Conflicts of Interest within Investment Banks: Analysts and Proprietary Traders David Haushalter Penn State University E-mail: gdh12@psu.edu Phone: (814) 865-7969 Michelle Lowry Penn State University E-mail: mlowry@psu.edu Phone: (814) 865-1483 March 21, 2008 Abstract: We examine the interaction between an investment bank’s trading, analyst recommendations, and advising activities around mergers. Banks advising an acquirer provide higher analyst recommendations to acquirers than non-advising banks around the time of the merger. On average the trading of an acquirer by the advisor bank following the announcement of an acquisition is in line with the recommendations made by the advisor’s analyst. However, additional tests show that this relation only holds for banks that do not rely heavily on investment banking as a source of revenue. The results are consistent with arguments that the conflicts faced by analysts differ predictably across firms. Traders within an investment bank are cognizant of the incentives of the in-house analysts, and knowledge of such incentives significantly affects the extent to which they rely on their recommendations when making trades. We thank Lubomir Petrasek for excellent research assistance. 1. Introduction The potential for conflicts of interest is prevalent within investment banks. Nearly all large investment banks offer underwriting services, provide analyst coverage, and have proprietary trading desks, and a wide body of evidence suggests that these divisions do not act independently. In fact, Leaven and Levine’s (2007) finding of a diversification discount for financial conglomerates and DeLong’s (2001) event study analysis of investment bank mergers suggests a valuation penalty for offering too many services under one roof. An understanding of these conflicts is important to institutions, to companies buying services from the banks, and also to the banks themselves. To study the conflicts between investment banking activities and other aspects of a bank’s operations, we focus on mergers. Although conflicts of interest can be ongoing, they are arguably particularly large when a bank is advising a merger. First, analysts can be important in both enabling the bank to land a merger deal and in increasing the probability that the deal will be completed (see, e.g., Becher and Juergens, 2005). Mergers are a large source of revenues for investment banks. For example, in 2006 alone, the top 20 investment banks earned almost $35 billion in fees from underwriting mergers and acquisitions.1 This is about half of the total fees that they earned from all investment banking activities. Second, the insights of analysts into the expected costs and/or synergies of the merger can make their resulting recommendations particularly valuable. The value of companies can change dramatically around mergers, and the ability of analysts within the investment banks to forecast these changes can have dramatic 1 See “The Good Times Roll,” Bloomberg Magazine, April, 2007. 1 effects on their personal reputations.2 The first portion of the paper explores the ways in which analysts support the M&A advisory business. Specifically, we examine changes in analyst recommendations prior to, around the time of, and following merger announcements and completions. We find that analysts of the advisor firm issue significantly more optimistic recommendations around the merger, compared to non-affiliated analysts. This finding is consistent with the findings of Michaely and Womack (1999) and Lin and McNichols (1998), among, others, who find that affiliated analysts are more optimistic than non-affiliated analysts around equity offerings. Both of these studies conclude that the affiliated analysts are being overly optimistic to support the investment banking business. However, the possibility that affiliated analysts truly have a rosier outlook regarding the prospects of the acquirer company cannot entirely be ruled out. It would not be surprising if acquiring companies were more likely to choose an advisor that thought well of the merger in question, as opposed to one that had a more negative outlook. The second portion of the paper attempts to distinguish between the potential reasons that affiliated analysts are more optimistic than non-affiliated analysts, i.e., to determine if advisor firm analysts are being overly optimistic in an effort to support the investment banking business or if the advisor firm analysts truly have more positive expectations regarding the merger. To shed light on this issue, we contrast the actions of the advisor firm analysts with those of the advisor firm traders. It seems reasonable to assume that traders within a given institution would have more information regarding the incentives and pressures that analysts within their own bank face, compared to analysts at other banks. If the bank’s traders perceive that their firm’s analysts act independently from their investment banking arm, then we would expect that its traders 2 As shown by Moeller, Schlingemann, and Stulz (2005), returns to acquirers around the announcement of mergers at the 5% and 95% level range from -6% to 7% between 1980 to 1997 and -19% to 13% between 1998 and 2001. 2 would buy more shares of upgraded stocks. In contrast, if the bank’s traders perceive analysts to upgrade acquirer stocks in an effort to win or support the investment banking business, we expect that its traders would ignore the recommendations of these affiliated analysts. Our comparisons focus on the recommendations of and trading in the acquirer firm. We find no evidence that the institutional traders of the advisor firm trade in line with their analysts’ recommendations (on the acquirer firm) prior to the merger. However this association changes markedly after the merger: there is a significant positive relation between an advisor’s analyst recommendations and its trading following the merger. Mergers in which an advising firm’s analyst upgrades the acquirer’s stock are associated with significantly larger increases in share ownership of the acquirer by the advisor bank. The finding that traders’ buys and sells are more closely tied to analyst recommendations following the merger is consistent with at least two different scenarios. First, advisor firm analysts may provide more accurate recommendations following the merger, perhaps due to better information. The in-house traders would be more likely to act on the analyst upgrades or downgrades if they were perceived as higher quality, resulting in a stronger positive relation between recommendations and trades in the quarters following the merger. Alternatively, both the analysts and the traders of the advisor bank may face a greater conflict of interest following the merger. Suppose both analysts and traders are pressured to support the investment banking business, for example by issuing positive recommendations and by purchasing the stock. It is possible that pressure from the investment banking business causes analyst upgrades and inhouse purchases of the stock at the same time, thereby producing a positive relation between the two in the quarters following the merger. To distinguish between these alternative explanations, we attempt to classify investment 3 banks into those for which the conflict of interest from investment banking is likely to be more or less severe. Specifically, we classify investment banks into those that rely heavily on investment banking revenues versus those for which investment banking is less important. If there exists some expectation that the traders and analysts both support the investment banking arm, then this expectation should be greater in those firms for which investment banking is a relatively more important source of revenue. That is, we would expect the post-merger relation between analyst recommendations and in-house trading to be strongest in investment banks that rely most heavily on investment banking as a source of revenue. Alternatively, if traders perceive the analysts to have more accurate forecasts following the merger (due to more available information), then we would expect this increase in accuracy to be greatest among those banks where the conflict of interest from investment banking is particularly low. In this case, we would expect the post-merger relation between analyst recommendations and in-house trading to be strongest in those investment banks that rely least on investment banking as a source of revenue. We find that the positive relation between analyst recommendation changes and in-house trading is only significant among those investment banks that rely least on investment banking as a source of revenue, i.e., in those banks where analysts are likely to face the lowest conflicts of interest. Within these banks, traders’ actions indicate that they perceive analysts’ post-merger recommendations to be particularly valuable. In contrast, in banks where investment banking revenue is an important source of revenue, i.e., where the conflict of interest faced by analysts is likely to be quite high, the traders show no significant tendency to buy and sell in line with analyst recommendations. The in-house traders are evidently aware of the conflicts faced by analysts, and as a result they disregard their recommendations. 