Competition in the market for takeover advisers.

Abstract:

We investigate factors that motivate bidders to engage advisers, we model adviser selection and we test whether value-adding advisers gain market share. Our sample includes 801 attempted takeovers over 1989-1998 in Australia. Our results indicate advisers are likely to be engaged if the takeover deal is large, hostile, and includes non-cash compensation. We find deal completion is not as closely correlated with adviser rankings as does Rau (2000) but we confirm his finding that adviser ranked high on market value of deals advised do not have a comparative advantage in adding value to firms. However, we document some (limited) evidence that value-adding advisers achieve an increase in subsequent deal flow. Our results are consistent with specialization among takeover advisers.

Keywords:

TAKEOVER ADVISERS, MERGERS AND ACQUISITIONS, ADVISER RANKING.

This paper has benefited from comments of David Emanuel (the discussant) and participants at the Third Annual Summer Research School hosted by The Accounting Foundation at the University of Sydney and the School of Accounting at the University of Technology, Sydney in February 2000. The authors also wish to thank Tig Ihnatko for econometric advice and Emily Rosier for help with data collection.

1. Introduction

Australia has a thriving, substantial market for takeover advisers. To illustrate, in 2003 over 1,800 transactions involving advisers were announced, with a total market value of $US69.9 billion. (1) We address three questions. What attributes of a takeover are associated with hiring of advisers? How do firms choose among advisers? Is there an association between adviser ranking on the widely cited M&A Adviser League Tables and the economic rent earned by their clients?

Our study is motivated, in part, by Rau's (2000) intriguing US-based finding that adviser market share is more closely associated with deal completion (i.e. acquisition of target) than wealth creation for the acquirer, a finding prima facie at odds with wealth-maximizing behaviour among market participants. We test the robustness of Rau's results in Australia, after adapting his research design to eliminate a potential weakness, namely, temporal stationarity in adviser rankings. One reason for testing the robustness of Rau's results is that the Australian share market is much smaller in than its US counterpart and the relative contribution of advisers to the wealth created or lost in a deal is arguably easier to observe among smaller firms. Two comparisons indicate the substantial difference in size in the two markets: As at December 2002, the total market capitalization of all companies listed the Australian stock exchange was equal to just 5.5% of the total market capitalization of all New York Stock Exchange (NYSE) listed companies and 75% of Australian listed companies had a lower market capitalization than the mediansized firm in the NYSE decile comprising smallest firms. (2) Rau (2000) aside, there is no legal requirement to use an adviser in takeover negotiations and so modelling the attributes of Australian takeovers connected with the hiring and choice of advisers and assessing the outcomes provides findings of interest to, inter alia, managers, shareholders and regulators, particularly given that takeovers trigger substantial shareholder-wealth changes.

We find that advisers feature more frequently in large, hostile deals, if a firm is listed, the bid is for a diversified firm in the same primary industry as the acquirer, and the bidder uses forms of payment other than cash. In sum, advisers are more likely to be hired in complex deals where advisers can reduce transaction costs, a finding consistent with Servaes and Zenner's (1996) study of US investment-banks' roles in takeovers. Our findings on the attributes of Australian advisers are less consistent with prior US evidence. Unlike Rau (2000), first-tier advisers (i.e. the handful of firms consistently ranked highly terms of share of value of deals advised) in Australia are not unequivocally better at completing deals than other advisers. However, consistent with Rau's findings, first-tier advisers are also not associated with higher market returns to their clients, lower-tier advisers hold their own and more in this regard. These findings are reinforced when we redo our analysis on the basis of individual rankings rather than the three-tier adviser classification scheme adopted by Rau. We conclude that adviser league tables constructed on the basis of market share are unreliable guides to the roles played by advisers in Australia. It is likely that advisers specialize so that even those with lower market share might be better than their higher-ranked counterparts at delivering valued outcomes (e.g., lower transaction costs and deal completion) in some cases. We find evidence consistent with this scenario in that lower-ranked firms are preferred in complex deals. In our final analysis we revise adviser rankings by basing them on value added to firms to investigate whether advisers who add value are rewarded with higher deal flow in subsequent years. Our results are consistent with this conjecture, albeit based on a small sample size due to data constraints.

The paper is structured as follows. Section 2 briefly describes theory and evidence on attributes of takeovers that prompt firms to hire advisers, followed by a longer, related discussion on the factors influencing the choice of adviser. The section closes with a discussion of the relevant considerations when using M&A Adviser League Tables as a measure of quality. Sections 3, 4 and 5 comprise, respectively, the description of data and research method, review of results and summary with concluding comments.

2. Takeover Advisers: Context, Theory and Prior Findings

Takeover advisers are intermediaries whose fees are commonly contingent on deal completion (McLaughlin 1990; Rau 2000). (3) Takeover advisers ostensibly identify takeover targets, value them, and construct a bid at lower cost than their clients (Benston & Smith 1976). These features of the market for advisers have prompted research into identifying the features of takeovers that promote the hiring of advisers and assessing whether advisers do add value.

2.1 Attributes of Takeovers Associated with Hire of Advisers

Servaes and Zenner (1996) contend that the benefits to acquirers from economies of specialisation, scale economies and reduction in search costs are greater when the value of transaction is large, the target operates in a different industry, the target is itself diversified, the bidder is inexperienced and the offer is unwelcomed (i.e. hostile). These factors increase the complexity of a takeover deal and advisers may lower the costs of managing the deal. Servaes and Zenner report results consistent with their conjecture. As the following discussion shows, lowering deal transaction costs may be a second order consideration in the market for takeover advisers, notwithstanding Servaes and Zenner's findings.

2.2 Choice of Adviser: Wealth Creation or Ability to Complete Deals?

Given that advisers are hired to reduce takeover deal costs it is curious that adviser fees are routinely contingent on deal completion. Fees contingent on deal completion promote an obvious moral hazard problem; advisers are likely to advocate increasing the offer price beyond the level where the acquisition is a positive net present value investment for the acquirer. A potential constraint on advisers gouging their clients is the risk of lowering their reputation and consequently market share by advising on deals that lose value. Thus, as Rau (2000) observes, wealth-seeking advisers have two potentially conflicting objectives. One is to complete all deals (i.e. the deal completion hypothesis) and the other is to create value for their clients (i.e. the superior deal hypothesis). The conflict is not invariable because deal completion may be consistent with value creation. Indeed, in a well functioning market we expect completed deals to be, on average, deals which create value for acquirers and targets. The question is how, if at all, is the conflict resolved for advisers in cases where deal completion conflicts with value creation?

Somewhat anomalously, given our priors that managers' interests are typically aligned with shareholders in competitive markets, Rau (2000) finds advisers' market share is more closely associated with their record in completing deals than in value creation. However, Rau's (2000) finding of insignificant difference in wealth generated to acquirers regardless of quality of adviser might be partly driven by the method he uses to rank advisers. The advisers are ranked based on their deal values and number of transactions announced. The classification is based on data over the full period surveyed and so adviser quality is effectively fixed over the period studied. Rau's results may mask the dynamics of the takeover adviser market. Our innovation to allow adviser quality ranking to more closely reflect its variation over time is described in the research method section.

2.3 Relevance of 'M&A Adviser Ranking League Table'

The proposition that firms select advisers on the basis of quality rests on the premise that quality is known or estimated. Chemmanur and Fulghieri (1994) argue that investors use financial intermediaries' past performance to assess credibility because the amount and quality of effort exerted by intermediaries for clients cannot be fully observed. A widely cited source of information on advisers' past performance is Thomson Financial Services Ltd via its M&A database, SDC Platinum. The credibility of the data derives from its being an ostensibly independent, verifiable source of information about adviser quality. Prominent media reference to M&A adviser rankings based on SDC data, (4) widespread use of SDC data in academic research, (5) and the common use of SDC information by advisers to promote their expertise in M&A work all attest that Thomson Financial Services' SDC database on M&As is a key element in the market for takeover advisers. (6)

Given the above, the use of SDC ratings to assess the association between adviser quality and value added to clients may appear straightforwardly justifiable. However, SDC ratings based on value of deals performed by advisers are not necessarily indicative of the quality of advice provided on a takeover. One reason is that M&A advisors are not always hired principally for their advice on the merits of a takeover. Access to the capital necessary to finance a deal, connecting people and negotiating a deal through regulatory guidelines are other services provided by M&A advisers, particularly those affiliated with banks (Allen, Jagtiani & Peristiani 2001). Servaes and Zenner's (1996) finding that higher ranked advisory firms in terms of value of transactions advised--who mostly also happen to be affiliated with large banks--are not the first port of call for acquiring firms undertaking complex deals might reflect recognition that value of deals conducted in the past is not an accurate indicator of the quality of pure advisory work performed. If takeover advisers are hired principally to facilitate deals via access to financing and guidance through the regulatory maze then it is not surprising that advisers' market share will not be related to the economic quality of deals they advise. The emphasis on deal completion is also explained. Nevertheless, the lack of association between adviser ranking on the conventional basis of value of deals transacted and value created for acquiring firms' shareholders remains to be proved in Australia and the third question investigated in this paper attends to this.

It is worth noting that even if value of deals advised is in fact a valid gauge of quality of advice provided, the method used to collect information on deals in which advisers have been involved arguably reduces the accuracy of the metric as an indicator of quality. SDC takeover adviser data are compiled from publicly available information. Annual rankings are based deals worked on by each adviser during the previous calendar year. Deals must have the following characteristics to be counted in a given year:

* The deal must be completed or unconditional during the calendar year;

* The target must be Australian; and

* The deal must have had at least one adviser involved.

Each adviser in a deal is given credit for the deal regardless of its role, or the size of its contribution. Work for bidders or targets count equally but withdrawn takeover offers are ignored. Total market is the sum of all takeovers where at least one adviser has been involved. No account is taken of deal complexity or level of risk assumed by the adviser. The value of all takeovers an adviser has worked on is divided by the total value of the market. Advisers are then ranked on market share, typically.

Given that credible, standard, independently verifiable measures of deal complexity and risk are unavailable, it is understandable that they are not taken into account in most publicly available adviser ranking schemes. However, in-so-far that deal complexity and risk are important considerations when assessing adviser quality, any ranking scheme that ignores them most likely is substantially reduced in accuracy. The reduction in accuracy is unknown. We return to this issue when analysing our results.

