Do investment banks have incentives to help clients make value-creating acquisitions?

Abstract

To many observers, it has long seemed evident that there is a potential conflict between the interests of the investment bankers that do M&A advisory work and the shareholders of the acquiring companies they advise. The potential conflict arises because advisory contracts are structured to reward the bankers for “getting deals done,” with much less reward for deals that do not get done. In other words, contracts are structured so that the bankers generate the lion's share of their fees from those transactions in which their corporate clients end up acquiring the companies they target—but negligible amounts for advisory work that does not lead to a transaction.

Unfortunately for shareholders of the acquiring corporation, overpaying for an acquisition is a fairly surefire way of ensuring that an acquisition takes place. Indeed, there is a body of evidence from the 1970s and 1980s that suggests that acquirers, on average, were willing to do just that.1 In the many M&A deals that got done during those decades, the shareholders of the companies acquired usually seemed to fare significantly better, on average, than the shareholders of the companies doing the acquiring.2

And yet, as that columnist went on to point out, Wasserstein's career on Wall Street did not seem to have suffered from his reputed indifference to the shareholders of his corporate clients. Early scholarly evidence on that question tended to support the notion that banks and bankers were not penalized for facilitating overpriced deals.

In a study that was recently published in the Review of Financial Studies, we re-examined the evidence on the questions: Do bankers pay any penalty for advising on value-destroying acquisitions? Or, conversely, is there any reward to bankers for creating value for their acquisition-minded clients? These questions would seem to be important given that, in the United States alone, corporate acquirers paid investment banks over $20 billion in advisory fees to facilitate their acquisitions during the decade 2002–2011.5

One of the first studies of advisory contracts in mergers and acquisitions was conducted by Robyn McLaughlin, while a finance professor at Boston College, and its findings were published in an article in the Journal of Financial Economics in 1990. McLaughlin studied advisory contracts in corporate tender offers from 1978 to 1985. He observed that the compensation advisors are paid—the advisory fees—were not contingent on whether the transaction creates value for the client, which is the acquirer. In the typical contract, more than 80% of the advisory fee was paid only if the acquisition was completed. He noted that such contracts appeared to create a severe conflict of interest in which the advisor had an incentive to complete the acquisition regardless of the valuation consequences for the acquirers’ shareholders.

When discussing his findings, McLaughlin went on to speculate that other mechanisms, such as the reputation of the advisor, could possibly work to limit the conflict of interest between the investment banker and its acquirer client. That is, value-creating acquisitions can generate reputational capital for good advisors that help them obtain future advisory mandates, while the opposite would be the case for advisors involved in value-reducing acquisitions.

But if the economic logic behind McLaughlin's conjecture was quite plausible, the empirical evidence that followed did not seem to support it. After publication of McLaughlin's study, two studies examined the possible role of reputation in the acquisition advisory business: a 2000 study by P. Raghavendra Rau, then of Purdue University; and a 2011 study by Jack Bao, then of the Ohio State University, and Alex Edmans, then of the Wharton School (Bao-Edmans hereafter). Rau's study, for example, looked at whether the stock price performance of the acquirer following the announcement of an acquisition was related to the future market share of the acquirer's advisor. The main hypothesis of the study was that if a banker's reputation for creating value is an important consideration for corporate acquirers when choosing their advisors, advisors in acquisitions that have created more value for their previous clients should enjoy higher future market share. Rau's study, however, reported no significant relation between post-acquisition stock price performance and future market share. What he found instead was a significant positive relation between the proportion of acquisitions that were completed and the future market share of advisors. Thus, Rau's findings appear to show no effects of the bankers’ reputation for creating value for clients on the likelihood of the bankers being chosen for future transactions. But the findings provide evidence that a reputation for getting deals done does provide a strong incentive for the banker to complete deals, consistent with the incentives provided by the advisory contracts documented by McLaughlin. And, as Rau interprets his findings, acquirers appear to disregard the stock market performance of the advisors’ previous clients either because value-creation in an acquisition is not an objective of the acquirer, or because the acquirers believe that the past performance of advisors’ clients is not indicative of the performance of future acquirer clients.

