Investment Banking has received a lot of attention and negative publicity as of late and is deemed responsible for the recent recession that followed the credit crisis in 2007. Nevertheless the advisory side of the business has been largely irrelevant to causing the crisis. Mergers & Acquisitions ( M&A) advice is vital in the world of corporate finance and creates an efficient market for corporate mergers. The fees charged by advisory practices for the services are sometimes seen as outrageously high and unjustified. Like any other salesmen, investment bankers charge a commission based on the price of the product. Due to the vast sums companies are worth, fees generated are large as well. Whilst the economic value of mergers and acquisitions has been analysed in great depth, advisory fees, and their effect on premiums paid by acquirors for the target firm, has been largely neglected by literature. This study aims to analyse whether the fees are justified by analysing whether more expensive advisors add value for the shareholders or not. We will research how advisors are chosen, and how fees are determined. Then we will continue to analyse how advisors, and the fees they charged impact the premium paid. This paper will focus on the transactions with any German involvement. This will proof valuable as Germany has not been covered in depth and will provide new insights. This study is highly relevant to economic theory and will focus primarily on the principal agent problem.
Concepts & Literature
The number of mergers and acquisitions taking place in the corporate world increased dramatically over the last two decades. Alongside this trend market analysts and academic researchers have been working to identify key factors affecting merger outcomes, and what differentiates a successful acquisition from a failed one. The role of investment banks and the relationship between client and advisor has been of particular interest. How this relationship affects merger outcomes, such as time to completion and premium paid has been subject to thorough investigation. Whilst it is generally agreed that investment banks provide valuable services in making the market for corporate mergers and acquisitions more efficient(Mortensen 1982, Diamond and Maskin 1979), the apparently excessive fees charged by advisors have come under criticism in recent years. It is unclear whether advisors add sufficient value to justify their high fees. This literature review aims to examine some of the previous papers investigating this topic. Merger fees for the purpose of this paper will be defined as fees paid to investment bank advisor. This paper will ignore other fees incurred during the transaction such as legal, consulting, or auditing fees and solely focus on the fees received by the investment bank responsible for facilitating the transaction. It is important to distinguish between target and acquirer fees. With regards to price paid for the target, the acquisition process is a zero sum game. The acquirer would like to pay as little as possible, whilst the target would like to achieve a price as a high as possible in order to maximize value for its shareholders. The target’s gain is the acquirers’ loss and vice versa. Thus fees paid by both parties will have to be looked at separately, in order to determine if higher fees generate value for the shareholders of the respective parties. There are principally two ways in which investment banks create value for their client and the client’s shareholders, the first is through creating an efficient market for transactions and suggesting suitable acquirers/targets for their clients. The second, and the one this paper will focus on, is through negotiating the best possible price for their client. Bowers and Miller find significant evidence that more prestigious and thus more highly paid banks generate higher returns for their clients through selecting suitable targets/acquirers. Other papers generally support these findings (Grossman and Hart 1980, Servaes and Zenner 1986). Whilst there is a vast amount of literature investigating the excess returns generated by bankers through suggesting suitable acquirers/targets for their client, there is very little done in terms of premium paid/received and choice of advisor. The most relevant paper for the purpose of this discussion is Chahine, Ismail (2005). This is because it is one of the very few papers that distinguish between fees paid by acquirer and by target. It focuses on the level of premium paid for the target, and the investment bank’s effort. Fees and advisor reputation are used as proxies for advisor effort. The paper focuses on “investment bank effort”, rather than fees as previous research has found that advisor choice is mainly driven by reputation(Walter, Yawson and Yeung, 2008)(and not necessarily by fee level, although reputable advisors will generally also charge higher fees)(Srinivasan 2001). Within this framework, Chahine and Ismail propose three Hypothesis; The choice of target investment bank reputation is positively related to the choice of the acquirer investment bank The fees paid to advisors is positively related to reputation, and the target advisor fee also depends on the perceived quality of the acquirers advisor The higher the acquirer advisor fee/reputation the lower the premium. The higher the target advisor fee/reputation the higher the premium. This suggests a sequential decision making framework. The acquirer chooses its investment bank first. Then the target chooses its bank, which will depend on how reputable (and thus expensive) the acquirer’s bank is. The more prestigious the acquirer’s advisers, the more of an incentive for the target to hire a reputable advisor. The main issue with trying to model the effect of advisor choice on fee is the vast number of variables driving premium and distinguishing between correlation and causation. Cash-rich acquirers maybe willing to hire highly expensive advisors and pay a very high premium because they have sufficient cash available, this would mean the effect of expensive acquirer advisors on premium may well be underestimated by a regression. Likewise troubled target firms, likely can only afford cheaper advisory firms, and will likely receive lower premiums as they are not performing optimally. This will likely overstate the effect of advisory quality on the target side. Another well researched issue, is the deal completion hypothesis. It states that investment banks may induce their client’s to pay a premium that ensures deal success, rather than optimizes value. This implies target advisors would negotiate a lower premium to ensure deal success, and acquirer advisors would negotiate higher premium. (Chahine& Ismail 2005). This creates a conflict of interest between acquirer and investment bank, as the latter may only be motivated by fee. Fee however mainly depends on deal success, thus the advisors may focus more on getting the deal done than on maximizing value for clients. This is negated to a certain extent tough since future business for an advisor is mostly dependent on its reputation (McLoughlin 1990), and thus advisors have an incentive to act in the best interests of their clients to ensure a good reputation and future deal flow.