3. Research Design and Data

We construct three models to investigate each of our three questions: What attributes of a takeover are associated with hiring of advisers? How do firms choose among advisers? In addition, is there an association between adviser ranking on the widely cited M&A Adviser League Table and the economic rent earned by their clients? The adviser use model and the adviser choice model are constructed from deal-by-deal data. The market share model uses annualised data.

3.1 Data

Our sample offers, sourced from SDC Platinum, were made between 1 January 1989 and 31 December 1998, and include only attempted takeovers where the bidder is Australian. A total of 4,172 takeovers are identified. We remove all rumoured takeovers, leaving 2,610 takeovers. Imposing a minimum transaction size of $500,000 reduces the sample to 1326 takeovers. We then cull all offer where the bidder did not intend to hold at least 20% of shares in the target after the takeover, to ensure our sample includes only offers potentially affecting a material transfer of control deals. (7) This reduces the sample to 865 takeover offers,

Our final restriction is to exclude government firms, joint ventures and several firms whose status is unknown. Government firms and joint ventures are omitted because their price data are unavailable and because their management does not have to respond to the same commercial incentives as non-governmental firms. The final sample contains 801 attempted takeovers. The sample of 801 takeovers includes 244 (30%) Asset Sales of which are 51 (21%) where the Acquirer used an advisor (compared with 50% in the non-Asset sample), and 34 (14%) targets who use advisors (compared with 38% in the non-Asset sample). Pertinently, the league tables compiled by Thomson Financials includes Assets Sales when the M&A Adviser rankings are determined. (8)

ASX Journals are used to identify the number of shares outstanding one month prior to the takeover. They are also used to identify listed firms, and the total assets of each listed firm. (9) Day end share prices for listed firms, adjusted for dividends and all changes in basis of quotation on, are sourced from SIRCA's Core Research Data (CRD). The All Ordinaries Accumulation Index (AOAI) daily values are sourced from FT Extel. The index represents 93% of value and 94% of turnover on ASX (ASX Fact Book, 1998). Returns to the AOAI approximate a buy-and-hold strategy with 100% dividend reinvestment.

3.2 Adviser Use Model

The following logit model tests the factors that influence the decision to use an adviser.

(1) Adviser Used = [[alpha].sub.0] + [[beta].sub.1]Attitude + [[beta].sub.2]Cash + [[beta].sub.3]Log(Transaction Size) + [[beta].sub.4]Log(Previous Acquisitions) + [[beta].sub.5]Related Industry + [[beta].sub.6]Log(# Target SICs) + [[beta].sub.7]Market Reaction + [[beta].sub.8]Bidder Listed + [epsilon]

where: The rationale for including each factor is explained in the discussion to preserve concision:

Adviser Used = a dummy variable equal to one if the bidder uses an adviser, and zero otherwise;

Attitude = a dummy variable equal to one if the acquisition is hostile, and zero otherwise;

Cash = a dummy variable equal to one if cash is the sole means of payment, and zero otherwise:

Log(Transaction Size) = where Log is the natural log, and transaction size is the price paid for the target net of debt;

Log(Previous Acquisitions) = the number of previous acquisitions performed by a given company, as at the date of announcement, and is a proxy for experience. (10) The range of values for this variable is zero to 16. The log is taken, (11) as the incremental experience gained by performing one extra takeover is not expected to be linear:

Related Industry = a dummy variable equal to one if the 3 digit primary SIC code is the same for both the target and the bidder, and zero otherwise. We follow Servaes and Zenner (1996) in using this variable as a proxy for information asymmetry;

Log(# Target SICs)= a complexity measure. The greater the number of industries the target operates in, the more work is needed to accurately assess the impact of the takeover;

Market Reaction = a dummy equal to one if the short run abnormal return is non-negative, and zero if the return is negative. Market Reaction is defined as the abnormal bidder return over the period [-5, +5 days], where day [0] is the day the deal is announced. An issue with this variable is that the market reaction is an ex post measure used to proxy for an ex ante decision input. (12) We assume that positive deals are more likely to be known by management and so market reaction is a proxy for managements' knowledge of the NPV of the takeover at time they hire an adviser; and

Bidder Listed = a dummy variable equal to one if the bidder is listed on the ASX, and zero otherwise. This dummy is included to account for the reputation at stake in success for the bidder.

3.3 Adviser Choice Model

Adviser choice is modelled using multi-logit regression. The model is based on Servaes and Zenner (1996), and includes variables from the Adviser Use Model. The sample comprises only takeovers where the bidder used an adviser, that is, 247 takeovers.

(2) Tier = [alpha] + [[beta].sub.1]Attitude + [[beta].sub.2]Cash Dummy + [[beta].sub.3]Log(Deal Size) + [[beta].sub.4]Log(Previous Acq.) + [[beta].sub.5]Log(#Target SIC Codes) + [[beta].sub.6]Market Reaction + [[beta].sub.7]Relative Size + [[beta].sub.8]Sign. Bidder + [epsilon]

where: Tier variable is equal to three if the adviser involved is third tier; two if the adviser is second tier; and one if the adviser is first tier. First tier advisers are the most prestigious. A negative coefficient on the independent variable signifies an increased probability of using a higher ranked adviser. Where there are multiple advisers, the adviser of the highest rank is chosen to represent the minimum prestige level of all advisers, consistent with Rau (2000) and Servaes and Zenner (1996). The procedures followed to establish adviser rank are described in section 3.4.

The Relative Size difference between the bidder in deal i using adviser j and the target in deal i can have an impact on the choice of adviser.

Sign. Bidder is a dummy variable equal to one if the bidder, prior to the takeover, owned five percent or more of the target's outstanding shares, and zero otherwise.

All other variables are as defined in the first model.

3.4 Ranking the Advisers

Two methods are used to rank advisers, the 'look-back' method and the 'cumulative ex-ante' method. Rau (2000) groups advisers on the basis of the number of times an adviser is ranked amongst the top five in the league table (first tier), the next 15 (second tier) or lower (third tier) on the basis of value of transactions advised during each calendar year. The particular rank used in the study is the overall ranking of an adviser which is based on the frequency with which it is in the top five, the next ten, or the rest during the 15 years which Rau's sample covers. The overall rank is calculated once only, that is, Rau implicitly assumes the ranking of an adviser is permanent. Servaes and Zenner (1996) also keep their rankings constant during their entire sample period. Ranking is therefore based partially on ex post information and, if rank is not constant, may not accurately reflect firms' perception of advisers' relative quality at the particular time they select an adviser. It is reasonable to assume that adviser rank does change over time. For instance, Maher and Cooper (1996) describe the change in the 'bulge bracket' and refer to the 'five contenders to the new bulge bracket'. This suggests that over the long run the ranking of advisers can change, hence the need for a ranking method that allows movement between tiers.

3.4.1 Look-Back Method We use a similar ranking method, but avoid Rau's ex-post ranking complication by estimating rankings on the takeover announcement date, not on the year of takeover completion. This ensures both successful and unsuccessful takeovers are included. The bidder has to be an Australian firm so that share returns are available. Adviser ranking is based on the market share achieved during the year. Following Bowers and Miller (1990), Servaes and Zenner (1996), and Rau (2000), the top five ranked advisers for each year are deemed to be first tier, the next ten are second tier, and the remaining advisers are deemed third tier.

The nature of the data is such that market share will appear to vary significantly. There may be years where the market share for an adviser is substantial, then the adviser will disappear from the table for a few years. To overcome the problem of advisers not featuring in the rankings table despite having a high reputation due to their deals in earlier years, a smoothing approach is adopted. In addition to the one-year rankings, the league tables are aggregated over rolling two- and three-year periods.

3.4.2 Cumulative Ex-Ante Ranking Using the 'look-back' ranking, the top five ranked advisers are given a score of three, the next ten are given a score of two, and the remaining advisers receive a score of one. An adviser is only ranked if it is active. All advisers are assumed inactive at the beginning of the data set (1989), once the adviser has participated in a deal it is considered active. The point allocation process is repeated for all active firms during the period 1989 to 1998. The rankings are calculated cumulatively during the period, which makes the ranking process dynamic and realistic. The measure does not handicap advisers if they are new, but penalises advisers who perform deals at irregular intervals. Advisers who drop out of the market completely will slowly have their score approach the minimum of one. The points allocation process builds into the model a degree of market memory.

3.5 Market Share Model

The league table ranks advisers on quality levels. The market share model tests the association between ostensible adviser quality and market share. The cross-sectional model set out below uses data annualised for each adviser, the independent variables being aggregates of the deals performed during the year. The model is based on previous research as well as intuition regarding what the market sees as being valuable in a takeover adviser.

(3) Market Share = [[alpha].sub.0] + [[beta].sub.0]Ability to Complete + [[beta].sub.1]Last Years Market Share + [[beta].sub.2]Accumulated Wealth Gain + [[beta].sub.3]Merger/Tender + [[beta].sub.4]Proportion of Acquisitions + [[beta].sub.5]Average Deal Size + [[beta].sub.6]Time to Complete + [[beta].sub.7]Attitude + [[beta].sub.8]SIC Compatibility + [epsilon]

where: The dependent variable Market Share is the level of market share achieved during the year, as in Rau (2000).

An adviser's Ability to Complete a deal is calculated as the proportion of successfully completed deals to total deals attempted. The deals completed and attempted by adviser j are counted each calendar year.

Last Years Market Share controls for the differences in starting points for the year, that is, a 2% increase in market share has different implications for advisers who had a 4% market share compared to one that had a 10% market share.

Accumulated Wealth Gain is calculated three ways: once as a raw wealth gain, and twice after different methods of standardisation. The variable estimates the wealth adviser j has assisted in generating during the calendar year (as in Bowers & Miller 1990). Accumulated Wealth Gain is calculated as the increase in bidder market value (adjusted for the market index) in the period from the day of announcement until the day the deal is withdrawn or completed. This is calculated for all listed bidding firms advised by adviser j during one calendar year.

