Media-Based Merger Arbitrage Matthias M. M. Buehlmaier and Josef Zechner∗ First Draft: October 2012 This Draft: September 28, 2013 Executive Summary Merger arbitrage is one of the most profitable investment strategies available to investment practitioners. Although it is mostly used by institutional investors such as hedge funds, it is increasingly being made accessible to retail investors as well through ETFs and mutual funds. While merger arbitrage is one of the mostprofitabletradingstrategies, itissurprisingthatthedeterminantsofmerger arbitrage success have received little attention in previous studies. This paper fills this gap by investigating a key determinant of merger arbitrage success: the financial media, for example financial newspapers and financial newswires. The financial media is important because it reveals new information that can increase theprofitabilityofthemergerarbitrageinvestmentstrategy. Consistentwiththis argument, we find that risk-adjusted alphas increase by more than 12 percentage points when trading on the information content of the financial press. Keywords: Financial media, merger arbitrage, hedge funds, market efficiency, mergers and acquisitions. JEL Codes: G34, G14, G11 ∗Matthias Buehlmaier (corresponding author) is Assistant Professor of Finance at the Faculty of BusinessandEconomics,TheUniversityofHongKong,[email protected],+85222194177. JosefZechner is Professor of Finance at WU Vienna University of Economics and Business, [email protected], +43 1 31336-6301. The authors thank members of the finance research seminars of City University of Hong Kong and Lingnan University for valuable suggestions. The work described in this paper was substantially supported by a grant from the Research Grants Council of the Hong Kong Special Ad- ministrative Region, China. Furthermore, this research has been partially supported by the following grant: HKU Seed Funding for Basic Research. 1 Abstract Using a large sample of merger announcements, this paper provides strong evidence that information in financial media is not fully incorporated in stock prices. Cross-sectional regressions show that a one standard deviation increase in the media-implied probability of deal completion results in an increase of 1.2% in the subsequent twelve-day return. Media content information released on the an- nouncement day contains information, not captured by announcement day stock returns, which are found to be largely unrelated to the probability of deal com- pletion. The results for media coverage are much weaker. A trading strategy based on media content increases annualized alphas by 12.5%, while the effect of media coverage on alphas is statistically insignificant. Finally, we find weak evidence in favor of a certification role of the media, with the top newswire and top newspapers contributing more information to the market. Keywords: Financial media, merger arbitrage, hedge funds, market efficiency, mergers and acquisitions. JEL Codes: G34, G14, G11 2 1 Introduction Akeypurposeofthefinancialmediaistoprovidenovelinformationtofinancialmarkets (Engelberg and Parsons (2011)). This information is likely to be particularly relevant during times when a corporate event takes place, since this introduces extra uncertainty andinformationalasymmetriesarelikelytobehigh. Duringsuchtimes, theinformation provided by the financial media may therefore be particularly valuable. A contrasting view is that the information contained in financial media is already re- flected in financial markets, or that it may be manipulated, and thus may be irrelevant or even mislead financial market participants (Gurun and Butler (2012), Ahern and Sosyura (2013)). Although there is a growing literature dealing with related questions, it is still unclear whether media provide information not yet incorporated in securities prices and, if this is the case, what the implications are for stock returns, corporate valuations, and shareholder wealth (Tetlock (2007), Tetlock et al. (2008), Barber and Odean (2008)). Furthermore, if information flows from the media to financial mar- kets, we would like to understand the specific channels through which such information transmission takes place. We investigate these questions by considering public announcements of a planned corporate merger or acquisition. During this event, one firm, the acquirer, tries to take over another firm, the target. The distinguishing feature of this event is the uncertainty about whether the planned acquisition will complete or fail. This uncertainty is used by so-called merger arbitrageurs, who place risky stock market bets that the acquisition will complete. By their nature, these bets are highly sensitive to new information about the likelihood of deal completion. We use this sensitivity as an identification strategy to test whether the financial media provides fundamental information about the merger or whether it is a potentially manipulated sideshow. To quantify media-based information, we use a novel empirical approach that al- lows us to directly calculate the media-implied likelihood of deal completion. This methodology allows us to extract the media-based information that is directly relevant to