The method used by Chahine and Ismail(2005) will serve as base but will need to be altered significantly. See below the three equations used in their paper, these equations will serve as a base for our own regressions. a) Acquirer Advisor Reputation = f(deal value, acquirer size, , deal attitude, geographic scope, method of payment, industry scope) b) Target Advisor Reputation = f(acquirer advisor reputation, deal value, acquirer size, deal attitude, geographic scope, method of payment, industry scope) a) Target Fee = (Acquirer Reputation, Target Reputation, deal value, deal attitude, geographic scope, method of payment, industry scope.) b) Acquirer Fee =(Acquirer Reputation, deal value, deal attitude, geographic scope, method of payment, industry scope.) 3. Premium = f (acquirer fee, target fee, acquirer-IB reputation, target-IB reputation, deal value, acquirer size, deal attitude, geographic scope, method of payment, industry scope) These equations will attempt to model the sequential decision framework. At first the acquirer decides it would like to buy a company, and then chooses its advisor. This choice will depend on the various factors listed above, we would expect it to be mainly driven by acquirer size. Because we define reputation in terms of league table placement, the advisors at the top of the league table must advise on many large transactions to reach this league table position. Hence they are likely the ones with most experience, and best connections to large clients. Hence these large clients are much more likely to hire a reputable bank. After the acquirer chooses its advisor, they together find a suitable target. When they approach their target, the target will hire its own advisor, dependent upon similar factors to the acquirer. However the target knows which advisor the acquirer picked and hence will likely try to pick an equally reputable one. Fees are determined simultaneously to the hiring of the advisor and are likely contractually laid out when the advisor is hired. The above variables are intended to gauge complexity of the deal which would require higher fees. The fees of the respective side will also dependent on the reputation of the advisor, with more reputable advisors charging higher fees. We would expect the reputation of the acquirer to affect the target fee levels, as this reputation is known to the target advisor when the contract is set up. Lastly then premium is determined by various factors as laid out above, the effect of the relevant advisor and fee variables have been discussed in the theories above. The key issue in this equation is RHS endogeneity and lack of sufficient exogenous instruments. The reputation of the acquirer’s advisor is endogenous as more complex, larger deals generally require a more reputable acquirer. However all of those factors may influence the premium as well. There is a lack of exogenous variables to use as instrument for this. We may attempt to use acquirer size as instrument, since in our own regression we found this to be least correlated to the premium. Whilst some previous literature(Rau, 2000) finds that larger acquirers may pay larger premiums, it is as of now the best instrument available. Target advisor reputation faces similar issues, they are compounded by the fact however that it also depends on the acquirers advisor reputation as we have laid out in the sequential decision making framework that the targets advisor choice is dependent on the acquirer. There are unfortunately no exogenous instruments available. Overall Chahine and Ismail serve as a template but their 2SLS methodology is flawed. They estimate reputation and fee separately, then use those estimates in a 2SLS regression to determine premium in the equation laid out above. However this is unnecessary as 2SLS does this itself already, hence their standard errors and significance levels are likely wrong. This paper will attempt to correct this flaw. Overall estimation will nevertheless be difficult due to severe RHS endogeneity, multicollinearity and lack of exogenous instruments.