The SDC database for Australia does not distinguish between mergers and tenders, but does indicate if the deal is either a merger or a tender, as opposed to an acquisition of assets. Therefore, the Merger/Tender ratio is calculated by dividing the number of mergers and/or tenders performed during the calendar year by the total number of deals performed. It is expected that sales of assets require a different skill set, and a greater client base to find an appropriate buyer, that is, higher ranked advisers are expected to have the expertise required to perform the sale of assets. Thus, less prestigious advisers will perform more merger and tender offers, therefore the coefficient is expected to be positive.

In the US, target advisers are on average paid 30% more than advisers acting for the bidder (McLaughlin 1990). For that reason some advisers may choose to act mainly for targets. The Proportion of Acquisitions variable measures the number of deals adviser j performs as an adviser for bidding firms as a proportion of the total number of deals. Acting for the target takes less time from an adviser's perspective as the target hires its adviser after the bidder does. Thus advisers who act for more targets can perform more deals in one year, hence they have a greater chance of moving up the league table. Therefore there is an expectation of a positive coefficient.

The Average Deal Size is included to control for any systematic differences in the size of deals performed by advisers. It is also included to determine if first tier advisers have their market position simply from performing large deals. If first tier advisers perform larger deals, then the coefficient should be positive.

The Time-to-Complete variable measures the average time (in days) taken to complete the takeovers (announcement date until withdrawal or acceptance of offer) for an adviser in a given year. This is a complexity measure. The longer a deal takes to complete the more complex it is. The expected sign of the coefficient is unknown as a positive coefficient may indicate a deal is complex, and that first tier advisers are the ones who perform the deals. Alternatively, a negative coefficient may mean the more prestigious advisers are better at dealing with complex takeovers and complete them more swiftly.

The Attitude of a deal may be an important determinant of the prestige level of adviser chosen. The annualised variable is calculated as the number of hostile deals divided by the total number of deals. Hostile takeovers have greater risk of not succeeding. Thus to reduce the time and cost of completing a hostile takeover the experience and resources available to more prestigious adviser is important.

The last complexity measure used is the SIC Compatibility ratio. The ratio also proxies for the level of information asymmetry between the bidder and its target. The dummy variable is set to one if the primary 3-digit SIC code of the two firms are identical, and zero otherwise. The ratio is calculated as the number of identical 3-digit SIC code deals performed by adviser j, divided by the total number of deals for adviser j. More prestigious advisers are expected to complete more complex takeovers. Since the SIC compatibility ratio is equal to one if the deals are all simple, then the relationship between the ratio and the level of prestige is expected to be inverse. Thus a negative coefficient is anticipated.

The cross-sectional market share model above is developed to highlight what bidders might consider important. The approach is unique as it is annualised for each adviser. The independent variables are therefore aggregates of the particular attributes of the deals performed during the year. The model should help unlock how an adviser develops and retains its reputational capital (rank).

4. Results

4.1 Descriptive Statistics 

Figure 1 shows the variation in volume of deals over time, and the proportion of takeovers with and without advisers. The figure is based on a larger sample than that used in the regression models. Several trends are salient. The total market for takeovers experienced a slump during the period 1989 until 1994, after which the volume of takeovers grew very rapidly. The greatest growth has occurred in the part of the market using an adviser. The data in figure 1 highlight a shift in the proportion of deals using an adviser around the year 1994/5. The proportion of takeovers performed without an adviser appears to have experienced a permanent downward shift, from around 40% prior to 1994 to slightly less than 20% from 1995. (13)

[FIGURE 1 OMITTED]

The statistics in table 1 describe the main dataset of 801 individual takeovers. There are 401 takeovers with no advisers, and 400 deals where at least one adviser was involved. In line with the US market, there is greater use of target advisers (328) relative to bidder advisers (247), 1991 being the only year where the percentage of bidder advised deals (26.19%) exceeds the percentage of target advised deals (21.43%). As expected, advisers feature more frequently in larger deals, for instance, in 1998, just 42.18% of takeovers in the sample used an adviser, yet figure 1 indicates the cumulative value of adviser assisted deals was over 80% of the cumulative value of deals transacted that year.

Table 2 includes comparisons of deals with and without advisers. For adviser-assisted completed deals the median time it takes to complete the deal (86 days) is almost three times as long as the median time to completion (31 days) of non-adviser assisted deals. This may be because the latter deals are smaller, generally friendly and less complicated. However, it may also be that friendly non-adviser assisted deals have been preceded by informal, unpublicised discussions over a longer period and these are not reflected in the data.

Consistent with Rau (2000), a higher proportion of takeovers are completed when the bidder uses an adviser (82.19% vs. 71.12%). First tier advisers lead both second and third tier advisers in completion rates, the respective rates being 89.87%, 73.08% and 83.33%. However, average deal size monotonically decreases with adviser prestige. Average size of deals advised by first tier advisers is $A100.1m, for second tier advisers it is $A83.0m, third tier adviser deals average $A42.7m, while deals without an adviser average just $Al0m in size. Deals where no adviser is used involve bidders that are twice the size of their targets (1.99). The largest size discrepancy is in deals performed by first tier advisers (4.84) followed by second tier (4.51) and finally third tier (2.26).

Table 2 indicates a bidder is more likely to hire an adviser if: the bidder is listed (70.44% vs. 57.94%), has previous experience (59.52% vs. 43.84%), does not pay cash (64.71% vs. 78.16%), and undertakes a hostile takeover (21.86% vs. 4.51%). Hostile takeovers are generally performed using an adviser (21.86% versus 4.51%). A high proportion of takeovers in which second tier advisers are involved are hostile (28.21%) relative to first (20.25%) or third tier (17.78%) advisers. Interestingly, first tier advisers advise the greatest proportion of bidders with previous experience (63.29%). One implication being that managers and directors of those acquiring firms best placed to assess the value of advisers consider that they are worth their fees. Shareholders also have reason to think well of advisers. Deals using an adviser are more often well received by the market (34.41%) than if no adviser is used (28.43%). However, a majority of deals prompt a negative return on announcement. Further, bidders in takeovers involving second (43.59%) and third (33.33%) tier advisers more often capture over 50% of the value generated than bidders with first tier advisers (21.84%) and in deals where there is no adviser (28.16%). Takeovers assisted by first tier ($A0.70m) and second tier ($A2.76m) advisers create value for the bidder more than 50% of the time.

Table 3 reports returns earned by sample firms around the bid announcement period. Significant returns are concentrated over the windows [-60:0], [-60:+2], [30:+2] days. The returns over periods that exceed 5 days after the takeover ([0: +5], [0:+10] and [0: to completed/withdrawn]) are not significant. None of the mean returns are significantly negative. Only bidders who use a second tier adviser earn significant positive abnormal returns around the time of the announcement. The returns are equally weighted, that is, a 5% abnormal return on a $100 million dollar market capitalisation carries the same weight as a 5% abnormal return on a $20 million firm. This may bias the results against first tier advisers, as it is perceived to be easier to generate a 5% increase in share price if the base is small. However, as first tier bidders are larger then the returns are expected to be relatively smaller.

In short, the broad brush descriptive statistics confirm that the market for advisers functions as one might expect; advisers are more likely to be hired in large deals that are likely also to be complex deals and experienced acquirers also make use of them which suggests that they do contribute to value creation, consistent with the finding that deals involving an adviser are more likely to be well received by the market. However, apparent anomalies appear in the details. For instance, second and third tier advisers capture more value for acquirers than first tier advisers. Our tests and discussion below tease out the issues in more depth.

4.2 The Adviser Use Model

Table 4 contains the results of seven regression estimates for the Adviser Use model. The model is estimated using a logistic regression hence [R.sup.2]'s are not reported. The F statistics and associated p-values are all highly significant, indicating a reasonable fit. Regression 1 is the full model. It contains both listed and unlisted firms. The results indicate a bidder is more likely to use an adviser if it is listed and intends to perform a large, hostile deal, for a diversified firm in the same primary industry, and uses forms of payment other than just cash. Previous experience in acquisitions is insignificantly related to the probability of use of adviser. This result diverges from Servaes and Zenner (1996) finding that less experienced acquirers are more likely to hire advisers. However, on the whole, the evidence is consistent with the hypothesis that advisers are hired to reduce the transaction costs of complex deals. The Market Reaction variable is a proxy for the acquiring firm's managers' estimate of the net present value of the takeover. (14) As such including the variable in the regression allows us to test if managers are more (or less) likely to hire advisers if they are more confident about the value generated by the deal. Regression 2 shows that when the Market Reaction variable is substituted for Bidder Listed, then the F-statistic on the model becomes smaller but remains highly significant. The positive and significant (at the 10% level) coefficient on Market Reaction indicates that the higher managers' expectation of NPV, the more likely that they will decide to hire an adviser. The evidence rules out that advisers are hired for deals with marginal (or negative) NPV. The sign and significance of the estimated coefficients for the other independent variables remain similar to those in regression estimate one.

One possible reason managers of acquiring firms hire advisers is to guard themselves against accusations from shareholders that they did not conduct sufficient due diligence. The need for this kind of 'insurance' is less likely among unlisted bidders which typically are closely controlled and managed by a few concentrated blockholders. Investigating the association between deal complexity and adviser use among unlisted bidder is therefore likely to be a stronger test of the transaction cost hypothesis. When the adviser use model is run for only unlisted bidders (regression 3), some previously significant factors become insignificant. It appears that if the bidder is unlisted an adviser is more likely to be chosen if the deal is hostile, the transaction is large, and the target is diversified, however the method of payment and the relatedness of the target's industry are insignificant. These findings are consistent with the transaction cost hypothesis.

Continuing with application of the adviser use model on subsets on our sample that are likely to give a sharper indication of the role of advisers in reducing takeover transaction costs, we find that bidders with no prior experience (regression 4) are more likely to hire an adviser if the deal is hostile, requires non-cash finance, is large, and if the target is diversified with the target's core industry being similar to the bidder's. If the bidder does have previous experience (regression 5), then the only difference to inexperienced bidders is that the method of payment becomes insignificant. The Bidder Listed variable is insignificant in regression 4 and 5, so there is no support for the information asymmetry hypothesis.