4 We also find that the relation between analyst recommendations and in-house trading varies according to the percentage of investment bank revenue that comes from trading activities. Among those banks that rely most heavily on trading as a source of revenue, there is a significantly stronger relation between analyst recommendation changes and in-house trading. These findings are consistent with an interpretation that analysts’ objective functions differ in predictable ways across institutions. Within investment banks that rely most heavily on trading, analysts are expected to provide the most accurate recommendations, and consequently traders within these institutions are much more likely to act on these recommendations. In contrast, within banks that rely most heavily on investment banking as a source of revenue, analysts are expected to support the investment banking business, and consequently the in-house traders pay little heed to their recommendations. Our findings contribute to several streams of literature. First, our study relates to the debate regarding analyst incentives and the extent to which conflicts of interest cause analysts to issue overly forecasts and recommendations. Michaely and Womack (1999), Dugar and Nathan (1995), and Lin and McNichols (1998), among others, find that analysts employed by underwriters of security offerings tend to be more optimistic than other analysts. However, Cowen, Groysber, and Healy (2006), Jacob, Rock and Weber (2003), Clarke, Khorana, Patel, and Rau (2004), and Agrawal and Chen (2005) find no evidence that conflicts of interest from investment banking make analysts more optimistic or less precise. We take a new approach to this problem, by considering the ways in which analyst incentives are likely to vary both over time and across investment banks. Second, although there are a number of papers on analyst actions around equity issues, there is relatively little evidence on analyst actions around mergers and acquisitions. M&A 5 activity is a substantial source of revenues for many investment banks, and our study increases our understanding on the ways in which investment banks potentially compete for this business. Our paper proceeds as follows. Section 2 reviews prior literature on conflicts of interest within investment banking. Section 3 outlines the data. Section 4 describes analyst recommendations and institutional ownership around the merger. Section 5 includes empirical tests on the relation between analyst recommendations and institutional trades, by the advisor investment bank. Section 6 investigates how the relation between analyst recommendations and institutional trades varies depending on the likely magnitude of the conflict of interest faced by analysts. Finally, Section 7 concludes. 2. Related Literature A substantial body of literature has examined the value of analyst recommendations. Givoly and Lakonishok (1979), Stickel (1991), Womack (1996), Barber, Lehavy, McNichols and Trueman (2001), Jegadeesh, Kim, Krische, and Lee (2004), Loh and Mian (2005), and Busse and Green (2002) all show that analysts’ earning forecasts and stock recommendations have investment value. Consistent with analyst recommendations being value relevant, research by Jackson (2005), Hong and Kubik (2003), and Mikhail, Walther, and Willis (1999) shows that analysts are motivated to increase their reputations by issuing the most informative forecasts and recommendations.3 In the merger framework, Becher and Juergens (2005) find that analysts have insight into the value of a merger, and as a result they can impact the outcome of a merger. Although considerable evidence suggests that analyst recommendations have value, there is also a large literature on the conflicts of interest that analysts face. Specifically, as discussed 3 Ljungqvist, Malloy and Marston (2006) find that the importance of accuracy for career outcomes has become more limited in recent years. 6 in detail by Mehran and Stulz (2007), analysts face pressure from within their firm to issue overly optimistic forecasts and recommendations to support the investment banking business. While there is broad consensus that analysts face conflicts of interest, the effects of such conflicts are disputed. Lin and McNichols (1998), and James and Karceski (2006), among others, find that analysts employed by underwriters of security offerings tend to be more optimistic than other analysts. Findings of Michaely and Womack (1999), Aggarwal, Purnanandam, and Wu (2005), Barber, Lehavy, and Trueman (2007) suggest that this optimism contributes to inflated stock prices. The findings of Dugar and Nathan (1995) and McNichols, O’Brien, and Pamukcu (2006) similarly show that affiliated analysts are more optimistic. However, they find that the market discounts the affiliated analysts’ recommendations. Agrawal and Chen (2005) and Bradley, Jordan and Ritter (2006) reach similar conclusions. Finally, papers by Cowen, Groysber, and Healy (2006) and Jacob, Rock and Weber (2003) find no evidence that a conflict of interest from investment banking causes analysts to issue overly optimistic or less precise forecasts. Similarly, Agrawal and Chen (2004) find no evidence that accuracy or bias in earnings forecasts are related to the importance of investment banking as a source of revenue to the financial institution.4 Part of the inconsistency in these streams of prior literature is potentially related to the fact that not all analysts face the same conflicts at all times with respect to all stocks. Ljungqvist, Marston, Starks, Wei, and Yan attempt to address this issue by separating stocks by the level of Note that Agrawal and Chen’s (2004) examination of earnings forecasts shows no evidence of a conflict of interest from investment banking resulting in overly optimistic forecasts. In contrast, Agrawal and Chen’s (2005) study of analyst recommendations yields the opposite conclusion. This difference is consistent with Mehran and Stulz’s observation that more evidence exists suggesting that recommendations are biased and less evidence suggesting that earnings forecasts are biased. This potentially reflects the fact that earnings forecasts are more verifiable and potentially have greater effects on an analyst’s reputation. 4 7 institutional ownership. Analysts’ career paths are largely influenced by the All-Star rankings, which are based on institutional investor feedback. Consequently, it follows that an analyst’s incentives to provide unbiased, accurate recommendations are highest in those stocks with the highest institutional ownership. Consistent with this conjecture, the authors find that recommendations relative to consensus are positively related to investment banking relationships and negatively related to ownership by institutional investors. Although all of the above studies examine analyst conflicts and are therefore obviously related to our research question, there are only a few prior papers that link analyst recommendations with institutional trading, as we do. Chen and Cheng (2002) find that quarterly institutional trades are correlated with consensus stock recommendations. However, they compare all institutional trading with consensus recommendations, rather than matching institutions with their own recommendations, as we do. The paper potentially most closely related to our own is Chan, Chang, and Wang (2005). Similar to us, they match quarterly trades of financial firms with in-house recommendations. However, their specification as well as the focus differs considerably from ours. The primary objective of our study is to examine conflicts of interest within investment banking. As discussed previously, we believe that quarters around mergers provide an ideal setting to examine such issues. In contrast, Chan et al focus more solely on the value relevance of recommendations in a more general setting, and correspondingly they examine all recommendations, not just those around a corporate event. Consistent with the analyst recommendations having value, they find that in-house trade is more positive around upgrades than downgrades. Differences between their findings and our own are discussed in more detail later. 