ThomsonOne Banker Database Company reports
Freeman algorithm estimates We primarily used ThomsonOne Banker database to gather all relevant transactions details. Only acquisitions larger than $1m and only non financial institutions (as valuation is vastly different from other industries) will be used. There are several issues that need to be addressed in data collection. For one there is heavy survivorship bias as only transactions that happened are recorded in the database and all failed negotiations are not accounted for. Further there is significant attrition as of an original 10,000 recorded transactions only 130 meet all sufficient criteria, thus we are only dealing with a very limited subsample of the entire transaction space.
Avg. Deal Size
Avg. Acquirer Size
% Cross Industry
% Reputable Target Advisor
% Reputable Acquiror Advisor
% Target fee of total
% Acquirer fee of total
# of Transactions
Our premiumspread is wide, and this sample has been truncated significantly to exclude any distressed situations (high negative premium), and outliers at the top end. All deals are control changing. Average transaction and acquirer size are large($4.1bn, $22.7bn) this is expected as we only included publicly listed companies, which tend to be large, hence both acquirers and targets will be large. Both acquirer and target fees(as % of total deal value) are similar. Most deals are cross industry or cross border. Most transactions are advised by reputable advisors, suggesting the market for these large cap transactions is split between few firms. We defined advisor reputation by taking the league tables 5 years prior to deal announcement. The top 10 firms in those league tables are declared reputable(assigned a 1 in the dummy), the rest is assigned a zero. Target advisors tend to be less reputable than acquirer advisors. There are minor inconsistencies in share price data. We tried to use 4week previous share prices in determining the premium level, however for some transactions(<5) there are only one week or one day prior data available that was used. We manually looked at their share price and found the available one week/one day data to be in line with the 4 week prior data. Further exchange rate fluctuations may alter valuations and sizes, as all data on Thomson one is in USD. Thomson one does not supply acquirer size data. It was put together using company financial reports were available. Revenue was used, rather than equity value, to allow for easier comparison between public and private companies, and to avoid having to make difficult valuation judgements. Where data at year of purchase was not available it was estimated using prior years. No original fee data is available, Thomson One relies on freeman algorithm estimates. This maybe an issue if the algorithm is flawed. The algorithm is generally accepted in industry, and there is no original data available for German transactions, hence we are forced to rely on these estimates.
probit acqadvrep lnacqsize border xindustry, vce(robust)
.304 .0683 4.45 0.000
-.568 .348 -1.63 0.103
.0316 .269 0.12 0.906
-1.642 .588 -2.79 0.005
.102 .0223 4.55 0.000
-.172 .0916 -1.88 0.061
.0106 .0905 0.12 0.907 We use a probit model to analyse determinants of acquirer advisor reputation. This is sensible as acquirer advisor reputation is binary. We found this specification to be adequate using various tests. We find only the natural log of the acquirer size and cross border to be significant. As expected the key determinant of the acquirer advisor reputation is the acquirer’s size, as discussed before the top of the league table advisors are at the top because they advise on a lot of large transactions, and hence have a strong reputation amongst large acquiring firms. The negative coefficient on border is counterintuitive as we would expect cross border transactions to be more complex and thus warrant more reputable advisors. One possible explanation is that cross border deals involve countries with significant differences in league table placement of advisors, and thus this leads to a strong negative bias on the border coefficient.
probit tgtadvrep acqadvrep lndealsize, vce(robust)
.708 .311 2.28 0.023
.520 .116 4.50 0.000
-.119 .315 -0.38 0.705
-.029 .283 -0.10 0.918
-3.45 .750 -4.61 0.000
.275 .118 2.33 0.020
.200 .042 4.73 0.000
-.046 .119 -0.38 0.703
-.011 .109 -0.10 0.918 Moving onto hypothesis 1 b), the target advisor reputation, we find a significant positive impact of acquirer advisor reputation on target advisor reputation. The natural log of deal size is also positive and significant. This confirms our above hypothesis of a sequential decision making framework. The log of deal size is highly important for the same reasons as above, top league table advisors will have most large transaction experience. What is interesting to see with both target and acquirer reputation is that deal/acquirer size appears to be the main drivers, and the variables measuring complexity of transaction (cross industry, cross border) do not have a significant effect (except for border for acquirer reputation, a counter intuitive and likely biased effect).