the merger arbitrage investment strategy. It thus provides further insights about the specific channels of information transmission to financial markets. We find that information in financial media is not fully incorporated in stock prices, even after twelve trading days. Our cross-sectional regressions show that a one stan- dard deviation increase in the media-implied probability of deal completion results in an increase of 1.2% in the subsequent twelve-day return, corresponding to a monthly increase of 2.2%. Furthermore, time series tests show that the twelve trading days after an announcement date yield annualized risk-adjusted alphas of 18.5% for deals with a high media-implied probability of deal completion, while the alpha drops to 6% for deals with a low media-implied probability of deal completion. While we find strong evidence that media content affects merger arbitrage returns, the result for media coverage, i.e. the frequency, with which the media report an 3 announced merger, is much weaker. For example, our time series regressions show that a one standard deviation increase in media coverage yields an increased return of only 0.3% whereas the same change in lagged media content yields a monthly in- crease in returns of 0.8%. Also, a trading strategy based on media content increases annualized alphas by 12.5%, while the effect of media coverage on alphas is statistically insignificant. So, while media manipulation may occur by increasing media coverage (Ahern and Sosyura (2013)), this does not seem to be the case for media content. Ourresultsshowthatmediainformationreleasedontheannouncementdaycontains information, not captured by announcement day stock returns. In fact, a regression of announcement returns on media variables show insignificant results. Thus, announce- mentreturnsseemtobeunrelatedtotheprobabilityofdealcompletionandmaylargely reflect other information such as an assessment of whether a completed deal would cre- ate or destroy value. Finally, we find weak evidence in favor of a certification role of the media, with the top newswire and top newspapers contributing more novel information to the market. The remainder of the paper is organized as follows. Section 2 reviews the related literature. We then introduce the merger arbitrage investment strategy in Section 3 and explain how we quantify media information in Section 4. In the remaining sections we describe an analyze the data: Section 5 details our data and shows summary statistics, while Section 6 presents our regression results. Finally, Section 7 concludes this paper. 2 Related Literature This research project is related to four strands of the literature. The first strand is on mergers andacquisitions (M&A). Manne(1965) and Jensen(1986) are two cornerstones of the current literature because they show how mergers can create value. Manne’s basic proposition is that the control of a firm constitutes a valuable asset and that through the M&A process an active market for corporate control exists. Jensen (1986) recognizes that control of a corporation is frequently in the hands of management instead of being in the hands of the corporation’s shareholders. This leads to agency costs when a company has large cash flows but few high-return investment projects. Jensen proposes debt as a potential solution to this agency problem and shows that takeovers play a crucial role in this debt creation. Building on this basic literature, Grossman and Hart (1980) investigate the so-called free-rider problem that occurs in tender offers during takeovers. In their model, no target shareholder tenders his shares because every target shareholder tries to free-ride on other shareholders. Bagnoli and Lipman (1988) show how to overcome this free-rider problem by making a subset of target shareholders pivotal. The basic idea is that some shareholders get to know that their tendering decision is essential for the takeover deal to complete. Finally, a large empirical literature deals with the question of whether mergers create or destroy value. 4 For example, Savor and Lu (2009) show that overvalued acquirer firms create value for their own shareholders by using acquirer stock instead of cash to pay for acquisitions. The second strand of the literature is on merger arbitrage. The key difference to the M&A literature discussed above is that the arbitrage literature deals with stock trading strategies of deal outsiders, who try to profit from the events surrounding mergers. Lar- cker and Lys (1987) show that merger arbitrageurs obtain private information about the outcome of a takeover deal and use this information to earn substantial returns. Many other studies confirm that merger arbitrage strategies earn high excess returns, sometimes more than 100 percent annually (Dukes et al. (1992), Karolyi and Shan- non (1999), and Jindra and Walkling (2004)). Potential explanations include market inefficiencies, limits to arbitrage (Baker and Savaşoglu (2002)), and trading costs and premia for providing liquidity during market downturns (Mitchell and Pulvino (2001) and Mitchell et al. (2007)). Cornelli and Li (2002) argue that merger arbitrageurs have an informational advantage relative to target shareholders