8 3. Data Our data consists of mergers and acquisitions between 1995 and 2004, as obtained from the Securities Data Company (SDC) database. To ensure that the merger is a material event for the acquiring firm, we require the market value of the target to be at least 5% of the combined market capitalization of the bidder and the target. Both targets and acquirers are public firms traded in the U.S., and the acquirer must be publicly traded for at least three years prior to the merger announcement. We require each bidder firm to be followed by analysts, as listed on the IBES recommendation database, and to have institutional ownership, as listed in the Spectrum 13(f) filings, one year prior to the announcement of the acquisition. Our analysis necessitates merging the SDC merger data, the IBES recommendation data, and the Spectrum institutional holdings data. For each merger, we identify the advisory investment bank from SDC. We match by hand the identity of this bank with the IBES broker code and with the Spectrum institutional name. In matching the institutions between the SDC, IBES, and Spectrum databases, we are careful to account for both mergers between investment banks and for banks reporting under different names (e.g., Smith Barney Inc. and Smith Barney & Co). We attempt to match every investment bank that served as an advisor in at least 10 deals over our sample period. The only banks that were not matched were those such as Houlihan, Lokey, Howard & Zukin and Greenhill & Co, LLC, neither of which have either a trading desk or analysts. Mergers in which the advisor either did not have an advisory arm (i.e., wasn’t listed in IBES), didn’t have a trading arm (i.e., wasn’t listed in Spectrum), or served as an advisor in less than ten deals are omitted from our sample. For our analysis of analyst recommendations, we obtain advisor firm recommendations 9 and consensus recommendations from IBES. The summary data are available monthly, on the 15th day of each month, and they represent the average of all outstanding and new recommendations made during the previous month.5 For the independent analyses of analyst recommendations, we use the analyst recommendations 1, 4, 7, 10, and 13 months prior to the merger announcement, and the recommendations 1, 4, 7, 10, and 13 months following the merger completion. In addition, to assess the effects of the merger completion, we compare analyst data one month following merger completion to one month preceding merger completion. Institutional holdings data are reported in Spectrum quarterly, on March 31st, June 30th, September 30th, and December 31st of each year. We calculate total shares held by each advisor institution and each non-advisor institution one through five quarters prior to merger announcement and one through five quarters following merger completion. Our interest in analysts’ response to the merger and also institutional traders’ response to the analysts dictates the merging of the three datasets. Specifically, we match each quarterly institutional reporting date to the prior analyst consensus. For example, the December 31st institutional data would be matched to the December 15th analyst recommendation data. Thus, for a merger on either October 12th or December 8th, quarter +1 relative to the merger announcement would represent the December 31st institutional data, and we would use the corresponding December 15th analyst recommendation data. Although these examples pose no serious difficulties, there are other cases for which comparing analyst recommendation data and institutional trading data around the merger proves to be especially problematic. Not surprisingly, these difficulties are driven primarily by the Consistent with IBES’ calculations of the consensus estimates, we match an advisor analyst’s recommendation (as obtained from the detail database) with the subsequent consensus recommendation (as obtained from the summary database). 5 10 infrequent interval at which the institutional trading data are available. For example consider a merger on September 23rd. Using the measurement intervals described above, quarter +1 would correspond to the September 30th institutional trading data. However, the preceding analyst consensus data would be from September 15th, a date that actually precedes the merger announcement. There is an obvious difficulty in capturing analyst recommendations to the merger and the corresponding trades of the institutions in such cases. We therefore eliminate any case where the date of the merger announcement falls in between the first available institutional reporting data and the corresponding (i.e., most recent) analyst consensus recommendation data As shown in Table 1, these requirements result in a sample of 726 mergers, of which 403 are stock acquisitions, 96 are cash, and 227 are mixed. Many of the mergers have more than one advisor. Due to our interest in conflicts of interest at the investment bank level, many of our analyses focus on advisor-level recommendations and trading. Our sample includes 816 advisorlevel observations. The sample is spread over time, with the largest number of transactions occurring in the late 1990s. This concentration is consistent with the finding in prior literature that M&A activity tends to be particularly high when the stock market is strong. Looking at the industry distribution, the largest number of mergers is in the business equipment and finance industries. Table 2 provides descriptive statistics on the full sample, the sample divided by the presence or absence of analyst coverage by the advisory firm, and the sample divided by the presence or absence of institutional ownership by the advisory firm. Median market capitalization is measured one month prior to the announcement of the merger, and all other financial data reflect medians measured at the end of the fiscal year preceding the announcement of the merger. Relative merger size represents the market capitalization of the target firm 11 divided by the sum of the market capitalizations of the target and acquirer firm, where all market capitalizations are measured one month prior to the announcement of the merger. The median market capitalization of the acquirer firm is approximately $1.8 billion. We find that the market capitalization is significantly larger for firms in which the advisor firm provides analyst coverage and in which the advisor firm owns shares. This likely reflects the fact that both analyst coverage and institutional ownership are greater in larger firms, as shown by Gompers and Metrick (2001) and Barth, Kasznik, and McNichols (2001). Similar inferences can be made based on the total assets and the sales of the acquirer firms. Market-to-book is significantly higher and working capital as a fraction of total assets is significantly lower for firms in which the advisory firm owns shares. The significant differences in market-to-book are consistent with Barth et al’s (2001) finding that analysts are more likely to cover growth firms. Finally, relative merger size is significantly lower for companies in which the advisor firm issues recommendations. This is potentially driven by differences in firm size – companies in which the advisor firm issues recommendations are significantly larger, meaning a given target size will be relatively smaller. 4. Analyst Recommendations and Institutional ownership around the merger As discussed in section 2, prior literature provides contradictory evidence on the extent of analyst optimism due to conflicts from investment banking. The majority of this prior literature has focused on analyst forecasts and recommendations either around equity offerings or across all firms. However, little is known about analyst recommendations around mergers. Notably, the magnitude of fees from advising M&A banks likely make it one of the most competitive 12 areas of investment banking.6 Therefore, M&A provides a setting in which the pressures from investment banking on a bank’s other activities are potentially the greatest. Tables 3 and 4 examine the extent to which analyst recommendations appear related to either expected or recent M&A advisory business by the investment bank. The first column of Table 3 shows the percent of all 816 advisors (across the 726 mergers) that issue recommendations in the acquirer company, from five quarters prior to the announcement of the merger to five quarters following the completion of the merger. That is, the analysis is at the advisor level rather than the deal level, meaning a deal in which there are two advisors would be represent two observations. We assume that the investment bank has little idea of an upcoming merger on which they could potentially advise five quarters ahead of time. In contrast, one quarter prior to the merger, an investment bank might know that a merger is likely and consequently decide to initiate coverage in the hopes of increasing the chances of winning the advisory business. Results show that the percent of advisors issuing recommendations on the acquirer company increase substantially over time, from 52% of advisors five quarters prior to the merger, to 68% one quarter prior to the merger, 73% one quarter after the completion of the merger, and 82% five quarters after the completion of the merger. Column 2 shows similar increases in the total number of recommendations per company. Moreover, the last column shows that the median market capitalization of the acquirer companies is also increasing substantially over time, suggesting that at least a portion of the increase in coverage is driven by increases in firm size. In an effort to isolate the portion of increases in advisor coverage that is driven by efforts to win M&A advisory business, column 3 shows the percent of total recommendations that 6 For example, as discussed above, fees from advising M&A in 2006 exceeded $35 billion. 