If a target adviser is present (regression 6), then the important determinants again are: whether the deal is hostile, if the deal is large, whether the two firms operate in the same industry, and whether the target is diversified. This suggests that if the target uses an adviser, then the negotiation skills are more important than being able to provide easy access to capital. However, if no target adviser is present (regression 7) then deal attitude, non-cash payment, the size of the deal, and the diversification of the target are important factors in choosing to use an adviser.

In sum, our results indicate the decision by Australian acquiring firms to hire an adviser are consistent with their function being to lower transaction costs, a conclusion that Servaes and Zenner (1996) also reach in their US-based analysis. In the next section, we explore this finding in greater dept by investigating whether ostensibly more competent advisers, as measured by share of dollar value of deals captured, are more likely to be hired in complex deals.

4.3 The Adviser Choice Model

The adviser choice model in table 5 is estimated using a multi-logit regression. The model is estimated four times, once for each of the look-back measures, and once for the ex-ante cumulative ranking measure. We have two intercepts because there are three unique values the dependent variable may take on, hence the estimation is performed in three-dimensional space, rather than the conventional two-dimensional space that standard logit operates in.

For all four models the F-statistic is insignificant, which suggests any inferences drawn from the coefficients of the model must be interpreted with caution. The only significant variables are Attitude and Cash. The positive coefficient on Attitude suggests less prestigious advisers are more often associated with hostile takeovers. The negative coefficient on the Cash variable suggests more complicated financing is performed by lower ranked advisers, and that higher ranked advisers are able to more often raise the required debt finance for its bidder. The lack of significance in all four models suggests the league table is not a good indication of the attributes of an adviser's ability or quality that we would expect to be relevant to bidders. An alternative explanation is that the relevant attributes have been measured with too much error for the results to be significant, for instance, using three tiers to indicate advisor quality might be too coarse.

To account for the above mentioned weakness, we repeat our analysis using adviser ranking, rather than class of tier, as the dependent variable. The rank used is the ex ante cumulative rank and is calculated, respectively, over one year scores, the cumulation of a two-year rolling average, and the cumulation of a rolling three-year average.

Table 6 shows that changing from tiers to rankings improves the goodness of fit, with the F statistic becoming significant for all seven estimated models. For regression 1, the size of the deal is the most important variable when deciding on what prestige level of adviser to use. The negative coefficient indicates that higher ranked advisers perform larger deals, perhaps unsurprising result given our measure of adviser ranking is based on market value of deals advised but this variable is not significant across all regressions. The marginally significant cash variable suggests complicated financing is performed by lower ranked advisers, and that the higher ranked advisers have their clients pay in cash. This may reflect the strong capital backing of the more prestigious advisers who may provide debt finance to fund cash bids.

When the rolling average is extended to two years (regression 2), size no longer is important, instead the significant variable becomes the log of previous acquisitions. The negative coefficient indicates the more deals performed previously, the more likely a bidder is to use a higher ranked adviser. Since size is insignificant there are two possible explanations for the significance of the Previous Acquisition variable. Firstly, repeat bidders understand the value an adviser can add and may consider that higher ranked advisers do a better job. Secondly, higher ranked advisers may be better at suggesting acquisition partners, thereby establishing a good rapport with the bidder, leading to repeat business between the two. In the two and three year models Market Reaction is also marginally significant. The positive coefficient indicates lower ranked advisers' deals are better received by the market. Extending the rolling average to three years (regression 3) sees the method of financing become insignificant, and size becoming significant again. Regression 4 extends the window used to calculate the abnormal announcement return. The window is extended from day -1 to +1 to an 11-day window spanning days -5 to +5. (15) The positive significant variable indicates lower ranked advisers work on takeovers that earn greater announcement period returns.

In regressions 6 and 7 dummy variables are used to test robustness of the findings. In 6, the log of previous acquisitions is removed, and in its place is a dummy to capture whether the bidder has any prior experience. The dummy is significant, the negative coefficient suggesting that more experienced bidders choose to use more prestigious advisers. The interesting observation is that size becomes a more significant variable, which suggests that Log(Previous Acquisitions) used in the previous estimates might be proxying for bidder size. In regression 7 the new variable is a dichotomous variable which is equal to one if the bidder has a significant holding in the target. According to the Corporations Law s707, a person holding more than 5% of the target is a significant shareholder. The lack of significance for the variable suggests the foothold the bidder has in the target is not a consideration for bidders when choosing advisers.

When the results across all seven regressions in table 6 are considered as a whole, they suggest the following inference. The market for corporate advisers may not be accurately characterized as one in which advisers are ranked in order of quality along one or two dimensions. Advisers, like firms in other industries, likely specialize so that they do not directly compete with each other and firms match their particular needs with the attributes of advisers. Large firms likely value the combination of deal financing and deal completion savvy that the higher ranked advisers are able to provide them. These relationships are likely to be reinforced over time so that repeat business is generated. Lower ranked advisers may be less well resourced in terms of providing a 'one-stop' shop for all takeover related business but they might compensate by specializing in particular sectors. The evidence shows that deals in which lower ranked advisers are involved are better received by the market but there is insufficient evidence to conclude that they typically provide a better service because our estimates of market reaction are more likely to be downward biased for the large, well followed firms that make the big acquisitions on which higher ranked advisers work on. We further explore these issues in the next section when investigating the factors that determine advisers' market share.

4.4 The Market Share Model

The market share model is estimated using two different market share bases. In one measure the denominator is the total volume of deals in the market, in the other measure the denominator is the total volume of advised deals, that is, a subset of the total market.

The results in table 7 suggest the method used to rank the advisers (i.e. market share) is not an indicator of other aspects of quality. The only variable that is significant is last year's market share. This suggests advisers are entrenched in their position. The model has significant F values, and a reasonable R2. To take into account the possibility that class of tier is a overly coarse measure of quality, we re estimates the model using rankings. We first review results using the simple look-back ranking measures described in 3.4.1, then review results after ranking advisers on a cumulative ex ante basis.

All models in table 8 have significant F statistics, and reasonable [R.sup.2]'s. For regression 1 the only variable that is significant is last year's market share. This is similar to the results of table 7. Extending the look-back time frame to two (regression 2) and three years (regression 3) suggests the proportion of wealth obtained by the bidder is a significant determinant of current market share. The negative coefficient suggests higher ranked advisers capture greater rents on a consistent basis for the bidder. This supports the superior deal hypothesis. In addition, the compatibility of the bidder and the target also appears important. The negative coefficient indicates more prestigious advisers perform relatively more takeovers where the bidder and the target operate in the same primary industry. This may be an indicator of industry specialisation by higher ranked advisers. Despite having to perform the more complicated deals, results reported in section 5.2 show second tier advisers are better at generating wealth for the bidder, indicating that reputation is built up by deal flow rather than value creation.

In the models that incorporate the cumulative ex ante ranking (regressions 4, 5, and 6), the F values increase. The ability to complete becomes a significant variable. Surprisingly, the coefficient is positive indicating that lower ranked advisers complete more deals attempted (more on this finding in section 4.5). Last year's market share remains an important explanatory of this year's rank. The longer the period of time over which the cumulative ex ante measure is calculated, the less significant the SIC compatibility variable become. The cumulative ex ante rank based on the two and three-year rolling average is negatively associated with the proportion of wealth captured by the adviser. This supports the superior deal hypothesis. The final variable that is significant across all three ex ante models is Attitude, which indicates advisers who perform hostile deals are more likely to be ranked higher. It is surprising to see that the SIC Compatibility changes sign when the ranking method changes. The significant positive variable in regression 4 suggests deals with less information asymmetry are performed by lower ranked advisers, however SIC compatibility is significant in only one of regressions 4, 5 and 6 which suggests the relationship is tenuous.

4.5 Completion Rates Between the Tiers

To test for significant differences in completion rates between the tiers, t-tests were calculated. (16) Completion rates are not reported in tables, however, first tier advisers (89.87%) have a higher completion rate than second tier advisers (73.08%), and a higher completion rate than deals with no advisers (71.12%). This supports the deal completion hypothesis, that is, higher ranked advisers complete more deals. However, there is no significant difference between the completion rates of first (89.87%) and third tier (83.33%) advisers. A possible reason is that smaller advisers operate in niche markets in which they specialise.

Second tier advisers (25.64%) withdraw relatively more deals than first (10.13%), and third tier (13.33%) advisers, and there are more withdrawn second tier advised deals (25.64%) than withdrawn deals involving no adviser (12.27%). (17) Thus second tier advisers appear less able to close out a deal. This provides some support for the hypothesis that bidders prefer first tier advisers because they are more capable of completing a deal. However, the number of observations in the table are small for some categories which should decrease our confidence in generalizing results.

Further analysis of completion rates is done after partitioning the sample into 'good' deals, and 'bad' deals. 'Good' deals are takeovers where the abnormal announcement day bidder return (day -1 to + 1), is non-negative. The results are not reported in detail, however deals performed with advisers have a higher completion rate for both 'good' and 'bad' deals. The only exception is that second tier advisers do not complete a significantly different number of 'bad' deals than deals with no adviser. The withdrawal rates for 'good' deals do not differ significantly. (18)

First tier advisers (88.89%) complete significantly more 'bad' deals than second tier (68.63%) advisers and more than if no adviser were present (73.77%). Third tier advisers complete 82.76% of their deals. Second tier advisers (29.41%) also withdraw more 'bad' deals than first (11.11%) and third tier (15.52%), as well as deals with no advisers (11.76%).

4.5.1 Capturing Wealth Between the Tiers To explore wealth creation further we analyse the proportion of wealth captured. The sample is partitioned by tiers and, within each tier, into two groups, one where the overall outcome of the deal is wealth creation, the other where overall wealth is destroyed. Wealth is defined as the change in market capitalisation of the bidder, divided by the absolute value of the sum of the change in market capitalisation of the bidder and the change in market capitalisation of the target. Change in market capitalisation is adjusted for index movements and is measured from announcement day until completion or withdrawal day.