because arbitrageurs, hiding among noise traders, know that they bought shares. The third strand of the literature is on the role of the media in finance. A first step is to consider how stock market indices incorporate new information from the financial press, as investigated by Tetlock (2007). He shows that, consistent with models of noise traders, the content of a Wall Street Journal column predicts downward pressure on the stock market, followed by a price reversal. Another hypothesis is that the media can alleviate informational frictions in the stock market, even if the media does not supply genuine news. Fang and Peress (2009) investigate this hypothesis and find that stocks with no media coverage earn higher returns than stocks with high media coverage. However, when considering the media and trading in the stock market, it is crucial to distinguish between the impact of media reporting and the impact of the events being reported. Engelberg and Parsons (2011) address this question and find that local media coveragecauseslocaltradingactivity. AhernandSosyura(2013)andBuehlmaier(2012) show that media content can be manipulated during corporate acquisitions. Dyck et al. (2008) consider corporate governance and find that media coverage leads to reversals of corporate governance violations. There is a fourth strand of the literature that uses text-based information in ar- eas unrelated to the financial media. For example, Hoberg and Phillips (2009) use text-based information from 10-K statements to develop a novel method of industry classification and apply it to product differentiation as well as mergers and acquisitions (Hoberg and Phillips (2011)). Das and Chen (2007) and Antweiler and Frank (2004) analyze the text of stock message boards on the internet. 5 3 Introducing Merger Arbitrage When a company (the acquirer) tries to “swallow” another company (the target), a so-called merger, acquisition, or takeover occurs. For the purpose of this paper, we use the terms merger, acquisition, and takeover interchangeably, although depending on their definition there can be subtle differences between them. Merger arbitrage is an investment strategy that bets on the outcome of a merger, i.e. whether or not the acquirer ends up “swallowing” the target. An alternative name for merger arbitrage is risk arbitrage. This section briefly reviews this investment strategy and clarifies the terminology surrounding it. 3.1 Outlining the Anatomy of a Typical Corporate Acquisition Although each corporate acquisition is different, a typical timeline can be described as follows. In the pre-announcement phase, the management and the boards of both the target and the acquirer are in private negotiations about the potential sale of the target to the acquirer and the conditions of this sale. This phase is often shrouded in secrecy, and even the employees of the target and the acquirer (with exception of the senior management) are not aware of the ongoing negotiations. The negotiations are usually facilitated by investment banks, who also have to treat any knowledge of the ongoing negotiations confidentially. The negotiations may break down at any point in time, or they may reach the next stage, at which point the deal is made public. The date when the deal is made public is called the announcement date. Its purpose is to inform regulators and the shareholders of the target and the acquirer that a deal is in the making. The goal of the subsequent post-announcement phase is to obtain approval from both regulators and shareholders to complete the planned merger. If it is possible to obtain approval, the deal is completed and the firms merge. Otherwise the deal status is said to be withdrawn and the two companies continue to exist as two separate entities. The resolution date is the date when the deal either completes or is withdrawn. Although most deals fit into the timeline above, there are some deals where the pre- announcement phase is skipped and no negotiations take place prior to the announce- ment date. These deals are called unsolicited since there are no prior negotiations and the acquirer directly approaches the target’s shareholders instead of the target’s management or board. Most unsolicited deals end up being hostile, meaning that the target’s management and board oppose the takeover. There are, however, also unso- licited deals that turn friendly, with the target’s board and management supporting the planned acquisition. In the simplest case, the method of payment is cash, meaning that the acquirer pays a given amount of money to target shareholders in exchange for each target share. A deal of this type is known as a cash deal or all-cash deal. Instead of using cash, 6 the acquirer may wish to issue new acquirer stock and use this newly-issued acquirer stock to pay for the acquisition. In this case, the target shareholders give up ownership of their target stocks (which are subsequently canceled) and in return receive newly- issued stock of the acquirer. A deal of this type is called all-stock deal. This type of deal, however, is not very frequent. More often one settles on a deal where target shareholders receive both cash and acquirer stock. A deal of