13 represent recommendations by the advisor firm. Interestingly, this percentage increases substantially from five quarters prior to the announcement to one quarter following the completion. The substantial increases around the time of the merger are consistent with efforts to win and/or support the M&A advisory business. Columns 4, 5, and 6 of Table 3 show similar statistics for advisor firm ownership in the acquirer company. Similar to inferences from analyst coverage, we find increases in the percentage of advisor institutions that own shares, from 58% five quarters prior to announcement to 69% five quarters following completion. However, there is no evidence that increases in the incidence of the advisor firm owning shares in the acquirer company are related to the merger. In fact, advisors as a percent of all institutions that own shares in the acquirer company actually decrease around the time of the merger, indicating that non-advisor institutions are investing in the acquirer firms for the first time faster than advisor institutions. The observed increases in advisor firm ownership are likely driven by increases in firm size. Finally, column 7 of Table 3 shows the percent of advisors that both issue recommendations and own shares in the acquirer company. Consistent with the other statistics, this percentage increases markedly over time, from 27% five quarters prior to the announcement to 58% five quarters following completion. Table 4 looks more specifically at the dynamics of analyst recommendations and institutional ownership in the period immediately surrounding the merger. Panel A focuses on the analysts’ recommendations, Panel B on institutional ownership across all institutions, and Panel C on institutional ownership across institutions that have an investment banking arm. Looking first at Panel A, we see that non-advisor analysts are more likely to both upgrade and downgrade the acquirer stock following announcement of the acquisition. One potential 14 reason for the greater activity in both directions among non-advisor analysts is that the announcement of the merger was more of a surprise to them. However, it is nevertheless surprising to observe a significantly greater portion of non-advisor analysts upgrading the stock. It is possible that the higher average rating among advisor analysts prior to the merger announcement explains part of this difference. (Moreover, the first two columns of Table 5 show that the average advisor rating continues to be higher following the merger.) More consistent with our expectations is the difference in the percentage of analysts downgrading the acquirer stock following the merger announcement, with 4% of non-advisor analysts downgrading, compared to only 1% of advisor analysts. The third row of Panel A shows the average recommendations as well as upgrades and downgrades following merger completion, where upgrades and downgrades are relative to the quarter prior to completion (but by definition subsequent to the merger announcement). Consistent with predictions, we find that more advisor analysts upgrade the acquirer stock following merger completion; however the difference is not significant. The percentage of analysts downgrading the stock is similar between advisors and non-advisors. Panel B examines similar issues as pertaining to the institutional ownership by the advisor versus non-advisor firms. Columns 2 and 3 of Panel B of Table 4 suggest that the advisor firm institutions are more active traders than non-advisor institutions. They are more likely to both purchase and sell shares in the acquirer company in the quarter prior to and following announcement. Panel C suggests that at least a portion of this difference reflects differences between investment bank versus non-investment bank institutions. It is possible that institutions behave differently if they have an investment banking business. Such institutions may be pressured to 15 support the investment banking business. Alternatively, the traders associated with investment banks may hold different positions for other reasons, for example lower trading costs, better information on more firms, etc. To examine the extent to which such issues affect our analysis, we restrict the sample of institutions to those that have an investment banking business. Specifically, we restrict the sample of institutions to those who are in our M&A sample, i.e., those that served as advisors in at least ten deals over our sample period and had both analysts and a trading desk. Once we restrict the sample of institutions to those with a significant investment banking business (i.e., Panel C), we find that institutions of the advisor firm are significantly more likely to buy shares in the acquirer company, but there is no difference in the probability of selling shares after the merger announcement. Finally, both panels B and C indicate that non-advisor institutions are more likely to purchase shares and less likely to sell shares following completion of the merger. This is exactly opposite our prediction. In an attempt to understand this result, we re-calculate these statistics based solely on cash mergers. Within the cash subsample, we find no evidence that non-advisor firms are more likely to buy shares following completion. This leads us to believe that this apparently puzzling result in the full sample reflects the exchange of shares in stock mergers, in particular the institutions that had held shares in the both the target and acquirer company now owning more shares of the combined entity. In sum, Tables 3 and 4 provide substantial evidence of both analysts and institutional traders behaving differently when they belong to the advisor versus non-advisor investment bank. The next section examines this conclusion in more depth, by investigating the relation between analyst recommendations and institutional trades. 16 5. Relation between analyst recommendations and institutional trades As a first step towards understanding the relation between analyst recommendations and institutional trades, Table 5 provides some descriptive evidence. Quarters -3 through -1, as labeled in the top row of each panel, represent the quarters prior to the merger announcement, and quarters +1 through +5 represent quarters following merger completion. For example, consider a merger announced on April 12th, 1998 and completed on August 20th, 1998. In this case, the quarter -1 institutional trading data would be measured on March 31st, 1998 and the quarter -1 analyst recommendation data would be measured on March 15th, 1998. For quarter -2, institutional trading data would be measured on December 31st, 1997 and analyst recommendation data would be measured on December 15th, 1997. Analogously, quarter +1 institutional trading data would be measured on September 30th, 1998 and quarter +1 analyst recommendation data would be measured on September 15th, 1998. Upgrades and downgrades are measured relative to the analyst’ recommendation one quarter prior. Looking at the first row of Panel A, we see that there are 18 instances of advisor firm analyst upgrades three quarters prior to merger announcement. Of these 18 cases, 44% are associated with decreases in institutional holdings by the advisor firm, relative to the previous quarter, and 50% are associated with increases in institutional holdings by the advisor firm. To the extent that analyst upgrades are viewed as positive information, we would expect them to be associated with a greater percentage of affiliated institutions buying shares, versus selling shares. However, we see little evidence of this at quarter -3. Evidence of traders following affiliated analysts recommendation changes is strongest in quarters -1 and +2. (Note that inferences on quarter +1 are difficult because the institutions are 17 so often net buyers of shares in stock acquisitions.) For example, upon an affiliated analyst upgrade in quarter -1, 67% of in-house institutions (i.e., institutions within the same investment bank) buy shares compared to only 27% selling. Similarly, in quarter +2 after the merger, an analyst upgrade is associated with 68% of in-house institutions buying compared to only 32% selling. However, little if any relation is seen in the quarters farther from the merger. Inferences are similar based on downgrades. To the extent that downgrades contain negative information, we would expect a greater frequency of selling by affiliated institutions, compared to buying. Consistent with this prediction, an affiliated analyst downgrade in quarter +2 is associated with 65% of in-house traders