And that brings us to the main focus of the Bao-Edmans study, which examines whether the performance of an advisor's clients persists through time. In other words, are the advisors on acquisitions that create value for their clients more likely to be advisors in future acquisitions that also create value for the acquirers? Bao-Edmans find that the client performance of advisors is persistent in the sense that the past performance of deals announced by a given advisor is a predictor of the performance of the future deals on which they are also the advisor. And this in turn implies that an advisor's reputation for helping companies create—or destroy—value through acquisitions provides reliable information that can be used by future clients when choosing their advisors.

Puzzled by these findings, Bao-Edmans then re-examined Rau's analysis, but focused on the short-run (3-day) announcement period stock market reaction of the acquirer rather than the “long-run” acquirer stock returns considered by Rau. And like Rau, they found no statistically significant relation between measures of value created for acquirers and the future market shares of the acquirer's advisors.

In a further effort to explain this puzzle, Bao-Edmans examined whether the value created in acquisitions by an investment bank's prior clients was related to the likelihood that the bank would be hired by the same or any other acquirer to advise on a future acquisition. And, consistent with Rau's findings, Bao-Edmans reported no significant relation. All this evidence appeared to confirm the hypothesis that acquirers, when choosing their advisors, disregard the information—and apparently quite valuable information—contained in prior client performance.

Nevertheless, since the main focus of their study is on the persistence of performance by an advisor's clients, Bao-Edmans present their findings about the effects of client performance on future business of the advisor with caution, commenting that “these results are only suggestive” due to the specifics of the sample used in their analysis.

With this caveat in mind, we undertook our own study of the relation between the performance of an advisor's clients and its success in winning future acquisition mandates. Our investigation began with four hypotheses that we expected to receive support from our statistical tests if the information contained in the stock price performance of an advisor's prior acquirer clients is a factor in a potential acquirer's future decision to hire an advisor. In our tests we measured acquirer performance as the abnormal stock returns around the announcements of the acquisitions. First, if the premise of our tests is correct, we expected that advisors whose prior clients experienced better stock price performance should be more likely to be chosen as advisors in the future. Second, if an acquirer engaged in several sequential acquisitions, that acquirer should be more likely to choose the same advisor the better performance in the prior transaction. Third, an advisor whose clients achieved better performance should experience an increase in its share of the advisory market in the future. (Note, this change in market share is distinct from the level of the advisor's market share, as examined by Rau.) And, fourth, investment banks whose clients experienced better stock price performance should also experience positive stock price responses at the time their clients’ acquisitions are announced.

In conducting our analyses, we used an extensive sample of corporate acquisition attempts that took place over the 28-year period 1984–2011.6 Thus, we included aborted as well as completed transactions. We focused only on those acquisition attempts that had at least the potential to result in a change in control—that is, all attempts in which the acquirers owned less than 50% of the target's stock prior to the attempt and were seeking to own more than 50% after the acquisition. Like Rau, we considered an advisor to the acquirer to be any bank that “acts as dealer manager,” “lead or other underwriter,” “provides financial advice,” “provides a fairness opinion,” “initiates the deal or represents shareholders, board[s] of directors, [or] major holder[s].”7 With our criteria in place, we ended up with a sample of 11,324 acquisition attempts in which acquirers had one or, in some cases, several identifiable financial advisors. By contrast—and this difference will become important later—the sample used by Bao-Edmans in their analysis of advisor choice was, for justifiable reasons, much smaller and included only 1224 acquisition attempts.