The results in table 9 suggest that first tier advisers do not perform 'superior' deals. Second tier advisers capture more of the wealth for the bidder in takeovers where wealth is created than first tier advisers (75.34% compared with 34.99%), and first tier advisers do not generate significantly more wealth creation than third tier advisers (11.56%), or deals without an adviser (-6.20%). This supports Rau (2000). Second tier advisers (75.34%) capture significantly more wealth than third tier advisers (11.56%). In takeovers where overall wealth is destroyed, the difference in wealth captured is not significantly different, except that first tier advisers (-137.71%) destroy more bidder wealth than takeovers performed without an adviser (-69.40%). Although there are significant differences in the 'withdrawn' column, the paucity of observations do not allow meaningful interpretation.

When wealth is created it appears that the gains are divided between the bidder and the target. Second tier advisers capture a majority of the wealth for their bidders, as the mean wealth captured in a completed wealth creating deal is 75.34%, that is, for every $A100 wealth created in the takeover, the bidder receives $A75.34. This is the only figure that exceeds 50%. For takeovers where wealth is destroyed it is evident that the bidder is left with the bulk of the losses. The smallest loss is made by deals performed without advisers, where -69.40% of the loss is captured by the bidder. For wealth destroying takeovers that involve an adviser, any gains are on average captured by the target. A percentage that exceeds 100% in the 'wealth destroyed' columns indicates that the bidder's share price decreased, and that wealth was transferred to the target. The 137.71% loss experienced by bidders who use second tier advisers indicates for every $37.71 increase in wealth by the target, the bidder lost $137.71. The fact that the loss is greater for second and third tier advisers suggests they are not able to limit the losses to the same extent a first tier adviser can, however the differences are insignificant. Overall there is some weak support for second tier advisers being able to capture slightly more value for the bidder. However if the overall deal wealth is negative, then second tier advisers appear unable to reduce the losses to the same extent as first tier advisers and others are able. This final observation is not, however, significant. These results indicate the ranking of advisers on league tables does not indicate the underlying quality of the advisers, if quality is estimated using wealth created.

One perspective on the above results is that they are consistent with our earlier observation that takeover advisors specialize and do not compete directly with each other. It is evident that first tier advisers do not gain their position as a consequence of creating greater wealth for their clients, however, it is likely that their relatively indifferent (at best) performance in this regard is not due to shortcomings in the quality of their advice but rather because acquiring firms secure their services primarily for their assistance in financing and completing deals to which they are committed. We do not offer direct evidence on this but such a reading is consistent with Allen, Jagtiani, Peristiani and Saunders' (2001) findings. One implication of this interpretation is that concern about systematic propensities to make value-decreasing takeovers--recall that we find, consistent with other studies, that market reaction to acquiring firms' announcements of takeover bids is typically negative--should be focused on the management of acquiring firms than their advisers.

Notwithstanding the above and the widespread use of the adviser league tables based on market, it might that adviser market share rankings are insufficiently correlated with the one factor that investors most care about-increases in their wealth. In the next section we test of ranking advisers on their wealth creating ability is associated with increases in their market share.

4.6 Wealth Creation Ranking

Our results indicate the SDC league table is not consistently capturing aspects of quality other than volume and perhaps completion rate. Managers represent shareholders, and should therefore strive to add value to the company, for this reason it may be more appropriate to rank advisers on their ability to capture wealth for the bidder.

The wealth measure we develop is:

Wealth Creation Score = [SIGMA] BidderWealthGain 1/# Advisers/ [SIGMA] DealValue 1/# Advisers

The measure is the division of the wealth added to the bidder, divided by the number of advisers working on a given deal. Wealth is measured as the change in market capitalisation of the bidder from the time the bid is announced, until the takeover is completed, or withdrawn. The division by the number of advisers in each deal is done to share the gains between the advisers so that a $10 million gain on a deal performed by one adviser is given twice the value of a $10 million gain on a deal performed with two advisers. The numerator is then summed before being divided by the weighted value of all deals the adviser has participated in.

The denominator, like the numerator, is a weighted total of all deals in which the adviser has been involved. Like the numerator, the denominator is summed prior to division, if this were not done, then the weighting would be superfluous, and the measure would simply become a summation of averages. The wealth creation scores are calculated on an annual basis for each adviser, after which the advisers' scores are ranked from highest to lowest.

The wealth measure has several merits. High ranking is achieved by adviser who add value to their clients. The focus shifts from the number and size of deals to the variable that firms arguably most care about. However, it only measures deals involving listed bidders (19) and work performed for targets is not considered.

The wealth creation model is tested by substituting the lagged ranking into the market share model. The market share model is estimated three times, once for each 'look-back'. For the 2 and 3-year 'look-backs', the lagged wealth rank is estimated over a two or three year periods, and lagged by two or three years. In order for the regressions to be estimated all variables have to be available. The reduction in sample size comes mainly from the fact that advisers have to be in the sample for both the year of the SDC ranking, and the previous year for the wealth creation ranking to be calculated.

The model estimates are shown in table 10. In regression 1 the positive significant variable on the lagged wealth rank suggests, that advisers who add value to their clients, are rewarded the following year with an increase in deal flow, that is, a higher rank on the league table. However, when the measure is lagged two years, and compared to the two year rolling average market share, then the measure remains significant, but the model itself is not well specified. If the ranks are averaged over three years, the model is again not well specified, with a negative [R.sup.2], and an insignificant F value.

Although the results suggest that the market rewards advisers who create value there are two caveats. The sample size is relatively small (between 32 and 49 takeovers) and the sharp decline in goodness of fit suggests the model might not be well specified.

5. Summary and Conclusions

Using a sample of 801 Australian takeovers contained in SDC Platinum attempted between 1989 and 1998, this paper investigated three questions. What deal attributes motivate bidders to engage advisers? What factors do bidders consider when choosing advisers? Does capturing greater value for the bidder result in greater market share?

Our results support Servaes and Zenner's (1996) conclusion from their study of US investment banks that one function they perform is to reduce the transaction costs of completing complex deals. In relation to factors that influence choice of adviser our results indicate higher ranked advisers complete more deals than second-tier advisers, but not third-tier advisers, a finding not well aligned with Rau's (2000) conclusion that first-tier advisers are hired for their ability to complete a deal. However, in line with Rau (2000) we find first-tier advisers are also more likely to force a 'bad' deal to completion. When the dependent variable is changed from tiers to ranks, the adviser-choice model highlights that the league table is not a good indicator of ability to complete a deal completion or wealth creation, neither factor is significant. The market-share model with rankings reaffirms this, showing that the most important determinant of this year's market share is last year's market share. The market-share model also shows that less prestigious advisers work on deals that earn greater announcement returns, and are able to capture a greater amount of the wealth created in a takeover on a consistent basis. Our results are consistent with a degree of specialisation among advisory firms such that lower-ranked advisers in terms of market share are able to hold their own in terms of valued added to their clients. A fruitful avenue for future research is to investigate the nature of advisers' specialisation.

The SDC-produced league table is a key marketing tool used by advisers to generate future revenue. The way the league table is calculated may be parsimonious and easy to interpret, but it does not capture key quality aspects such as completion rates or shareholder wealth creation. We find support for the proposition that an adviser ranking system based on wealth created is able to predict changes in adviser deal flow in subsequent periods, but our results are based on a small sample size.

(Date of receipt of final transcript: March, 2004. Accepted by Garry Twite, Special Issue Editor.)

Table 1
Distribution of Advisers

Distribution of 801 Australian takeovers from 1989 through to 1998.
The proportions in the All Takeovers column, are the percentages
relative to the total number of takeovers performed (801). The
proportions in the 'With an Adviser', 'Without and Adviser',
'With an Bidder Adviser', and 'With a Target Adviser' columns,
are the proportions that group of advisers represent for that
year, that is, relative to the 'All Takeovers' Column e.g. in
1989 eleven takeovers in with an adviser represented 44% of
the twenty five takeovers in 1989.

Year      All Takeovers            With an              Without an
                                 Adviser (a)            Adviser (b)

        Number   Proportion   Number   Proportion   Number   Proportion

1989      25         3.12%       11      44.00%        14      56.00%
1990      23         2.87%        5      21.74%        18      78.26%
1991      42         5.24%       14      33.33%        28      66.67%
1992      34         4.24%       23      67.65%        11      32.35%
1993      51         6.37%       30      58.82%        21      41.18%
1994      50         6.24%       27      54.00%        23      46.00%
1995      89        11.11%       58      65.17%        31      34.83%
1996      92        11.49%       65      70.65%        27      29.35%
1997     101        12.61%       43      42.57%        58      57.43%
1998     294        36.70%      124      42.18%       170      57.82%
Total    801       100%         400       n/a         401       N/a

Year      With an Bidder        With a Target
           Adviser (c)           Adviser (d)

        Number   Proportion   Number   Proportion

1989       6       24.00%        7       28.00%
1990       3       13.04%        3       13.04%
1991      11       26.19%        9       21.43%
1992      17       50.00%       18       52.94%
1993      22       43.14%       22       43.14%
1994      19       38.00%       25       50.00%
1995      30       33.71%       50       56.18%
1996      48       52.17%       61       66.30%
1997      26       25.74%       33       32.67%
1998      65       22.11%      100       34.01%
Total    247        n/a        328        n/a

Note: Percentages may not sum to 100 because of rounding.

(a) With an Adviser: the deal includes at least one adviser
involved in the deal. The adviser can work for either the
target or the Bidder;

(b) Without an Adviser: neither the Bidder nor the target
employ the services of an adviser during the course of
the takeover;

(c) With a Bidder Adviser: The deal is performed using
at least one Bidder adviser; and

(d) With a Target Adviser: The deal is performed using
at least one target adviser.

Table 2
Descriptive Statistics

Descriptive statistics for 801 Australian takeovers
between January 1989 and December 1998.