selling, compared to only 23% buying. Statistics are similar for quarters 3 and 4, albeit weaker. However, there is no evidence of any relation in other quarters. In sum, Table 5 shows a substantially stronger relation between analyst upgrades/downgrades and in-house trades immediately around the merger, and little if any relation at other times. Our evidence is somewhat inconsistent with the findings of Chan, Cheng, and Wang (2005) who find a significant relation between analyst recommendations and in-house trading throughout time. It is possible that their larger sample size (which they obtain by looking at all firms across a ten-year sample period) gives them more power to find significant differences. Nevertheless, it remains the case that our analysis suggests a stronger relation around the merger event. Tables 6 and 7 examine this relation between analyst recommendations and changes in institutional holdings in a regression framework. The matching of the merger data, institutional holdings data, and analyst recommendations data is as described above. The dependent variable is the change in advisor shares held from quarter t-1 to t, divided by the quarter t-1 holdings. For 18 each firm, regression observations include t-4 to t-1 (relative to announcement) and t+1 to t+5 (relative to completion). We omit the period of time between the announcement and completion because the time varies so substantially across firms, from 0 days (in cases where the announcement and completion date are the same) to over 18 months. In each regression, the independent variable of interest is the change in advisor analyst recommendation, defined as the advisor recommendation outstanding immediately prior to the quarter t holdings date minus the advisor recommendation outstanding immediately prior to the quarter t-1 holdings date. This change is then multiplied by -1, so that a positive recommendation change can be interpreted as an upgrade and a negative recommendation change as a downgrade. Control variables include dummies for the level of the advisor recommendation at the end of quarter t-1, to account for the fact that an analyst with a strong buy cannot upgrade. We only include dummies for strong buy, buy, and hold, because there are so few observations with lower recommendations (sells and strong sells). Finally, we also include the change in market capitalization of the acquirer firm, to account for the strong positive relation between institutional ownership and firm size. Institutions are arguably more likely to increase shareholdings when firm size is increasing. Looking first at Column 1, we see a significant positive relation between changes in advisor institution shareholdings and changes in advisor analyst recommendations. It appears that advisor firm institutions are trading consistent with advisor firm analyst recommendations. However, column 2 shows that this relation is actually only significant for the quarters following merger completion. The interaction term, change in advisor analyst recommendation * post merger dummy, is significantly positive, with a t-statistic of 3.07. In contrast, the interaction term advisor analyst recommendation * pre merger dummy is not significant. Results suggest 19 that traders buy and sell in line with analyst recommendation changes following a merger, but not before. Finally, column 3 compares the effects of advisor firm versus non-advisor firm analysts on advisor firm institutional trading. In the post merger period, advisor firm traders appear to pay close attention to their own analysts, but disregard non-advisor analysts. In contrast, in the pre merger period, the advisor firm traders pay significant attention to the non-advisor consensus analyst estimate, but disregard their own firm’s analysts. Regressions in Table 7 are specified similarly to those in Table 6, except for the fact that the sample is restricted to the quarters following the completion of the merger, the only period where we find evidence of a relation between analyst recommendations and institutional trades by the advisor firm. Table 7 includes various control variables that might affect institutional trading, such as method of payment (cash versus stock), relative size of the merger, and market capitalization of the acquirer firm. Inferences are all similar, showing a significant positive relation between changes in advisor analyst recommendations and changes in advisor institutional holdings. Results in tables 6 and 7 are consistent with advisor firm analysts having particularly valuable information on the acquirer firm following a merger, and consequently the in-house traders acting on their recommendations. In contrast, prior to the merger the advisor firm analysts have no better information than analysts at other banks, and consequently the traders are more likely to look at consensus recommendations. While this interpretation fits our results, it does not incorporate any conflicts of interest, which analysts are likely to face. We investigate in more detail the role of analyst information versus analyst conflicts of interest in the next section. 20 6. Information sharing versus conflict of interest Results from tables 5, 6, and 7 indicate that advisor firm institutions trade in line with advisor firm analyst recommendation changes in the quarters immediately following the merger. To the extent that the advisor firm analysts have particularly valuable information, this is exactly what we would expect. However, as discussed earlier, there is also substantial evidence that affiliated analysts face serious conflicts of interest – the investment bank advising the acquirer firm might encourage its analysts to upgrade the acquirer stock. In this section, we investigate in more detail the effects of such conflicts of interest. To examine the conflict of interest motivation for advisor analyst upgrades, we classify investment banks into various categories based on their source of revenue. We posit that analysts working for institutions in which investment banking is a more important source of revenue will face greater conflicts of interest, for example stronger pressures to upgrade stocks of companies for which the bank has recently served as advisor on an acquisition For each advisor investment bank, we download the income statement from the bank’s 10K. Data limitations restrict this sample to those investment banks that are publicly traded. This limits us to 25 of the investment banks. However, these 25 banks served as advisors in the vast majority of our acquisitions. Investment banks are required to describe the source of their revenues, and the banks generally break down the revenues into those from investment banking, those from proprietary trading, and also those from various other activities on which we are not focusing. Thus, for each bank and each year, we are able to determine the percent of revenues from investment banking versus proprietary trading. For each year, we classify firms with above-median percent of revenues from investment banking as high investment banking firms. Similarly, firms with above-median percent of revenues from trading are classified as high 21 trading firms. If the analysts from the high investment banking firms face more serious conflicts of interest, then we would expect these analysts to be more optimistic regarding the acquirer firms, particularly in the period following the merger. Table 8 provides evidence consistent with this prediction. Similar to previous analyses, all recommendations are for the acquirer company and they are by the analysts working for the acquirer’s advisor. The first row of the table shows that the average recommendation by the analysts working at low IB banks (i.e., banks where investment banking is a less important source of revenue) is 2.00, compared to 1.87 for analysts working at high IB banks. Recalling that lower numbers represent more optimistic recommendations, this indicates that the analysts working for banks where the conflict of interest from investment banking is likely to be the greatest offer more optimistic recommendations. The difference of 0.12 is significant at the 1% level. Rows 2 and 3 of Table 8 indicate that the difference between the recommendations of analysts at high IB versus low IB banks is only significant over the period following the merger. Prior to the merger, the recommendations of the two groups of analysts do not differ significantly. However, following the merger, the average recommendation of the low IB banks is 2.04, compared to 1.86 for the high IB banks. The difference of 0.17 is significant at the 1% level. Results in Table 8 suggest that analysts at the high IB banks are more optimistic regarding the acquirer, possibly in an effort to support the investment banking business. If this is the case, then we would expect traders at these banks (i.e., the high IB banks) to be less likely to act on their analysts’ recommendation changes. Table 9 examines this proposition. Similar