretching from t=-180 to t=-61 relative to the acquisition public announcement date. An indicator variable equal to one if the Spearman correlation of the acquirer and target firm stock returns is more than one standard deviation greater than zero, or else one standard deviation below zero. Otherwise the variable is equal to zero. The correlation is measured using daily returns adjusted for the value-weighted index from days t=-300 to t=-181 relative to the acquisition public announcement date. The target firm s cumulative daily abnormal return (CAR) starting with the negotiation period through the public acquisition announcement date (t=-60 to t=+2 days relative to the acquisition public announcement date). Abnormal returns are computed as the daily returns adjusted for the value-weighted index. The target firm s cumulative daily abnormal return (CAR) starting with the negotiation period until 16 days prior to the public acquisition announcement date (t=-60 to t=-16 relative to the public announcement date). Abnormal returns are computed as the daily returns adjusted for the value-weighted index. 32

34 Figure 1 Timeline of a typical merger The figure below shows the length of time in each stage of the merger process as used in our tests. Numbers represent trading days relative to the public announcement of the merger. 33

35 Figure 2 Tone of acquirer press release articles during merger negotiations In the figure below, the x-axis represents time (number of days relative to the acquisition s public announcement date) and the y-axis represents the cumulative ratio of good minus bad news days based on press articles issued by the acquiring firm. In this figure, each group s Good - Bad News Day Ratio is adjusted for its average value during the prenegotiation period. The sample consists of 464 cash-only acquisitions. The average negotiation period length is 45 days (beginning at t=-60 and ending at t=-16 days relative to the public announcement date). 34

36 Figure 3 Tone of acquirer press release articles following merger announcements In the figure below, the x-axis represents time (number of days after the acquisition s public announcement date) and the y-axis represents the cumulative ratio of good minus bad news days based on press articles issued by the acquiring firm. In this figure, each group s Good - Bad News Day Ratio is adjusted for its average value during the negotiation period. The sample consists of 464 cash-only acquisitions. 35

37 Figure 4 Target firm cumulative abnormal returns during mergers In the figure below, the x-axis represents time (number of days until the acquisition s public announcement date) and the y-axis represents the cumulative abnormal returns of the target firm, adjusted for the value-weighted index. The sample consists of 464 cash-only acquisitions. The average negotiation period length is 45 days (beginning at t=-60 and ending at t=-16 days relative to the public announcement date). 36

38 Table 1 Sample selection Description All-Cash Deals Sample period Deals > $10M where acquirer and target are U.S. public firms 464 Total number of Factiva press articles 70,577 Length of prenegotiation period for each deal (in days) 120 Average length of negotiation period (in days) 45 Total number of good news days 14,217 Total number of bad news days 6,294 Total number of neutral (or no) news days 53,231 Total number of days in sample (prior to deal announcement) 73,742 37

39 Table 2 Descriptive statistics This table presents descriptive statistics for firms and deals in each of our three groups of interest: zero spillover deals (the benchmark group), positive spillover deals, and negative spillover deals. The groups are formed by partitioning the sample of 464 all-cash deals on the basis of the daily returns correlation between the acquiring and target firm. Daily Returns Correlation is the Spearman correlation of daily stock returns for the acquiring and target firm. Acquirer Market Cap is the market capitalization of the acquiring firm, in millions. Target Market Cap is the market capitalization of the target firm, in millions. Acquirer/Target Same Industry represents the proportion of deals in which the acquiring and target firm are members of the same SIC 2-digit industry. Good - Bad News Day Ratio is the proportion of days in the period for which the positive/negative words gap is above the sample average, minus the proportion of days in the period for which the positive/negative words gap is below the sample average. Good News Day Ratio is the proportion of days in the period for which the positive/negative words gap is above the sample average. Bad News Day Ratio is the proportion of days in the period for which the positive/negative words gap is below the sample average. Zero Spillover Positive Spillover Negative Spillover Col (1) - (2) Col (1) - (3) Mean Mean Mean t-stat t-stat Number of Deals Daily Returns Correlation Acquirer Market Cap ($ millions) 2,870 2,500 3, Target Market Cap ($ millions) Acquirer/Target Same Industry Factiva Press Articles: Good - Bad News Day Ratio Good News Day Ratio Bad News Day Ratio

40 Table 3 Changes in acquirer press release tone during negotiation Panel A in this table presents the results from examining the change in the tone of Factiva press articles during deal negotiations, conditional on the covariance of the acquiring and target firm returns. Panel B also presents the results from examining changes in acquirer disclosure tone, but examines good news and bad news separately. All variables are defined in the Variable Appendix. All specifications include deal fixed effects. The standard errors are clustered at the deal level and are robust to heteroskedasticity. *,**,*** indicate statistical significance at the 10%, 5%, and 1% levels, respectively, using a two-tailed t-test. Panel A: Acquirer Good - Bad News Day Ratio Dependent Variable Pr. Sign Good - Bad News Day Ratio Negotiation Period *** * (1.24) (2.70) (0.66) (1.97) Negotiation Period Positive Spillover *** *** (-3.26) (-2.85) Negotiation Period Negative Spillover *** 0.037*** (3.23) (2.93) F-Test for joint significance (β 1 + β 2 ) F-Test for joint significance (β 1 + β 3 ) < 0.01 < 0.01 F-Test for joint significance (β 2 + β 3 ) 0.06 Deal fixed effects Yes Yes Yes Yes S.E. clustered by deal Yes Yes Yes Yes No. of Observations No. of Acquisitions Adj. R-Squared 79.2% 79.4% 79.5% 79.6% Panel B: Acquirer Good News Day Ratio and Bad News Day Ratio 39