We supplemented the data on acquisitions with information from several other data sources. For each acquirer and for each acquirer's financial advisor for which data were available, we collected daily stock returns and market capitalizations from the Center for Research in Security Prices (CRSP) database. We also obtained information about each acquirer's equity and debt issuances and the lead underwriters for each issuance from the SDC's New Issues database. For acquirers that were subsidiaries of a public company, when such data were available, we obtained daily stock returns and market capitalizations of the acquirer's “immediate” or “ultimate” parent. We also obtained information on the analyst coverage of the acquirer from the Institutional Brokers Estimate System (I/B/E/S) to derive a measure of advisors’ security analyst coverage.

In Table 1, we report summary statistics for the sample to provide an overview of the acquisitions that we examined. The acquiring companies were roughly six times the size of the targets in terms of asset book value. The acquisitions involved primarily publicly traded companies, with 86% of the acquirers and 64% of the targets having publicly traded stock at the time of the acquisition attempt. And 89% of the attempts resulted in a completed transaction.

Moreover, in almost 14% of the attempts, the acquirer had used the same advisor in a prior acquisition attempt within 5 years of the current attempt. This statistic, however, is not indicative of the general propensity to rehire the same advisor for a subsequent acqussition, which is around 50%. In most acquisitions in the sample, the acquirer had not attempted a prior acquisition or, if it had, it had not used an advisor. And in 8.9% and 5.4% of the attempts, the acquirer had used the advisor in a prior equity or debt offering.

The main variable in our analyses was a measure of value created for acquirers in prior acquisition attempts. The value created for acquirers in prior deals was estimated by calculating the acquirer's cumulative abnormal return, CAR, over the 5-day interval centered at the announcement date. The CAR was calculated as the cumulative announcement period stock return minus the return on a benchmark portfolio.8

We used two measures of the value created by an investment bank's acquirer clients in prior deals. The first was derived from Rau's study, in which the CAR for each acquirer client was converted to a dollar value by multiplying the CAR by the market capitalization of the acquirer's common equity as of 60 days prior to the announcement. For each advisor, the dollar values so calculated for its clients were summed over the 1 year, that is, 365 calendar days, and 3 years, that is, 1095 calendar days, prior to the announcement of the acquisition in question and divided by the total equity market capitalization of these clients. The second measure, which was used by Bao-Edmans, is an equally weighted average of the CARs of the advisor's clients over the same 1- or 3-year interval. We referred to the first of these measures as the “value-weighted CAR” (VWCAR) of the advisor's prior clients and the second as the “equal-weighted CAR” (EWCAR) of the advisor's prior clients. We referred to these measures collectively as “prior client performance.”

As reported in panel B of Table 1, the average 1- and 3-year prior VWCARs for the acquisition attempts we examined were −0.5% and −0.4%, respectively; and the mean 1-year and 3-year prior EWCARs were 0.1% and 0.2%. The average CAR during all of the 11,324 acquisition attempts in our sample was −0.2%, and the median CAR was −0.7%. These results are consistent with the evidence from the 1970s and 1980s cited above.

Having collected the data and established our measures of prior client performance, we began to investigate the fundamental question of the paper: When choosing their acquisition advisors, does the acquisition performance of the advisors’ prior clients “matter”?

The first question that we empirically studied was whether an acquirer's choice of an advisor is sensitive to the prior client performance of potential advisors. An affirmative answer to this question is consistent with the idea that a reputation for creating value for clients “matters” for acquirers when choosing their acquisition advisor.

To address the question, we examined the relation between an investment bank's prior client performance and the likelihood that the bank was chosen as an advisor in subsequent acquisition attempts. We set up a “choice of advisor” model in which we estimated the likelihood that an investment bank was chosen as an advisor relative to the likelihood that the bank was not chosen as the advisor. The choice model is a logistic regression with 1 indicating that the bank is chosen as the advisor and 0 indicating that the bank is not chosen. The primary explanatory variable of interest is prior client performance (i.e., 1- and 3-year VWCARs and EWCARs).