                                               All          No
                                            Takeovers   Adviser (a)

Observations                                 801         401
Median time to complete (d)                   55          31
Median time to withdrawn (e)                  85          79
Completion ratio (f)                          74.53%      71.12%
Withdrawal ratio (g)                          13.48%      12.27%
Median deal size (net of debt)                21.3        10.0
  $A millions
Median relative Size (h)                       2.28        1.99
Proportion of bidders listed (i)              61.80%      57.94%
Identical 3-digit SIC code                    35.83%      33.57%
Same ultimate parent (j)                       8.99%       9.03%
Bidders with previous experience              48.68%      43.84%
Hostile deals                                  9.86%       4.51%
Pure cash acquisitions                        74.03%      78.16%
Percent owned prior to acquisition             8.91%       8.98%
Percent owned after acquisition               66.73%      63.90%
Percent seeking to own after                  91.22%      88.93%
  acquisition (k)
Market reacts positively (l)                  31.46%      28.43%
Bidder captures >50% of wealth (m)            22.78%      28.16%
Median value added to bidder ($A million)     -0.09       -0.19
Median value added to target ($A million)      1.36        0.08

                                              Bidder      First Tier
                                            Adviser (b)   Adviser (c)

Observations                                  247            79
Median time to complete (d)                    86            87
Median time to withdrawn (e)                   90            84
Completion ratio (f)                           82.19%        89.87%
Withdrawal ratio (g)                           16.19%        10.13%
Median deal size (net of debt)                 72.0         100.1
  $A millions
Median relative Size (h)                        3.33          4.84
Proportion of bidders listed (i)               70.44%        72.15%
Identical 3-digit SIC code                     40.89%        41.77%
Same ultimate parent (j)                        8.91%        10.12%
Bidders with previous experience               59.52%        63.29%
Hostile deals                                  21.86%        20.25%
Pure cash acquisitions                         64.71%        65.82%
Percent owned prior to acquisition              8.75%         6.36%
Percent owned after acquisition                73.07%        77.19%
Percent seeking to own after                   96.37%        93.18%
  acquisition (k)
Market reacts positively (l)                  34.41%         32.91%
Bidder captures >50% of wealth (m)            10.73%         21.84%
Median value added to bidder ($A million)     -0.18           0.70
Median value added to target ($A million)      6.43          13.24

                                            Second Tier   Third Tier
                                            Adviser (c)   Adviser (c)

Observations                                   78            90
Median time to complete (d)                    92            83
Median time to withdrawn (e)                   86           102
Completion ratio (f)                           73.08%        83.33%
Withdrawal ratio (g)                           25.64%        13.33%
Median deal size (net of debt)                 83.0          42.7
  $A millions
Median relative Size (h)                        4.51          2.26
Proportion of bidders listed (i)               66.67%        70.00%
Identical 3-digit SIC code                     42.31%        38.89%
Same ultimate parent (j)                        8.97%         7.78%
Bidders with previous experience               61.54%        54.44%
Hostile deals                                  28.21%        17.78%
Pure cash acquisitions                         73.08%        56.67%
Percent owned prior to acquisition             10.41%         9.42%
Percent owned after acquisition                65.04%        76.41
Percent seeking to own after                   97.39%        98.30%
  acquisition (k)
Market reacts positively (l)                   32.05%        37.78%
Bidder captures >50% of wealth (m)             43.59%        33.33%
Median value added to bidder ($A million)       2.76         -1.31
Median value added to target ($A million)       6.40          2.9

Note: (a) No Adviser: neither target or bidder hired advisers;

(b) Bidder Adviser: at least one bidder adviser participated;

(c) The descriptive statistics are for deals with at least
one bidder adviser;

(d) Time to complete: the number of days from announcement
until deal is complete;

(e) Time to withdraw: the number of days from announcement
until deal is withdrawn;

(f) Completion ratio is calculated as: the number of deals
completed, divided by the number of deals attempted;

(g) Withdrawal ratio is calculated as: the number of deals
withdrawn, divided by the number of deals attempted (note
there are 96 deals still 'pending' in the sample, hence the
completion and withdrawal ratios do not sum to 100%);

(h) Relative size is Total Assets of the Bidder, divided
by Total Assets of the Target;

(i) This is the percent of deals where the bidder is a
publicly listed company;

(j) The percentage indicates the proportion of deals where
the target and the bidder have the same ultimate parent;

(k) The stake the bidder would hold if all shares bid for
were acquired. The percentage is the sum of the shares
owned prior to the acquisition and the number of shares
seeking to be purchased during the takeover;

(l) The percent of deals where the abnormal announcement
day return (day -5 to +5) is positive. Only listed
bidders included; and

(m) If the dollar return of the bidder exceeds the dollar
return of the target then more than 50% of wealth has been
captured by the bidder. Sample size reduced as only
takeovers with listed bidder and target can be used.

Table 3 Announcement Period Returns

The table below presents the mean abnormal buy-and-hold returns
earned by bidders for the periods indicated using the various
level of advisers.

     Window           All Listed Bidders       No Bidder Adviser
(Days relative to
  announcement)

                     Return %    Student-t    Return %    Student-t
                    (n = 486)                (n = 314)

-60 to 0             4.007 ***     3.422      3.938 **      2.524
-60 to +2            4.421 ***     3.643      4.302 ***     2.626
-30 to +2            2.900 ***     3.097      2.221 *       1.882
-10 to +10           1.216 **      2.186      0.578         0.854
-5 to +5             1.132 **      2.244      0.587         0.967
-2 to +2             0.695 *       1.65       0.706         1.204
-1 to +1             0.342         0.987      0.049         0.108
0 to +2              0.465         0.207      0.398         0.763
0 to +5              0.387         0.929      0.110         0.206
0 to +10             0.366         0.792     -0.013        -0.022
0 to Completed       1.408         0.817      2.191         0.871
0 to Withdrawn      -0.501        -0.293     -2.319        -1.331

     Window          Bidders Using First    Bidders Using Second
(Days relative to     Tier Advisers           Tier Advisers
  announcement)

                    Return %   Student-t    Return %   Student-t
                    (n = 57)                (n = 52)

-60 to 0             4.081 *      1.746     3.972 *      1.790
-60 to +2            4.443 *      1.786     4.936 **     2.185
-30 to +2            3.843        1.325     3.037        1.539

-10 to +10           1.137        0.625     1.929        1.478
-5 to +5             1.114        0.793     2.793 *      1.858
-2 to +2             0.064        0.095     1.365 *      1.850
-1 to +1            -0.250       -0.471     2.100 **     2.646
0 to +2              0.276        0.451     1.023 *      1.689
0 to +5              1.140        0.887     0.726        0.979
0 to +10             0.914        0.597     0.790        0.854
0 to Completed       0.303        0.078     4.671        1.396
0 to Withdrawn       8.976        0.879     2.246        0.701

     Window          Bidders Using Third
(Days relative to     Tier Advisers
  announcement)

                    Return %    Student-t
                    (n = 63)

-60 to 0             4.319        1.167
-60 to +2            4.569        1.276
-30 to +2            5.322 *      1.858
-10 to +10           3.881 **     2.187
-5 to +5             2.494        1.478
-2 to +2             0.657        0.574
-1 to +1             0.887        0.759
0 to +2              0.508        0.578
0 to +5              0.807        0.646
0 to +10             1.411        1.122
0 to Completed      -3.490        1.013
0 to Withdrawn       7.313        0.780

Note: *** Significant at the 1% level:

** Significant at the 5% level: and

* Significant at the 10% level.

Table 4
Adviser Use Model

The adviser use model is estimated using a logit regression. The
dummy is equal to one if the bidder used an adviser, and zero
otherwise. Adviser Used = [[alpha].sub0] + [[beta].sub.1]Attitude
+ [[beta].sub.2]Cash + [[beta].sub.3]Log(Transaction Size) +
[[beta].sub.4] Log(Previous Acquisitions) + [[beta].sub.5]Related
Industry + [[beta].sub.6]Log(# Target SICS) + [[beta].sub.7]Market
Reaction + [[beta].sub.8]Bidder Listed + [epsilon]. Numbers in
parentheses are standard errors. Significance levels are
determined using one-tailed t-test cut-off values, as
there are a priori expectations on the variables.

                          Full Model     Full Model      Unlisted
                          No Returns    With Returns      Firms

Regression estimate         1             2              3
Intercept                  -3.871 ***    -3.543 ***      4.542 ***
                           (0.428)       (0.471)        (0.840)
Slope coefficients on:
  Attitude                  1.561 ***     0.999 ***      2.692 ***
                           (0.281)       (0.330)        (0.540)
  Cash                     -0.378 **     -0.419 **      -0.161
                           (0.204)       (0.227)        (0.508)
  Log(Transaction Size)     0.487 ***     0.502 ***      0.518 ***
                           (0.054)       (0.070)        (0.093)
  Log(Previous              0.028         0.028          0.088
    Acquisitions)          (0.084)       (0.100)        (0.161)
  Related Industry          0.474 ***     0.665 ***      0.270
                           (0.184)       (0.2286)       (0.337)
  Log(#Target SICs)         0.380 ***     0.377 ***      0.544 **
                           (0.151)       (0.187)        (0.266)
  Market Reaction             --          0.037 ***        --
                                         (0.015)
  Bidder Listed             0.393 **        --             --
                           (0.203)
F-statistic               191.635 ***   112.091 ***     83.312 ***
Observations              801           486            315

                          No Previous   With Previous
                          Experience     Experience

Regression estimate         4             5
Intercept                  -4.021 ***    -3.490 ***
                           (0.611)       (0.623)
Slope coefficients on:
  Attitude                  1.671 ***     1.484***
                           (0.452)       (0.361)
  Cash                     -0.617 **     -0.233
                           (0.325)       (0.264)
  Log(Transaction Size)     0.500 ***     0.473 ***
                           (0.076)       (0.077)
  Log(Previous                --             --
    Acquisitions)
  Related Industry          0.466 **      0.494 **
                           (0.272)       (0.251)
  Log(#Target SICs)         0.511 **      0.280 *
                           (0.232)       (0.201)
  Market Reaction             --             --

  Bidder Listed             0.299         0.248
                           (0.288)       (0.300)
F-statistic                90.835 ***    88.014 ***
Observations              411           390

                          Target Adviser      No Target
                             Present       Adviser Present

Regression estimate           6                7
Intercept                    -3.139 ***       -3.662 ***
                             (0.621)          (0.652)
Slope coefficients on:
  Attitude                    1.170 ***        1.278 **
                             (0.349)          (0.548)
  Cash                       -0.092           -0.480 *
                             (0.280)          (0.326)
  Log(Transaction Size)       0.480 ***        0.296 ***
                             (0.080)          (0.081)
  Log(Previous                0.176 *         -0.052
    Acquisitions)            (0.126)          (0.130)
  Related Industry            0.510 **         0.030
                             (0.262)          (0.301)
  Log(#Target SICs)           0.296 *          0.450 **
                             (0.217)          (0.236)
  Market Reaction               --               --

  Bidder Listed              -0.017             0.743 **
                             (0.302)           (0.317)
F-statistic                  75.967 ***       39.660 ***
Observations                328              473

Note: *** Significant at the 1% level;

** Significant at the 5% level; and

* Significant at the 10% level.