ivregress 2sls tgtfee lndealsize border xindustry (acqadvrep = lnacqsize)
-.166 .335 -0.50 0.621
-.231 .035 -6.54 0.000
-.076 .103 -0.74 0.460
-.152 .084 -1.80 0.072
.250 .131 1.91 0.057
2.471 .198 12.49 0.000 We use a 2sls to account for the endogeneity of acquirer advisor reputation. Unfortunately we cannot account for the endogeneity of target advisor reputation as we lack exogenous instruments. We see that the reputation of the acquirer advisor is insignificant, the log of size of the deal is highly significant and negative, the reputation of the target advisor is positive and significant. As tgtfee here is defined as fee paid to target advisor as a %, it is expected for this % to decrease as the size of the deal rises. This is because on very large deals advisors tend to charge lower fees as % of total deal, as the fees in absolute terms are still more than sufficient to compensate the advisor. We find that more reputable advisors charge higher fees as %, this was also to be expected. Contrary to our hypothesis, the reputation of the acquirer advisor has no effect on the target advisor fees. This may be due to the endogeneity issues with target advisor reputation.
ivregress 2sls acqfee lndealsize border xindustry (acqadvrep = lnacqsize)
.483 .254 1.90
-.296 .0314 -9.40
-.099 .0795 -1.25
-.103 .0640 -1.61
2.571 .139 18.49 Similar regression as above, we find only acquirer advisor reputation, deal size, and cross industry to be significant. Coefficient on acqadvrep is positive and significant as expected, more reputable advisors charge higher fees. Fees decrease as a % in dealsize, which was also to be expected. The coefficient on cross industry is interesting, as it is counter intuitive. One would expect advisors to charge higher fees on cross industry transactions, as they are more complex. One explanation maybe that cross industries deals are primarily undertaken by financial buyers, who are less willing to pay high fees. Alternatively strategic buyers gain less synergies in cross industrie deals and thus maybe less inclined to pay high fees.
reg lnpremium border xindustry acqfee acqadvrep tgtadvrep ownedaftertransaction tgtfee pharma TMT natres, vce(robust)
.031 .214 0.15 0.884
.114 .167 0.68 0.496
.059 .211 0.28 0.782
-.285 .207 -1.37 0.172
.143 .231 0.62 0.539
.024 .008 3.17 0.002
-.457 .272 -1.68 0.096
.555 .193 2.88 0.005
.664 .280 2.37 0.020
.974 .535 1.82 0.072
1.151 .679 1.70 0.093 We use an OLS model as there is too much endogeneity to account for, hence we chose to ignore it, and discuss limitations of our results later on. We also find various specification issues which we could not address, including omitted variables. These are likely due to the endogeneity issues. See appendix for relevant tests. We find the % of shares owned after the transaction highly significant, and positive. This is to be expected as buyers have to pay a significant control premium. Whilst our sample only includes transactions in which there was a change in control (shares owned before must be less than 50%, shares owned after transaction must be more than 50%), there is an additional premium for owning larger stakes in firms, especially for owning up to 100%. This is because minority shareholders still hold significant rights and need to be compensated highly to sell their shares and forfeit those rights. This is well documented in the existing literature ( Dyck and Zales, 2004). The primary argument is that large blocks of shares grant its owners private benefits, not available to diluted shareholders. These mainly arise through improved control over management, amongst other factors. The positive coefficients on pharma, natural resources(natres) and TMT is in line with theory. Those three industries traditionally command higher premiums than the remaining industries. In the case of TMT this is primarily driven by technology companies which historically achieve very high premia due to high future upside. The same applies to pharmaceutical companies. This is mainly because most pharmaceutical transactions involve larger companies acquiring smaller firms with a promising pipeline. Large firms are able to pay high premiums over market value because the pipeline, when realized, can lead to huge profits for them as they possess a vast distribution network, which the small firm lacks. Relevant to our core theory we unfortunately only find target fee to be relevant, and to have a negative effect. This is not necessarily counter intuitive, as it is in line with the deal completion hypothesis. Because league tables are driven by closed deals, those on top of the tables(and thus those charging higher fees), got to the top through closing a lot of transactions. A transaction is more likely to close the lower a premium the target advisor demands. Also fees are only paid in case of deal closure, so advisors have a short term incentive to close a deal to reap fees, although in the long run this may damage reputation(however only if reputation is not primarily determined by league tables, as league tables are driven by closing deals). An issue we face is potential Endogeneity. Our theoretical framework established a highly endogenous relationship between all independent variables (i.e. tgtfee depends on acquirer reputation and target reputation which all drive premium). We found some relationship between acquirer advisor reputation, target advisor reputation and the fees charged by them. Our issue is we cannot address it in a 2SLS framework as we lack exogenous instruments. This means our OLS estimates maybe biased and inconsistent, making interpretation questionable. In the Ramsey reset test we also find there to be omitted variables, however even upon inclusion of all available variables in various combinations we have found no adequate specification, which again is likely caused by the endogeneity.