To implement the model, we identified the bank or banks that were chosen as advisors and the banks that could have been under consideration for advisors but were not chosen. To construct the latter group, we selected investment banks that were likely active in the acquisition advisory market at the time of the acquisition announcement and, therefore, could have been considered as potential advisors. We defined an investment bank as active in the advisory market if it was an advisor in an acquisition attempt in the 1- or 3-year interval (depending on the interval used to construct prior client performance) preceding the acquisition and at least one acquisition attempt after the current acquisition.

We also recognized that other factors may play a role in determining the choice of an advisor and included various “control” variables in the model. Inclusion of these variables was motivated by prior research indicating that such factors can play a role in determining relationships between acquirers and their investment banks.9 Thus, we controlled for prior advisory or underwriter relations between the bank and the acquirer, the bank's record of completing acquisitions in which it was an advisor, the bank's share of the advisory market, the expertise of the bank's stock analysts in the acquirer's industry, and the expertise of the bank in advising acquisitions in the same industry as the target of the current acquisition attempt.

This set-up gives rise to four regressions, one regression for each measure of prior client performance and each measurement interval (i.e., VWCAR and EWCAR each over 1- and 3-year intervals). The results are reported in Table 2. In each regression, prior client performance is positively and significantly associated with the likelihood that the investment bank was chosen as an advisor. That is, investment banks whose prior clients experienced more positive stock price reactions during announcement of their acquisition attempts were more likely to be chosen as advisors in future acquisitions in comparison with other banks whose clients experienced less favorable stock price reaction to their acquisition announcements. This result suggests that when choosing advisors for an acquisition, acquirers are sensitive to the value created for the advisor's prior clients and they are more likely to hire advisors whose clients experienced more favorable stock price reaction.

While the relation between prior client performance and advisor choice is statistically significant, an important question is whether it is economically significant. After all, we also found that other factors—prior relationships between acquirers and investment banks, the bank's analyst coverage, and market share—were all also significant determinants of the choice of an advisor. As a way to assess the economic significance of the effect, we examined the relative change in the outcome (i.e., the likelihood of being chosen as an advisor) that would result from a typical change in the explanatory factor (i.e., prior client performance). When so doing, we estimated that a one-standard-deviation increase in prior client performance led to a 0.13%–0.15% increase in the likelihood that the bank would be chosen as an advisor. This effect should be compared to the bank's unconditional likelihood of being chosen, which is 1.5%. Relative to this unconditional probability, a one-standard-deviation increase in prior client performance improves a bank's likelihood of being hired by 8.7%–10.0%. That increase is not inconsiderable given the potential reward associated with being chosen as an advisor. Perhaps McLaughlin was right in his speculation.

The second question that we studied focuses on acquirers that had hired an advisor for an acquisition attempt and attempted a second acquisition within 5 years of the first. Referring to these acquirers as “serial,” we asked: is a serial acquirer's performance during the announcement of the first acquisition related to the likelihood that the same advisor would be retained for the second acquisition? Essentially, we examined whether acquirers were sensitive to their own performance in prior acquisitions assisted by an advisor when deciding whether to retain that advisor or to choose a different advisor for a subsequent acquisition.

As we noted earlier, the likelihood that an advisor was retained for the subsequent acquisition—assuming an advisor was hired—is almost 50%. Thus, having a prior advisory relationship with an acquirer appears to provide the advisor with a significant edge. So, the question we were really asking was this: did the favorable prior experience with the acquirer improve these odds?

To answer this question, we estimated a model of “advisor retention,” which estimates the likelihood that the same advisor was retained for the second acquisition versus the likelihood that a different advisor was retained. With our sample, we identified 934 pairs of acquisitions by serial acquirers. While the acquirer remains the same, advisors may, and often do, change. And in some cases, the acquirer decides to use no advisor at all for the second acquisition.

In this instance, we estimated a bivariate probit model with 1 indicating that the same advisor was used and 0 indicating that a different advisor was used.10 The key explanatory variable is the first acquisition's CAR.