Table 5
Adviser Choice Model--Using Adviser Tier as Indicator of Quality

The adviser choice model tests [H.sub.3]. The model is estimated using
the following multi-logit regression. Tier = [alpha] + [[beta].sub.1]
Attitude + [[beta].sub.2]Cash Dummy + [[beta].sub.3]Log(Transaction
Size) + [[beta].sub.4]Log(Previous Acq.) + [[beta].sub.5]Log(#Target
SIC Codes) + [[beta].sub.6]Market Reaction + [[beta].sub.7]Relative
Size + [[beta].sub.8] Significant Bidder + [epsilon]. The dependent
variable varies depending on the ranking method. The dependent
variables is one if  a first tier adviser is used, two is a second
tier adviser, and three is a third tier adviser. The data  used for
this regression are from a sample of 247 takeovers where bidder
adviser has been hired. A sample  firm is included only if all
variables are available. Numbers in parentheses indicate the
standard errors. There are a priori expectations for all the
variables, thus one tailed student-t values are used.

Dependent Variable      1 Year      2-Year      3-Year      Ex-Ante
                       Look-Back   Look-Back   Look-Back   Cumulative

Regression estimates     1           2          3            4
Intercept 1              2.256 *     1.756      2.666 *      1.187
                        (1.519)     (1.529)    (1.568)      (1.510)
Intercept 2              0.643       0.606      1.120       -0.089
                        (1.488)     (1.511)    (1.525)      (1.501)
Slope coefficients
  Attitude               0.723       0.903 *    1.401 **     0.276
                        (0.619)     (0.634)    (0.671)      (0.629)
  Cash                  -0.836 *    -0.741     -0.959 *     -0.819 *
                        (0.575)     (0.585)    (0.595)      (0.577)
  Log(Transaction       -0.190      -0.168     -0.287       -0.101
    Size)               (0.227)     (0.228)    (0.232)      (0.226)
  Log(Previous           0.124      -0.109     -0.191       -0.051
    Acquisitions)       (0.249)     (0.258)    (0.270)      (0.250)
  Ln(#Target SIC        -0.190      -0.138     -0.086        0.173
    Codes)              (0.428)     (0.444)    (0.448)      (0.430)
  Relative Size         -0.003      -0.001     -0.010        0.020 *
                        (0.010)     (0.010)    (0.012)      (0.012)
  Market Reaction        0.017       0.039      0.028        0.013
                        (0.043)     (0.043)    (0.044)      (0.043)
  Significant Bidder    -0.054      -0.036     -0.089        0.376
                        (0.615)     (0.628)    (0.638)      (0.622)
Observations            56          55         54           56
F-statistic              5.135       5.303      9.836        5.450

Note: *** Significant at the 1% level;

** Significant at the 5% level; and

* Significant at the 10% level.

Table 6
Adviser Choice Model--Using Adviser Rankings as Indicator of Quality

The adviser choice model Rank = [alpha] + [[beta].sub.1]Attitude +
[[beta].sub.2]Cash Dummy + [[beta].sub.3]Log(Transaction Size) +
[[beta].sub.4]Log(Previous Acq.) + [[beta].sub.5]Log(#Target SIC
Codes) + [[beta].sub.6]Market Reaction + [[beta].sub.7]Sign. The
data used for this regression is based on a sample of 801 takeovers.
The observations contained in the 'Observations' column indicate
the number of takeovers in each regression. Numbers in
parentheses indicate the heteroseedasticity adjusted
standard errors (White 1980).

Dependent Variable               1-Year        2-Year        3-Year
                                 Ex Ante       Ex Ante       Ex Ante
                                  Rank          Rank          Rank

Regression estimates             1             2             3
Intercept                       23.952 ***    19.962 ***    17.485 ***
                                (4.413)       (3.473)       (2.794)
Slope coefficients on:
  Attitude                       1.162         1.353         0.824
                                (2.483)       (2.036)       (1.646)
  Cash                          -2.782 *      -2.382 *      -1.547
                                (1.875)       (1.534)       (1.270)
  Log(Transaction Size)         -1.393 **     -0.681        -0.656 *
                                -0.646        -0.534        -0.435
  Log(Previous Acquisitions)    -1.106        -1.474 **     -1.326 **
                                (0.851)       (0.726)       (0.610)
  Previous acquisition             --            --            --
    dummy
  Log(#Target SIC Codes)         0.268        -0.210         0.129
                                (1.540)       (1.219)       (1.016)
  Market Reaction (-1 to +1)     0.107         0.125 *       0.104 *
                                (0.156)       (0.096)       (0.078)
  Market Reaction (-5 to +5)       --            --            --

  Market Reaction (-10 to          --            --            --
    +10)
  Shares held prior to           0.012         0.019        -0.007
    takeover                    (0.040)       (0.036)       (0.029)
  Significant shareholder          --            --            --
Observations                   171           169           167
F-statistic                      1.990 *       2.236 **      2.280 **
Adjusted [R.sup.2]               3.89          4.87          5.09

Dependent Variable               3-Year        3-Year
                                 Ex Ante       Ex Ante
                                  Rank          Rank

Regression estimates             4             5
Intercept                       17.224 ***    16.960 ***
                                (2.764)       (2.752)
Slope coefficients on:
  Attitude                       0.981         1.056
                                (1.627)       (1.605)
  Cash                          -1.736 *      -1.394
                                (1.244)       (1.272)
  Log(Transaction Size)         -0.616 *      -0.576 *
                                (0.438)       (0.406)
  Log(Previous Acquisitions)    -1.282 **     -1.320 **
                                (0.599)       (0.602)
  Previous acquisition             --            --
    dummy
  Log(#Target SIC Codes)         0.067        -0.008
                                (1.003)       (0.991)
  Market Reaction (-1 to +1)       --             --

  Market Reaction (-5 to +5)     0.090 **         --
                                (0.048)
  Market Reaction (-10 to          --          0.090 **
    +10)                                      (0.044)
  Shares held prior to          -0.001        -0.001
    takeover                    (0.029)       (0.028)
  Significant shareholder          --            --
Observations                   167           167
F-statistic                      2.522 **      2.594 **
Adjusted [R.sup.2]               6.00          6.26

Dependent Variable               3-Year        3-Year
                                 Ex Ante       Ex Ante
                                  Rank          Rank

Regression estimates             6             7
Intercept                       18.352 ***    16.929 ***
                                (2.811)       (2.729)
Slope coefficients on:
  Attitude                       0.925         1.062
                                (1.606)       (1.583)
  Cash                          -1.593        -1.393
                                (1.257)       (1.274)
  Log(Transaction Size)         -0.753 **     -0.572 *
                                -0.432        -0.442
  Log(Previous Acquisitions)       --         -1.327 **
                                              (0.601)
  Previous acquisition          -2.999 **         --
    dummy                       (1.435)
  Log(#Target SIC Codes)        -0.021        -0.010
                                (0.999)       (0.995)
  Market Reaction (-1 to +1)       --             --

  Market Reaction (-5 to +5)       --             --

  Market Reaction (-10 to        0.088 **      0.090 **
    +10)                        (0.043)       (0.045)
  Shares held prior to          -0.001            --
    takeover                    (0.028)
  Significant shareholder          --          0.064
                                              (1.324)
Observations                   167           167
F-statistic                      2.598 **      2.595 **
Adjusted [R.sup.2]               6.28          6.27

Note: *** Significant at the 1% level;

** Significant at the 5% level; and

* Significant at the 10% level.

Table 7
Market Share Model--Using Adviser Tier as Indicator of Quality

The market share model tests [H.sub.4], [H.sub.5], and [H.sub.6], and
is estimated using an OLS regression.  Market Share = [[alpha].sub.0]
+ [[beta].sub.0]Ability to  Complete + [[beta].sub.1]Last Years Market
Share +  [[beta].sub.2]Accumulated Wealth Gain + [[beta].sub.3]Merger/
Tender + [[beta].sub.4]Proportion of Acquisitions + [[beta].sub.5]
Average Deal Size + [[beta].sub.6]Time to Complete + [[beta].sub.7]
Attitude + [[beta].sub.8]SIC Compatibility + [epsilon]. The dependent
variable in all cases is the market share of the adviser. The data
used for this regression is aggregated for each adviser on a
per-annum basis. Numbers in  parentheses are standard errors
adjusted for heteroscedasticity (White 1980). The t-tests
are one tailed.