Limitations, Extensions & Conclusion
The validity of this study and any study trying to find the impact of advisor fees/reputation on shareholder value is questionable. There are too many reasons as to why such investigations are riddled with flaws, we will briefly try to touch on the few key points. The first issue is the problem of deciding what a fair price is, and what the “correct premium” should be. Determining premium levels involves valuation. Valuation is an inherently subjective matter. Thousands of finance professionals and academics obsess over this daily, analysis are often as quantitative as they are qualitative. Valuation is not an exact science and there is always significant room for debate. This will lead to random variations in premiums paid amongst similar companies. This is because the market for companies is highly illiquid and thus accurate pricing is not achieved. Even if there was a generally agreed upon formula to value businesses, unobserved heterogeneity would still be a major issue. There are an infinite number of qualitative factors affecting the valuation of any one business. Whilst it is possible to include the most obvious ones (i.e. severe expected litigation charges, recent damage to the brand, distressed situation), to conduct in-depth valuation of each transaction this is beyond the scope of this paper(investment banking professionals may spent 40 hours or more on each firm valuation) and as discussed above would not necessarily yield great results. Even if we could take them all into account, the actual price paid for a corporation is still subject to massive variance because different buyers can afford different prices, which will lead to bias in the coefficients for advisor effect. This is an issue because of the illiquid nature of the market. We attempt to account for financial buyers paying lower prices(although find the coefficient to be insignificant), although even two strategic buyers may be able to pay vastly different premiums for similar businesses. This leads to even more heterogeneity and random variation of premium. Moving onto our last issue, even if we had perfectly homogenous companies and a way to value them exactly, we still have one major enemy in our undertaking, the date of transaction. Timing of transactions influences valuations and premiums in a vast number of ways, many unobservable, we will attempt to briefly discuss the most common ones. Stock market valuation differ massively throughout time, and often do not reflect fundamentals. A premium of 100% maybe perfectly adequate for one pharmaceutical company if its price is depressed due to market irrationality, whilst a premium for the same pharmaceutical company of 10% maybe outrageously high even just one year later. Next to the stock market aspect we also face a credit aspect. Depending on availability of credit and interest rates financial buyers may be able to pay significantly higher premiums(as seen in 2007). Another issue is not only the overall cycle, but also the individual industry cycle affects premium. Lastly there are not only problems associated with the market pricing of firms, global macro conditions also affect the fundamentals of companies. When you add the interplay of these two factors you could realize that comparing a transaction today to a transaction five years ago is incorrect. We have tried adding a time variable to account for this, but found them to be insignificant.
As you see from the limitations mentioned above, the key extension is to massively increase sample size. As we have exhausted all German transactions, further research may be done by adding more European countries to the transaction list. This would allow it to deal with the various issues of unobserved heterogeneity better. Although we would face the additional problem of differences between countries, we should however be able to account for those easily through use of appropriate dummy variables. With vast amounts of additional time and resources analysing every transaction in depth, attempting to model in all the qualitative factors could lead to better results as well.