As with the choice of advisor analysis, we controlled for other factors that may determine the decision of whether to use an advisor. The full set of these control variables is given in Table 3 along with the results of the regression and includes such variables as the fraction of prior acquisitions with the advisor, an indicator variable for whether the advisor provided analyst coverage to the acquirer, the number of years between acquisitions, the number of prior acquisitions by the acquirer, an indicator for whether the target in the subsequent acquisition was a publicly traded company and other factors.

The results, as shown in Table 3, indicate that, after controlling for other potential determinants of the decision to retain an advisor, the likelihood that the same advisor is retained for a second acquisition attempt is positively and statistically significantly related to the acquisition performance of the acquirer in its prior acquisition attempt. As for the economic significance of the relation, a one-standard-deviation increase in the acquirer's CAR during the announcement of the acquisition is associated with an increase in the likelihood of advisor retention by 13.6%. Therefore, when it comes to the decision to retain an advisor for the subsequent acquisition, acquirers are sensitive to what happened to their stock price the last time around—and advisors do have an edge in retaining clients when past performance has been more positive. Again, the evidence is consistent with advisors having a market-based incentive to assist in acquisitions that create value for their clients.

Having established that acquirers appear to be sensitive to prior client performance in making their advisor choices, we revisited the lack of a significant relation between prior client performance and advisors’ market shares that has been reported by both Rau and Bao-Edmans. We noted that because an investment bank's market share is a fairly stationary variable, we should not look for a correlation between prior client performance and the level of market share but should rather examine the change in market share. A smaller regional advisor/bank, even if the few acquisitions that it advised during a year performed exceptionally well, is unlikely to become a top-tier bank over the next year, or even 2 or 3 years. The more likely outcome, assuming a bank's reputation for value-creating deals matters, is that such a bank will experience an increase in advisory appointments coming its way and, perhaps, even a large increase in its relatively smaller market share.

With that in mind, we examined whether client acquisition performance and the change in client acquisition performance during a calendar year was correlated with the future change in the market shares of advisors over the subsequent 1- and 3-year periods. In terms of simple statistics, the average 1-year change in future market shares of advisors whose clients had positive acquisition performance was between 1.1% and 10.4%; and for those whose clients had negative acquisition performance, the average change was between –13.9% and –19.2%. Thus, in simple terms, better client acquisition performance is associated with a greater change in future market share for the advisors.

To control for other factors that may affect the change in future market share, we estimated regression models in which the dependent variable was the future change in market share over 1- and 3-year intervals following the calendar year of the acquisition, and the primary explanatory variable was either client acquisition performance or the change in client acquisition performance. The control variables include factors associated with the advisor such as the fraction of acquisitions completed, the fraction of hostile acquisitions attempted, the fraction of contested acquisitions attempted, and the average fraction of cash used as consideration in acquisitions. We also controlled for the bank's prior share of the advisory market and for the change in the bank's share of the advisory market in the year in which client performance was measured. The results of these regressions, of which there are eight, are presented in Table 4.

As shown in the table, the coefficients of client performance and the change in client performance (i.e., both VWCAR and EWCAR and the change in VWCAR and EWCAR), are positively and statistically significantly related to both the advisors 1- and 3-year changes in future market share of acquisition advisory services. The regressions confirmed that both the level of client acquisition performance, and the change in client performance are positively related to the future change in market shares of advisors.

As in other analyses, we also examined the economic significance of the effect. Reassuringly, the economic impact of the relation between client performance and future change in market share was similar to those in the prior two analyses. A one-standard-deviation increase in the performance of the bank's acquirer clients resulted in an average 8.7%–9.8% increase in the bank's share of the acquisition advisory market over the following 1-year period. Again, the results are consistent with the advisors having an incentive to advise on deals that create value for the acquirer's shareholders.

Having established that client performance predicts whether an investment bank is hired by acquirers for their future acquisitions, we asked a related question: Does the stock market recognize, at the time of the announcement of the acquisition, that an advisor in an acquisition that creates value for the client will land more future advisory appointments, will experience a gain in market share, and will ultimately earn more advisory fees? If it does, the prediction is that the bank's stock price will incorporate that information at the time of the announcement of the acquisition attempt by the bank's client.