to Table 7, Table 9 shows regressions of the changes in advisor firm institutional 22 holdings on changes in advisor firm analyst recommendations in the five quarters following the merger. In fact, column 1 of Table 9 is nearly identical to Column 1 of Table 7, the only difference being that it is based on the smaller sample size used throughout Table 9, i.e., on those mergers for which we have sources of revenues for the advisor investment bank. Similar to prior findings, we find a significant positive relation between changes in advisor analyst recommendations and changes in advisor institutional shareholdings. In column 2, we replace the change in advisor analyst recommendation with two interaction terms, the change in advisor analyst recommendation times the high investment bank dummy and the change in advisor analyst recommendation times the low investment bank dummy. (Recall that the high (low) investment bank dummy equals 1 for those advisor firms for which revenue from investment banking is above (below) the median, 0 otherwise). Results suggest that the significantly positive relation between advisor firm recommendations and trading only exists among those banks that receive a relatively low portion of their revenue from investment banking. It seems that the investment bank traders avoid following the in-house analyst recommendations around mergers in those cases where the conflicts of interest from investment banking are likely to be the greatest, i.e., in those banks that generate the largest portions of their revenues from investment banking. The regression in column 3 is similar to that in column 2, except that it divides the recommendation changes according to the importance of trading revenue for the bank. Interestingly, results indicate that the positive relation between advisor analyst recommendation changes and in-house trading is concentrated among those firms that rely most heavily on trading as a source of revenue. Perhaps banks that rely most on trading revenues provide different incentives to their analysts. Results suggest that such banks encourage their analysts to make the 23 most informative recommendations, presumably to help the traders, rather than the most optimistic recommendations that might aid the investment banking business. Finally, the regression in column 4 shows that inferences on the importance of investment banking revenue and of trading revenue are robust to including interaction terms for both within a single regression. 7. Conclusion Conflicts of interest are pervasive within investment banks. Prior literature has not reached a consensus on the extent to which such conflicts affect analyst recommendations. We take a new approach to this problem, by examining the interaction between analyst recommendations and in-house trading. In addition, we broaden the examination of conflicts of interest, by looking at the extent to which such conflicts affect both the analysts and the proprietary traders. Because we are interested in conflicts of interest, we choose a setting in which such conflicts are likely to be severe: mergers and acquisitions. We find that advisor firm analysts tend to be significantly more optimistic about the acquirer firms than non-advisor analysts. Moreover, the advisor-firm trading desk buys and sells on advisor firm analyst recommendations around the time of the merger. However, findings suggest that this positive relation is restricted to those settings where the conflict faced by analysts is likely to be less severe. In the banks that rely most on investment banking business, traders are significantly less likely to listen to their analysts. In contrast, in banks that rely most heavily on trading as a source of revenue, traders rely significantly more on their analysts’ recommendations, suggesting that these analysts’ recommendations are less biased. 24 References Aggarwal, R., Purnanandam, A., Wu G., 2005. Underwriter manipulation in initial public offerings. Unpublished working paper, University of Minnesota. Agrawal, A., Chen, M., 2004. Analyst conflicts and research quality. Unpublished working paper, University of Alabama. Agrawal, A., Chen, M., 2004. Do analyst conflicts matter? evidence from stock recommendations. Forthcoming, Journal of Law and Economics. Barber, B., Lehavy, R., McNichols, M., Trueman, B., 2006. Buys, holds, and sells: the distribution of investment banks’ stock ratings and the implications for the profitability of analysts’ recommendations. Journal of Accounting and Economics 41, 87-117. Barber, B., Lehavy, R., Trueman, B., 2007. Comparing the stock recommendation performance of investment banks and independent research firms. Journal of Financial Economics 85, 490-517 Barth, M., Kasznik, R., McNichols, M., 2001. Analyst coverage and intangible assets. Journal of Accounting Research 39, 1-34. Becher, D., Juergens, J., 2005. Analyst recommendations and mergers: do analysts matter? Unpublished working paper, Arizona State University. Bradley, D., Jordan, B., Ritter, J., 2006. Analyst behavior following IPOs: the ‘bubble period’ evidence. Review of Financial Studies, forthcoming. Busse, J., Green, T.C., 2002. Market efficiency in real time. Journal of Financial Economics 65, 415-437. Carhart, M., 1997. On persistence in mutual fund performance. Journal of Finance 52, 57-82. 25 Chan, K., Chang, C., Wang, A., 2005. Put your money where your mouth is: do financial firms follow their own recommendations? Unpublished working paper, Cornell Unversity. Chen, X., Cheng, Q., 2002. Institutional holdings and analysts’ stock recommendations. Unpublished working paper. Clarke, J., Khorana, A., Patel, A., Rau, R., 2004. The good, the bad and the ugly? Differences in analyst behavior at investment banks, brokerages, and independent research firms. Unpublished working paper, Purdue University. Cowen, A., Groysberg, B., Healy, P., 2006. Which types of analyst firms are more optimistic? Journal of Accounting and Economics 41, 119-146. DeLong, G., 2001. Stockholder gains from focusing versus diversifying bank mergers. Journal of Financial Economics 59, 221-252. Dugar, A., Nathan, S., 1995. The effects of investment banking relationships on financial analysts’ earnings forecasts and investment recommendations. Contemporary Accounting Research 12, 131-160. Fama, G., French, K., 1993. Common risk factors in the returns on stocks and bonds. Journal of Financial Economics 33, 3-56. Givoly, D., Lakonishok, J., 1979. The information content of financial analysts’ forecasts of earnings: some evidence on semi-strong inefficiency. Journal of Accounting and Economics 1, 165-185. Gompers, P. and A. Metrick, 2001. Institutional investors and equity prices. Quarterly Journal of Economics 116, 229-259. 26 Hong, H., Kubik, J.D., 2003. Analyzing the analysts: career concerns and biased earnings forecasts. Journal of Finance 58, 313-351. Jackson, A.R., 2005. Trade generation, reputation, and sell-side analysts. Journal of Finance 60, 673-717. Jacob, J., Rock, S., Weber, D., 2003. Do analysts at independent research firms make better earnings forecasts? Unpublished working paper, University of Colorado, Boulder. James, C., Karceski, J., 2006. Strength of analyst coverage following IPOs. Journal of Financial Economics 81, 1-34. Jegadeesh, N., Kim, J., Krische, S., Lee, C., 2004. Analyzing the analysts: when do recommendations add value? Journal of Finance 59, 1083-1124. Leaven, L., Levine, R., 2007. Is there a diversification discount in financial conglomerates? Journal of Financial Economics 85, 331-367. Lin, H.-W., McNichols, M., 1998. Underwriting relationships, analysts’ earnings forecasts and investment recommendations. Journal of Accounting and Economics 25, 101-127. Ljungqvist, A., Marston, F., Starks, L, Wei, K., Yan, H., 2007. Conflicts of interest in sell-side research and the moderating role of institutional investors. Journal of Financial Economics 85, 420-456 Loh, R., Mian, G.M., 2005. Do accurate earnings forecasts facilitate superior investment recommendations? Journal of Financial Economics, forthcoming. McNichols, M., O’Brien, P., Pamukcu, O., 2006. That ship has sailed: unaffiliated analysts’ recommendation performance for IPO firms. Unpublished working paper, Stanford University. 27 Mehran, H., Stulz, R., 2007. The economics of conflicts of interest in financial institutions. Journal of Financial Economics 85, 267-296 Michaely, R., Womack, K., 1999. Conflict of interest and the credibility of underwriter analyst recommendations. Review of Financial Studies 12, 653-686. Mikhail, M., Walther, B., Willis, R., 1999. Does forecast accuracy matter to security analysts? Accounting Review 74, 185-200. Mitchell, M., Stafford, E., 2000. Managerial decisions and long-term stock price performance. Journal of Business 73, 287-329. Moeller, S., Schlingemann, F., Stulz, R., 2005. Wealth destruction on a massive scale? a study of acquiring-firm returns in the recent merger wave. Journal of Finance 60, 757-782. Ramnath, S., Rock, S., Shane, P., 2006. A review of research related to financial analysts’ forecasts and stock recommendations. Unpublished working paper, Georgetown University. Stickel, S., 1991. Common stock returns surrounding earnings forecast revisions: more puzzling evidence. Accounting Review 66, 402-416. Womack, K., 1996. Do brokerage analysts’ recommendations have investment value? Journal of Finance 51, 137-167. 