this type is called stock deal, where it is important to note that many stock deals also have a cash component. In this sense, a stock deal means that some of the payment to target shareholders (but not necessarily the whole payment) consists of acquirer stock, with the remaining part usually consisting of cash. 3.2 Profiting from Merger Arbitrage The distinction between cash deals and stock deals is important for merger arbitrageurs since it determines which investment strategy they use. The simplest case for the arbitrageur is the cash deal, in which he simply buys the target stock on the stock market immediately after the public announcement of the deal. If the deal completes, the arbitrageur makes a profit that consists of the difference between the (cash) bid price and the price of the target stock when he bought it. If the deal is withdrawn, the arbitrageursellshistargetstock,usuallyataloss. Giventhisinvestmentstrategy,which isonlybasedonpublicinformation, thearbitrageurmakesaprofitifthedealcompletes, and he may make a loss if the deal is withdrawn. The key takeaway is that conditional on the success of the takeover bid (i.e. when the deal closes), the arbitrageur makes a riskless profit. As in the textbook definition of arbitrage, this profit is already known in advance. (Sometimes the bid price is even revised upwards after the announcement date, which further increases the arbitrageur’s profit.) Of course, it is unknown on the announcement date whether the deal will complete, which introduces a risk element to the arbitrage strategy. This is the reason why this investment strategy is commonly calledrisk arbitrage, althoughthisterminologyissomewhatofamisnomer. Throughout this paper we use the term merger arbitrage since this term, while also not being strictly correct, tends to create less confusion than the term risk arbitrage. In contrast to a cash deal, a stock deal makes the life of a merger arbitrageur slightly more complicated. In addition to buying the target stock on the stock market, he also needs to short-sell the acquirer’s stock in order to lock in a risk-free profit conditional on deal completion. Consider for example an all-stock deal. In this deal, the negotiations between the target and the acquirer result in a so-called exchange ratio being specified. This exchange ratio, denoted by δ, defines how many acquirer stocks a target stockholder may obtain for each target stock he holds. For example, if the exchange ratio is δ = 2, then each target stock will be converted into two newly-issued acquirerstocksincasethedealcompletes. Forthearbitrageurtolockinarisk-freeprofit conditionalondealcompletion, hethusshort-sellsδ acquirerstocksforeachtargetstock 7 he buys. If the deal completes, he exchanges each target stock into δ acquirer stocks and uses these acquirer stocks to cover his short position. Since δ reflects a premium over the target’s stock price at the announcement date, the arbitrageur makes a riskless profit if the deal closes. Suppose for example that shortly after the announcement date, the target’s stock price is $90 and the acquirer’s stock price is $50. The arbitrageur buys one target stock and short sells δ = 2 acquirer stocks, leaving him with a positive cash flow of $10 = 2·$50−$90. When the deal closes, his cash flow is zero, since he can cover his short position in the acquirer stock by exchanging the target stock. He thus makes a riskless profit of $10 per long-short position conditional on deal completion. In contrast, if the deal is withdrawn, the arbitrageur often loses money since he has to unwind his long position in the target and his short position in the acquirer, often at a loss. As mentioned earlier in this section, there are only relatively few all-stock deals. Many deals that have a stock component also have a cash component. The merger arbitrageur’s investment strategy in these so-called stock deals (which may also contain a cash component) is treated in the existing empirical literature in the same way as in all-stock deals (Mitchell and Pulvino (2001)). That is, although less than 100% of the payment is made in stock, the literature considers an investment strategy that shorts δ acquirer stocks for each target stock bought. The reason for this simplification is mainly data availability, since Thomson Reuters SDC Platinum, the most commonly used database on M&A, only has a dummy variable that indicates whether there is some stock component; it does not contain a variable that encodes how large the stock component is. In this paper, we thus follow the existing literature in constructing merger arbitrage returns for stock deals, with one minor extension. Often the exchange ratio is not available, even when the deal is marked as a stock deal. To approximate the exchange ratio in these cases, we use the average of the collar target ratios, in case they are available (Adolph and Pettit (2007)). Building upon these merger arbitrage investment strategies, we calculate so-called long-short merger arbitrage returns as follows by distinguishing between stock deals and cash deals. For stock deals with known exchange ratio δ, the long-short return is given by the target’s return minus δ times the acquirer’s