To consider that prediction, we examined the correlation between the CAR of the acquirer's stock around the announcement of an acquisition and the CAR of the advisor's stock during the same time interval. We were able to identify 502 observations in which the stock return data of both the acquirer and the acquirer's advisor were available at the time of the announcement of the deal. To examine that relation, we estimated a regression in which the dependent variable is the bank's CAR and the key independent variable is the client's CAR. The regression also included control variables that were related to the bank's announcement period CAR. These variables included the ratio of target size to acquirer size, an indicator variable for whether the parties are in related industries, an indicator for whether the acquisition was perceived as hostile by the target, an indicator for whether cash was offered as consideration, the count of the number of alternative bidders, and an indicator for whether the target was a publicly traded company.

As reported in Table 5, advisor abnormal stock returns are, indeed, positively correlated with acquirer abnormal stock returns during the acquisition announcement period. The stock market appears to recognize that value created for clients does end up benefiting their advisors, and the market incorporates that information into the advisor's stock price. In economic terms, for every dollar that the acquisition creates (destroys) for an acquirer, the advisor's market value increases (declines) by $0.208.

We performed several tests that were designed to examine the sensitivity of our general findings to differences among our sample companies. First, we examined whether the relation holds in certain samples of acquisitions that exhibit differences in other contexts. We partitioned our sample by the target's public/private status11 and by the method of payment for the target.12 Then we examined whether prior client performance was related to the choice of advisor in each of the resulting subsamples. This allowed us to examine whether, for instance, client performance in acquisitions of public targets affects the choice of an advisor for an acquisition of a public target. We performed similar analyses for acquisitions of private targets, for acquisitions in which consideration was exclusively cash, and for acquisitions in which consideration was all stock. The relation between client performance and advisor choice remained significant in these subsamples, with the single exception of the “cash-only” sample. In that particular sample, the number of observations dropped by 90% relative to the full sample, which may be the reason for our inability to detect a significant relation.

Second, most of our advisor choice analyses involved a lead/lag relation between prior client performance and the awarding of advisor mandates. Therefore, we estimated the choice model using a variety of lead/lag relations, ranging from 1- to 5-year intervals. In every instance, the sign of the coefficient of interest was positive and statistically significant. These robustness tests reassured us that the significant relation between client performance and advisory choice persists in various contexts.

In our study of over 11,000 M&A attempts made during the period 1984–2011, we found that the stock market responses to the announcements of acquisition bids are positively associated with both the probability that the advisor investment banks will be chosen by the same or other acquirers for their next deals, and with changes over time in the banks’ shares of the M&A advisory business. What these findings tell us is that, contrary to a common perception, when choosing their advisors, acquiring companies consider the past performance of the advisors’ other corporate clients; and when deciding whether to stick with the advisors they used on earlier deals, they consider the market reactions to those deals. Advisors in deals that receive more positive market reactions are more likely both to be rehired by prior acquirer clients, and to be hired for the first time by “new” acquirers. In this way, market forces provide advisors with incentives to assist their corporate clients in identifying and completing value-increasing acquisitions.

What's more, the reward to advisors for assisting in value-increasing deals comes not only in the form of increases in market share and higher client retention rates, but from increases in their own market values. That is, when acquisitions that increase value for the acquirers are announced, the stock prices of their advisors tend to increase as well (with each $1 increase in the bidder's market value accompanied, on average, by a 20-cent increase in the value of its advisor's stock). This accompanying value increase for the advisor can be explained as not just the value associated with its relationship with its current client, but also from a projected increase in the bank's future market share of the M&A advisory business.

In sum, our results suggest that market forces counteract, at least to some extent, the potential conflict of interest built into acquirer advisory contracts that appear to reward advisors mainly just for closing deals.