28 Table 1: Descriptive Statistics on M&A Sample The sample consists of 726 mergers over the 1995 to 2004 period. For a merger to be included in the sample, the acquirer firm must be followed by at least one analyst, as listed in the IBES database, and be owned by at least one institutional investor, as listed in the Spectrum database. In addition, the target market capitalization must be at least 5% of the combined market capitalization of the target plus acquirer, where all market capitalizations are measured one month prior to the merger announcement. A merger with two advisors is treated as two advisor-level observations; there are 816 advisor observations across the 726 mergers. Mergers are classified into industries based on the Fama-French 12 industry groupings. Relative Size > 5% Number of advisor observations 816 Number of unique mergers 726 Stock 403 Cash 96 Mixed 227 Year # Mergers Industry # Mergers 1995 71 Consumer Nondurables 13 1996 72 Consumer Durables 8 1997 120 Manufacturing 65 1998 126 Oil, gas, coal extraction 40 1999 85 Chemicals and allied products 14 2000 80 Business Equipment 149 2001 47 Telephone & TV transmission 19 2002 34 Utilities 22 2003 56 Wholesale, Retail 55 2004 35 Healthcare, Med. Eqpt, Drugs 60 Finance 221 Other 60 29 Table 2: Descriptive Statistics – 5% sample – Medians Descriptive statistics are provided for the sample of 726 mergers over the 1995 – 2004 time period. All variables, with the exception of relative merger size, refer to the acquirer firm, and all statistics represent medians. Market capitalization (in millions) is measured one month prior to the announcement of the merger. All other financial variables are measured at the fiscal year end preceding the merger announcement. Market-to-book equals the equity market capitalization divided by the book value of equity. Book leverage equals the sum of short-term and long-term debt, divided by total assets. Market leverage equals the sum of short-term and long-term debt divided by the total firm market value, where total firm market value equals total assets plus market value of equity minus the book value of equity. Total assets, sales, sales/TA, EBIT/TA, and WC/TA are computed using the relevant Compustat data items. Relative merger size equals the target market capitalization divided by the combined market capitalization of the target plus acquirer, where all market capitalizations are measured one month prior to the merger announcement. Statistics are computed for the whole sample, conditional on whether or not the advisor bank to the acquirer firm has an analyst issuing recommendations on the acquirer, as listed on IBES, and conditional on whether or not the advisor bank to the acquirer firm owns shares in the acquirer firm, as reported on Spectrum. Asterisks denote whether the advisor analyst vs. no advisor analyst statistics are significantly different, and similarly whether then advisor institutional ownership vs. no advisor institutional ownership are significantly different (*, **, *** represent the 10, 5, and 1% levels of significance). Whole Sample Advisor Analyst Following (n=473) No Advisor Analyst Following (n=253) Advisor Institutional Ownership (n=443) No Advisor Institutional Ownership N=283) Market Cap (mil) 1,855 2,274 1,201*** 3,007 906*** Total Assets (mil) 1905 1974 1806* 2832 1,141*** Sales (mil) 865 1,007 1,329 496*** Sales / TA 0.64 0.67 0.57 0.65 0.64 MB 2.52 2.57 2.36 2.64 2.20*** Book leverage 0.20 0.21 0.18 0.21 0.18 Market leverage 0.13 0.13 0.12 0.13 0.13 EBIT / TA 0.08 0.08 0.07* 0.08 0.07 WC / TA 0.21 0.22 0.21 0.18 0.24** Relative Merger Size 28% 27% 32%** 29% 646*** 27% 30 Table 3: Incidence of advisor recommendations and share ownership in the acquirer companies This table provides information on the incidence of advisor recommendations and advisor institutional ownership in the acquirer company, from five quarters prior to the announcement of the merger to five quarters following the completion of the merger. Percent of advisors represents the percentage of the 816 advisor-level observations in which the advisor bank to the acquirer had an analyst following the acquirer. Average number of recs per company represents the average number of analysts covering each acquirer firm. Percent of total recs by advisor equals the number of advisors covering each firm divided by the total number of recs in each firm, averaged across the 726 mergers. Percent of advisors that own shares represents the percentage of the 816 advisor-level observations in which the advisor bank to the acquirer owned shares in the acquirer. Average # insts invested in co equals the total number of institutions invested in each acquirer firm, averaged across all mergers. Advisors as a % of total equals the number of advisors owning shares in each firm divided by the total number of institutions owning shares in each firm, averaged across the 726 mergers. Percent of advisors that issue recs and own shares equals the percent of the 816 advisors to the acquirer firms that both have an analyst following the acquirer and own shares in the acquirer. Company mkt cap equals the median market capitalization of the acquirer firm. Issuance of Recommendations Ownership of Shares % of Advisors that Issue Recs and own Shares % of Advisors Avg # Recs per Company % of Total Recs that are by Advisor % of Advisors that Own Shares Avg # Insts invested in co Advisors as % of total insts 5 qtrs pre- ann’t 4 qtrs pre- ann’t 3 qtrs pre- ann’t 2 qtrs pre- ann’t 1 qtrs pre- ann’t 52% 57% 60% 64% 68% 10.3 10.5 10.8 11.2 11.4 7.8% 8.3% 8.9% 9.1% 9.4% 58% 59% 59% 61% 62% 173.9 180.1 187.2 195.3 202.6 0.58% 0.55% 0.52% 0.52% 0.48% 27% 34% 37% 39% 43% $1,537 $1,637 $1,747 $1,800 $1,995 1 qtr post-completion 2 qtrs post-completion 3 qtrs post-completion 4 qtrs post-completion 5 qtrs post-completion 73% 76% 79% 81% 82% 11.5 12.1 12.5 12.7 13.0 10.3% 10.3% 10.2% 9.9% 9.7% 66% 68% 69% 70% 69% 241.9 247.1 250.2 251.3 250.3 0.43% 0.43% 0.45% 0.46% 0.47% 51% 53% 55% 58% 58% $2,495 $2,571 $2,602 $2,592 $2,698 Company Mkt Cap ($mil) 31 Table 4: Changes in the characteristics of advisor recommendations and share ownership in the acquirer companies In Panel A, statistics are computed one month prior to the merger announcement, one month after the merger announcement, and one month following the merger completion. For each time period, analyst upgrades and downgrades are relative to the prior month (i.e., for the one month after completion row, upgrades are relative to the month preceding merger completion). An analyst upgrade represents a case where an analyst changed its recommendation on the acquirer company to be more positive (e.g., from a 2 to 1, which represents a buy to a strong buy). In Panel B, statistics are computed one quarter prior to the merger announcement, one quarter after the merger announcement, and one quarter following the merger completion. For each time period, the percent of institutions that purchase additional shares or sell shares is measured relative to the previous quarter, similar to Panel A. Asterisks represent significant differences between the advisor and non-advisors for each catetgory different (*, **, *** represent the 10, 5, and 1% levels of significance). Statistics in Panel C are similar to those in Panel B, except that non-advisors only include investment banks in our sample (as opposed to all institutions). Panel A: Advisor vs. Non-Advisor Analyst recommendations – All Analysts Average Recommendations 1 mth prior to ann’t 1 mth after ann’t 1 mth after completion Advisor Non-Advisor 1.88 1.88 1.85 2.04*** 2.01*** 1.99*** % Analysts with upgrades NonAdvisor Advisor 1.1% 2.3%** 1.7% 5.5%*** 3.9% 2.6% % Analysts with downgrades Advisor Non-Advisor 1.8% 1.4% 2.5% 2.3% 3.9%*** 2.3% Panel B: Advisor vs. Non-Advisor Institutional Ownership – All Institutions Average Ownership by: 1 qtr prior to ann’t 1 qtr after ann’t 1 qtr after completion Total Institutional Ownership 59.0% 60.9% 60.5% Advisor 0.47% 0.47% 0.46% NonAdvisor 0.47 0.45 0.37** % Inst’s that purchased additional shares % Inst’s that sold shares Advisor NonAdvisor Advisor NonAdvisor 55% 54% 63% 39%*** 39%*** 66% 41% 42% 37% 38%* 38%* 27%*** 32 Table 4 (cont.) Panel C: Advisor vs. Non-Advisor Institutional Ownership – Investment Bank Institutions Only Average Ownership by: 1 qtr prior to ann’t 1 qtr after ann’t 1 qtr after completion Total Institutional Ownership 57.3% 59.1% 59.1% Advisor 0.47% 0.47% 0.46% NonAdvisor 0.43% 0.43% 0.36% % Inst’s that purchased additional shares % Inst’s that sold shares Advisor NonAdvisor Advisor