return. In stock deals, if the acquirer’s return is missing or if the exchange ratio is missing, the long-short return is also assigned a missing value. For cash deals, the long-short return is given by the target’s return, and is thus a long-“short” return only in a degenerate sense. Following this recipe, we can calculate long-short merger arbitrage returns for all deals, where it is important to keep in mind that for the cash deals this “long-short” investment strategy is degenerate. With the exchange ratio δ being set to zero for cash deals, we can write the long-short strategy returns shorthand as: r −δr , Tar Acq where r , r denote the target stock return and the acquirer stock return, respec- Tar Acq 8 tively. In addition to this long-short merger arbitrage strategy, we also consider a simplified investment strategy that always invests in the target stock, independent of whether the acquisition is a cash deal or a stock deal. To ensure that there is no look-ahead bias in our sample, we follow the common practice in the literature to open merger arbitrage trading positions on the first trading day after the deal’s announcement day. This is motivated by two considerations. First, somedealsareannouncedafterthemarketsclose, whichmakesitimpossibletotradeon this information on the same day. Second, even if a deal is announced during trading hours, our databases dot not contain the exact announcement time of the day, and furthermore it is difficult to gauge how long it would take hedge funds to open their trading positions after the announcement. To be conservative and to follow the prior literature, we thus open trading positions only after the announcement day, calculating this day’s return based on the opening stock price and closing stock price of that day. The left-hand side of Figure 1 shows these merger arbitrage returns for the target, the acquirer, and the long-short merger arbitrage strategy. The solid lines represent completed deals, while the dashed lines represent withdrawn deals. The right-hand side of Figure 1 shows the same returns, with the difference that these returns include the announcement day. The right-hand side returns thus do not correspond to a tradable investment strategy based only on public information. Considering the left-hand column in Figure 1, we see that profiting from merger arbitrage is difficult if the announcement day is omitted. For completed deals, the target’s cumulative stock return (shown in the top row) is close to zero without much variation. The target stock return for withdrawn deals becomes increasingly negative as time progresses. For the acquirer (shown in the middle row) a similar picture emerges, withthedifferencethatevencompleteddealshaveslightlynegativereturns. Incontrast, the long-short merger arbitrage strategy (shown in the bottom row) makes money if the deal completes, and loses money if the deal is withdrawn. Figure 1 vividly illustrates that it makes a fundamental difference for the merger arbitrageur whether he invests in a deal that will be completed or in a deal that will be withdrawn. While deal completion/withdrawal becomes known with certainty only at the very end of a deal, Figure 1 shows that prices incorporate this information gradually. For merger arbitrageurs, it is thus essential to stay ahead of the curve by trying to obtain new information about the likelihood of deal completion. Since the financial media processes and disseminates merger-related information, we describe in the next section how this information can be quantified to be useful for the merger arbitrageur. 9 Figure 1: Arbitrage Returns for Completed Deals and Withdrawn Deals This figure shows the cumulative event-time returns of the target’s stock (row one), the acquirer’s stock (row two), and the long-short strategy (row three). The long-short strategy consists of shorting the acquirer stock and going long the target stock in stock deals where an exchange ratio exists, and going long the target in cash deals. The returns in the left column of figures show the merger arbitrage returns. These returns start on the first trading day after the announcement day. For comparison, the returns in the right column of figures show hypothetical merger arbitrage returns that could be obtained if, hypothetically, the merger arbitrageur could capture the announcement day returns as well. 0.10 0.10 0.05 CWoitmhdprlaewtend ddeeaallss 0.05 Return 0.00 Return 0.00 Target's −0.05 Target's −0.05 Completed deals −0.10 −0.10 Withdrawn deals −0.15 −0.15 0 5 10 15 20 25 30 0 5 10 15 20 25 30 Trading Days after Annoucement Date Trading Days after Annoucement Date 0.10 0.10 0.05 CWoitmhdprlaewtend ddeeaallss 0.05 CWoitmhdprlaewtend ddeeaallss n n Retur 0.00 Retur 0.00 Acquirer's −0.05 Acquirer's −0.05 −0.10 −0.10 −0.15 −0.15 0 5 10 15 20 25 30 0 5 10 15 20 25 30 Trading Days after Annoucement Date Trading Days after Annoucement Date 0.10 0.10 Acquirer 0.05 Acquirer 0.05 Return: Long Target, Short −0.10−0.050.00 CWoitmhdprlaewtend ddeeaallss Return: Long Target, Short −0.10−0.050.00 CWoitmhdprlaewtend ddeeaallss −0.15 −0.15 0 5 10 15 20 25 30 0 5 10 15 20 25 30 Trading Days after Annoucement Date 10 Trading Days after Annoucement Date