This paper examines the choice of advisor, the fees charged by advisors, and the determinants of premium in M&A transactions. We proposed the same three hypothesis as Chahine & Ismail (2005). The acquirer first chooses its advisor based on deal complexity, then the target plays a best response to the acquirers choice, and lastly the premium is a function of advisor reputation and some deal specific variables. We also followed the methodology of Chahine & Ismail (2005), albeit we did not address the endogeneity in their 2SLS way, as we did not find any adequate exogenous instruments, did not discover as much endogeneity as they did, and as their methodology was flawed in this regard. We found evidence for the advisor choice framework we proposed. The acquirer first chooses its advisor, then approaches a target, and then the target chooses its advisor. We also found support for our hypothesis that more reputable advisors charge higher fees, although we were unable to find support for the hypothesis that more reputable acquirer advisors cause the target advisor to charge higher fees. Lastly we found support for the deal completion hypothesis, that is more expensive(and hence more reputable) target advisors lead to lower premiums. This is contrarian to our original hypothesis. The validity of the final regression is questionable due to endogeneity issues, unobserved heterogeneity and our limited sample size. Lastly if we were to further investigate we would mainly aim to get a much larger sample to address the unobserved heterogeneity issue, but we still lack exogenous variables to address the endogeneity.
Bowers H, Miller R. Choice of Investment Banker and Shareholders’ Wealth of Firms Involved in Acquisitions.A Financial ManagementA [serial online]. Winter 1990;19(4):34-44. Carter, R. and Manaster, S., 1990. Initial public offerings and underwriter reputation. Journal of Finance 45, pp. 1045-1067 Chahine S, Ismail A. Premium, Merger Fees and the Choice of Investment Banks: A Simultaneous Analysis.A Quarterly Review of Economics and FinanceA [serial online]. May 2009;49(2):159-177. Diamond, P. and E. Maskin. 1979. An equilibrium analysis of search and breach of contract, I: Steady, States. Bell Journal of Economics , no. Spring: 282-311. Gaughan, P., 1996. Mergers, Acquisitions, and Corporate Restructurings. , Wiley, New York. Grossman, S. and O. Hart. 1980. Takeover bids, the Free-Rider problem and the theory of the corporation. Bell Journal of Economics , no. Spring: 42-64. Johnson, J. and Miller, R., 1988. Investment bank prestige and the underpricing of initial public offerings. Financial Management 17, pp. 19-29 Loughran, T., Vijh, A., 1997. Do long-term shareholders benefit from corporate acquisitions? Journal of Finance 52, 1765-1790. McLaughlin, R., 1990. Investment-banking contracts in tender offers: an empirical analysis. Journal of Financial Economics 28, pp. 209-232 McLaughlin, R., 1992. Does the form of compensation matter? Investment banker fee contracts in tender offers. Journal of Financial Economics 32, 223-260 Michel, A., Shaked, I. and Lee, Y.-T., 1991. An evaluation of investment banker acquisition advice: the shareholders’ perspective. Financial Management 20, pp. 40-49 Mortensen, D. 1982. Matching as a non-cooperative bargaining game. in The Economics of Information and Uncertainty, Edited by John J. McCall. Chicago: University of Chicago Press . Nanda, V. and Yun, Y., 1997. Reputation and financial intermediation: an empirical investigation of the impact of IPO mispricing on underwriter market value Rau, P. (2000)A A A Investment bank market share, contingent fee payments, and the performance of acquiring firms. Journal of Financial Economics, 56,293-324A Rau, P. and Vermaelen, T., 1998. Glamour, value and the post-acquisition performance of acquiring firms. Journal of Financial Economics 49, pp. 223-253. Saunders, A., Srinivasan, A. InvestmentA BankingA RelationshipsA andA MergerA Fees (October 2001). NYU Working Paper Servaes, H. and M. Zenner. 1996. The role of investment banks in acquisition. The Review of Financial Studies 9, no. Fall: 787-815. Sirri, E. and Tufano, P., 1998. Costly search and mutual fund flows. Journal of Finance 53, pp. 1589-1622 Travlos, N., 1987. Corporate takeover bids, methods of payment and bidding firms’ stock returns. Journal of Finance 42, pp. 943-963 Walter T, Yawson A, Yeung C. The Role of Investment Banks in M&A Transactions: Fees and Services.Pacific-Basin Finance JournalA [serial online]. September 2008;16(4):341-369.A Zingales, L. and Dyck, A. and (2004), Private Benefits of Control: An International Comparison. The Journal of Finance, 59:A 537-600.
ThomsonOne Banker Database Company reports