NonAdvisor 55% 54% 63% 44%*** 46%*** 67%*** 41% 42% 37% 42% 42% 29%*** 33 Table 5: Relation between advisors’ recommendations and institutional holdings For Panel A, each quarter, from three quarters prior to the merger announcement to five quarters following the merger completion, we compute the total number of advisor analyst upgrades (i.e., upgrades to the acquirer stock by the investment bank that is advising the acquirer) relative to the end of the previous quarter. For each of these upgrades, we compute the percentage of cases in which the advisor firm sold shares in the acquirer stock vs. bought shares in the acquirer stock during that same quarter. Panel B shows analogous statistics for advisor analyst downgrades. Panel A: Changes in Advisors’ institutional holdings when Advisor analysts Upgrade the acquirer Quarter -3 -2 -1 1 2 3 4 5 Sell Shares 44% 40% 27% 27% 32% 53% 47% 37% Buy Shares 50% 60% 67% 73% 68% 47% 53% 63% 18 20 15 37 28 17 19 16 N Obs Panel B: Changes in Advisors’ institutional holdings when Advisor analysts Downgrade the acquirer Quarter -3 -2 -1 1 2 3 4 5 Sell Shares 47% 31% 18% 38% 65% 50% 67% 46% Buy Shares 47% 63% 76% 62% 23% 40% 28% 46% 15 16 17 29 26 20 40 35 N Obs 34 Table 6: Determinants of change in acquirer shares held by the advisor investment bank This table shows fixed effects regressions, where the dependent variable is the change in shares owned by the advisor investment bank in the acquirer company. Regressions are estimated over the 816 advisor-level observations, for five quarters prior to the merger announcement to five quarters following the merger completion, excluding those quarters in between the announcement and completion. The first independent variable is the change in the advisor analyst recommendation of the acquirer company (measured over the same quarter but observed prior to the measurement of institutional ownership). This recommendation is interacted with both a pre-merger dummy and a post-merger dummy (equal in 1 in the quarters prior to and following the merger completion, respectively, 0 otherwise). Dummies for the level of the advisor recommendation at the beginning of the quarter are included (strong buy, buy, and hold). The change in market capitalization represents the change in the market capitalization of the acquirer company over the quarter. The last column also includes the change in the consensus recommendation of the non-advisor analysts over the quarter and the level of the consensus recommendation among non-advisor analysts at the beginning of the quarter. All recommendation variables are multiplied by negative 1, such that higher recommendations and increases in recommendations can be interpreted as more positive recommendations. T-statistics are shown in parentheses. Model 1 Model 2 Model 3 Rec * Pre- Merger dummy (Advisor) 0.30 (0.78) 0.34 (0.87) Rec * Post Merger dummy (Advisor) 0.75*** (3.07) 0.63** (2.52) Rec (Advisor) *** 0.57 (2.75) Rec * Pre Merger dummy .(Non-Adv) 1.80** (2.56) Rec * Post Merger dummy (Non-Adv) -0.02 (-0.05) Strong Buy Dummy (Advisor) Buy Dummy (Advisor) Hold Dummy (Advisor) 0.63 (0.68) 0.39 (0.44) 0.04 (0.05) 0.64 (0.70) 0.43 (0.49) 0.11 (0.12) Rec level non-advisor Mkt Cap R-Squared N Obs 0.02* (1.85) 12% 3876 0.02* (1.85) 12% 3876 0.19 (0.20) 0.10 (0.12) -0.06 (-0.07) 0.92*** (2.61) 0.02* (1.72) 13% 3846 35 Table 7: Determinants of change in acquirer shares held by the advisor investment bank, Quarters following the merger completion This table shows fixed effects regressions, where the dependent variable is the change in shares owned by the advisor investment bank in the acquirer company. Regressions are estimated over the 816 advisor-level observations, over the first five quarters following the merger completion. The first independent variable is the change in the advisor analyst recommendation of the acquirer company (measured over the same quarter but observed prior to the measurement of institutional ownership. Dummies for the level of the advisor recommendation at the beginning of the quarter are included (strong buy, buy, and hold). Rec non-advisor equals the change in the consensus recommendation of the non-advisor analysts over the quarter, and Rec level non-advisor equals the level of the consensus recommendation among non-advisor analysts at the beginning of the quarter. The change in market capitalization represents the change in the market capitalization of the acquirer company over the quarter, and the level of market capitalization equals the acquirer company market capitalization measured at the beginning of the quarter. The stock dummy equals 1 if the merger was paid for with stock, 0 otherwise. The cash dummy is measured analogously. Relative merger size equals the target market capitalization divided by the combined market capitalization of the target plus acquirer, where all market capitalizations are measured one month prior to the merger announcement. All recommendation variables are multiplied by negative 1, such that higher recommendations and increases in recommendations can be interpreted as more positive recommendations. Tstatistics are shown in parentheses. Rec (Advisor) Strong Buy Dummy (Advisor) Buy Dummy (Advisor) Hold Dummy (Advisor) Model 1 Model 2 0.76** (2.48) 0.51 (0.35) 1.19 (0.88) 0.22 (0.17) 0.78** (2.52) 0.57 (0.39) 1.23 (0.91) 0.23 (0.18) 0.02* (1.82) 0.03** (2.03) -0.02 (-0.93) 2.67 (0.16) 2.45 (0.13) -4.28 (-0.15) Rec non-advisor Rec level non-advisor Mkt Cap Mkt Cap Stock dummy Cash dummy Relative merger size R-Squared N Obs 20% 2112 20% 2112 Model 3 0.59* (1.88) -0.36 (-0.25) 0.53 (0.39) -0.11 (-0.08) 0.61 (1.06) 2.64*** (3.94) 0.02* (1.75) 21% 2100 36 Table 8: Analyst Recommendations, conditional on source of investment bank revenue Average analyst rankings of the acquirer company by the advisor investment bank are compared across categories of investment banks. Low (high) IB banks represents those banks whose investment banking revenues as a percent of total revenues fall below (above) the median (when ranked across all investment banks in that year). Average rankings are reported over the entire event period (five quarters prior to the merger announcement through five quarters after the merger completion), the pre-merger period (the five quarters prior to the merger announcement), and the post-merger period (the five quarters following the merger completion). The last column reports the difference in recommendation level, between low IB banks and high IB banks. Asterisks report the significance of the difference (*, **, *** represent the 10, 5, and 1% levels of significance).. Low IB Banks High IB Banks Difference All Quarters 2.00 1.87 0.12*** Pre-Merger 1.96 1.90 0.06 Post-Merger 2.04 1.86 0.17*** 37 Table 9: Determinants of change in acquirer shares held by the advisor investment bank, Conditional on sources of advisor investment bank revenue This table shows fixed effects regressions, where the dependent variable is the change in shares owned by the advisor investment bank in the acquirer company. Regressions are estimated over the 816 advisor-level observations for which we are able to obtain sources of revenue data for the advisor investment bank, over the first five quarters following the merger completion. The first independent variable is the change in the advisor analyst recommendation of the acquirer company (measured over the same quarter but observed prior to the measurement of institutional ownership). The recommendation change is interacted with a high IB dummy, equal to one if the advisor bank’s investment banking revenues as a fraction of total revenues fell above the median (when ranked across all investment banks in that year) for the year of the merger, and zero otherwise. The recommendation change is also interacted with a high Trade dummy, equal to one if the advisor bank’s trading revenues as a fraction of total revenues fell above the median (when ranked across all investment banks in that year) for the year of the merger, and zero otherwise. Similarly, recommendation changes are also interacted with low IB dummies and low trade dummies. Dummies for the level of the advisor recommendation at the beginning of the quarter are included (strong buy, buy, and hold). The change in market capitalization represents the change in the market capitalization of the acquirer company over the quarter. T-statistics are shown in parentheses. Model 1 Rec (Advisor) Model 2 Model 3 0.55* (1.90) Model 4 -0.87* (-1.67) Rec * Low IB (Advisor) 0.94** (2.50) Rec * High IB (Advisor) 0.14 (0.37) 1.58*** (2.82) Rec * Low Trade (Advisor) 0.24 (0.69) Rec * High Trade (Advisor) 1.06** (2.49) 1.63*** (2.81) Strong Buy Dummy (Advisor) 0.01 (0.01) -0.01 (-0.01) -0.08 (-0.06) -0.22 (-0.17) Buy Dummy (Advisor) 1.04 (0.89) 1.06 (0.91) 0.93 (0.80) 0.88 (0.76) Hold Dummy (Advisor) 0.07 (0.06) 0.12 (0.10) -0.11 (-0.09) -0.18 (-0.16) Mkt Cap 0.02 (1.48) 0.02 (1.60) 0.02 (1.34) 0.02 (1.45) R-Squared N Obs 20% 1388 20% 1388 20% 1388 21% 1388 38