Dependent Variable         Market share   Advised deals     Advised
                           for advised    standardised    deals wealth
                              deals          wealth        proportion

Regression estimates         1               2              3
Intercept                    7.676 **        7.842 **       7.445 *
                            (4.430)         (4.203)        (4.517)
Slope coefficients on:
  Ability to complete       -1.896          -2.009         -1.816
                            (2.763)         (2.727)        (2.733)
  Last years market          0.579 ***       0.609 ***      0.576 ***
    share                   (0.118)         (0.115)        (0.115)
  Accumulated wealth        -0.000              --             --
    gain                    (0.000)
  Bidder wealth per deal        --          -0.845             --
                                            (1.079)
  Proportion of wealth          --              --          0.053
    created                                                (0.098)
  Merger/Tender              0.600           0.711          0.678
                            (2.191)         (2.033)        (2.193)
  Proportion of             -3.223          -2.747         -2.853
    Acquisitions            (6.904)         (6.527)        (7.441)
  Average deal size          0.002           0.001          0.002
                            (0.006)         (0.005)        (0.005)
  Time to complete          -0.003          -0.001         -0.003
                            (0.006)         (0.006)        (0.006)
  Attitude                   1.917           1.504          2.006
                            (2.616)         (2.547)        (2.566)
  SIC compatibility         -1.068          -0.810         -1.031
                            (2.447)         (2.428)        (2.338)
Observations                71              75             71
F-statistic                  4.132 ***       5.050 ***      4.144 ***
Adjusted [R.sup.2]          28.43           32.69          28.49

Dependent Variable            Market        All deals     All deals
                           share for all   standardised     wealth
                               deals          wealth      proportion

Regression estimates          4               5            6
Intercept                     4.902 *         4.579 *      4.779 *
                             (3.302)         (3.098)      (3.367)
Slope coefficients on:
  Ability to complete        -0.916          -0.901       -0.874
                             (1.912)         (1.892)      (1.911)
  Last years market           0.633 ***       0.662 ***    0.631 ***
    share                    (0.133)         (0.130)      (0.131)
  Accumulated wealth         -0.000               --          --
    gain                     (0.000)
  Bidder wealth per deal         --          -0.414            --
                                             (0.834)
  Proportion of wealth           --              --        0.033
    created                                               (0.075)
  Merger/Tender             -0.007            0.285        0.035
                            (1.589)          (1.493)      (1.589)
  Proportion of             -2.152           -2.283       -1.924
    Acquisitions            (5.164)          (4.835)      (5.583)
  Average deal size          0.003            0.002        0.003
                            (0.005)          (0.005)      (0.005)
  Time to complete          -0.001            0.001       -0.001
                            (0.005)          (0.005)      (0.015)
  Attitude                   1.088            0.754        1.131
                            (1.864)          (1.846)      (1.840)
  SIC compatibility         -0.174            0.250       -0.163
                            (1.837)          (1.842)      (1.764)
Observations                71               75           71
F-statistic                  5.189 ***        6.108 ***     5.197 ***
Adjusted [R.sup.2]          34.72            38.01         34.73

Note: *** Significant at the 1 % level;

** Significant at the 5% level; and

* Significant at the 10% level.

Table 8
Market Share Model--Using Adviser Rankings as Indicator of Quality

The market share model Rank = [[alpha].sub.0] + [[beta].sub.0] Ability
to Complete + [[beta].sub.1] Last Years Market Share + [[beta].sub.2]
Accumulated Wealth per Deal + [[beta].sub.3] Merger/Tender +
[[beta].sub.4] Proportion of Acquisitions + [[beta].sub.5] Average
Deal Size + [[beta].sub.6] Time to Complete + [[beta].sub.7] Attitude +
[[beta].sub.8] SIC Compatibility + [epsilon] is estimated using an OLS
regression. The dependent variable in this sensitivity analysis is the
rank attained by the adviser. The data used for this regression is
aggregated for each adviser on a per-annum basis. Numbers in
parentheses are heteroscedasticity adjusted standard errors (White
1980). The Rank measure aggregates the deals done by the adviser over
a 1, 2 or 3 year period. The ex Ante rank is based on the rank
achieved by an adviser based on its points scored during the years it
has been active. As there are a priori expectations for all the
independent variables, the significance level cut-offs are calculated
based on one-tailed t-tests.

   Dependent Variable       Rank 1-Year   Rank 2-Year   Rank 3-Year
                             Look-Back     Look-Back     Look-Back

Regression estimates         1             2             3
Intercept                   21.525 ***    27.235 ***    26.518 ***
                            (5.559)       (5.500)       (7.813)
Slope coefficients on:
  Ability to complete       -1.501        -2.754        -0.208
                            (3.987)       (4.148)       (5.742)
  Last years market share   -0.551 ***    -0.739 ***    -0.953 ***
                            (0.100)       (0.130)       (0.184)
  Bidder wealth per deal    -0.794        -3.085 **     -2.624 **
                            (1.256)       (1.502)       (1.501)
  Merger/Tender             -1.406        -2.684         0.793
                            (2.480)       (3.192)       (3.633)
  Proportion of              2.766         2.641        -2.406
    acquisitions            (6.332)       (7.738)       (8.026)
  Average deal size         -0.007        -0.009 *      -0.007
                            (0.007)       (0.006)       (0.008)
  Time to complete           0.010         0.021 **      0.005
                            (0.007)       (0.009)       (0.015)
  Attitude                  -4.312        -3.906        -3.021
                            (3.461)       (4.431)       (4.127)
  SIC compatibility         -1.868        -9.635 ***    -7.122 *
                            (3.037)       (3.804)       (4.300)
Observations                75            75            73
F-statistic                  4.154 ***     3.327 ***     2.945 ***
Adjusted [R.sup.2]          27.46         21.83         19.34

   Dependent Variable        Ex Ante      Ex Ante      Ex Ante
                            Rank 1-Yr    Rank 2-Yr    Rank 3-Yr
                            Look-Back    Look-Back    Look-Back

Regression estimates         4            5            6
Intercept                   16.073 ***   16.497 ***   17.244 ***
                            (5.413)      (4.354)      (3.747)
Slope coefficients on:
  Ability to complete       10.306 ***    6.627 **     5.550 **
                            (3.074)      (2.807)      (2.376)
  Last years market share   -0.959 ***   -0.869 ***   -0.954 ***
                            (0.093)      (0.088)      (0.099)
  Bidder wealth per deal    -0.348       -2.397 ***   -1.367 **
                            (0.954)      (0.892)      (0.626)
  Merger/Tender              2.785        0.653        2.201 *
                            (2.479)      (2.318)      (1.657)
  Proportion of             -1.698        1.847       -2.401
    acquisitions            (6.993)      (6.335)      (4.991)
  Average deal size         -0.004       -0.004       -0.001
                            (0.005)      (0.003)      (0.003)
  Time to complete           0.002        0.005       -0.010
                            (0.007)      (0.006)      (0.008)
  Attitude                  -8.222 **    -4.947 ***   -2.603 *
                            (3.449)      (2.634)      (1.785)
  SIC compatibility          5.815 **    -1.396       -1.783
                            (3.181)      (2.462)      (1.761)
Observations                75           75           73
F-statistic                 10.176 ***    8.331 ***   11.057 ***
Adjusted [R.sup.2]          52.41        46.80        55.36

Note: *** Significant at the 1% level;

** Significant at the 5% level; and

* Significant at the 10% level.

Table 9
Wealth Captured by Various Adviser Tiers for Deals that Create or
Destroy Overall Wealth

The table assesses if there is a difference between the wealth
retained between tiers, and between deals where wealth is created, or
destroyed. The student t-statistic calculated for each paring
indicates whether there is a significant difference between means of
the two tiers. The classification method used is based on the 1-year
look-back method. Wealth is created when the sum of the dollar gain to
the bidder plus the dollar gain to the target exceeds zero. The sample
is slightly reduced, as the wealth measure requires both the target
and  the bidder to be listed.

                              Overall Wealth Created

                           Completed          Withdrawn

First Tier Adviser      34.99%   (25)       69.14%   (2)
Second Tier Adviser     75.34%   (22)      -12.34%   (7)
t-statistic (p-value)    2.118   (0.040)    -1.259   (0.251)

First Tier Adviser      34.99%   (25)       69.14%   (2)
Third Tier Adviser      11.56%   (16)      -53.73%   (6)
t-statistic (p-value)   -0.730   (0.473)    -3.092   (0.021)

First Tier Adviser      34.99%   (25)       69.14%   (2)
No Adviser              -6.20%   (57)       60.08%   (20)
t-statistic (p-value)   -0.714   (0.478)    -0.304   (0.768)

Second Tier Adviser     75.34%   (22)      -12.34%   (7)
Third Tier Adviser      11.56%   (16)      -53.73%   (6)
t-statistic (p-value)   -1.968   (0.062)    -0.570   (0.582)

Second Tier Adviser     75.34%   (22)      -12.34%   (7)
No Adviser              -6.20%   (57)       60.08%   (20)
t-statistic (p value)   -1.410   (0.164)     1.069   (0.316)

Third Tier Adviser      11.56%   (16)      -53.73%   (6)
No Adviser              -6.20%   (57)       60.08%   (20)
t-statistic (p-value)   -0.280   (0.780)     2.559   (0.028)

                              Overall Wealth Destroyed

                            Completed            Withdrawn

First Tier Adviser      -114.94%   (17)       -70.75%   (4)
Second Tier Adviser     -137.71%   (15)       -79.30%   (1)
t-statistic (p-value)      0.464   (0.650)      0.136   (0.900)

First Tier Adviser      -114.94%   (17)       -70.75%   (4)
Third Tier Adviser      -125.52%   (22)      -119.33%   (3)
t-statistic (p-value)      0.190   (0.851)      1.357   (0.233)

First Tier Adviser      -114.94%   (17)       -70.75%   (4)
No Adviser               -69.40%   (69)       -63.84%   (15)
t-statistic (p-value)     -2.487   (0.018)     -0.218   (0.864)

Second Tier Adviser     -137.71%   (15)       -79.30%   (1)
Third Tier Adviser      -125.52%   (22)      -119.33%   (3)
t-statistic (p-value)     -0.171   (0.865)      0.902   (0.462)

Second Tier Adviser     -137.71%   (15)       -79.30%   (1)
No Adviser               -69.40%   (69)       -63.84%   (15)
t-statistic (p value)     -1.427   (0.174)     -0.263   (0.797)

Third Tier Adviser      -125.52%   (22)      -119.33%   (3)
No Adviser               -69.40%   (69)       -63.84%   (15)
t-statistic (p-value)     -1.027   (0.316)     -2.084   (0.105)

Table 10
Market Share Model with Lagged Wealth Ranking

The market share model Rank = [[alpha].sub.0] + [[beta].sub.0] Ability
to Complete + [[beta].sub.1] Lagged Wealth Rank + [[beta].sub.2]
Merger/Tender + [[beta].sub.3] Proportion of Acquisitions +
[[beta].sub.4] Average Deal Size + [[beta].sub.5] Time to Complete +
[[beta].sub.6] Attitude + [[beta].sub.7] SIC Compatibility + [epsilon]
is estimated using an OLS regression. The dependent variable in all
cases is the rank of the adviser (based on attempted deals),
calculated over the past 1, 2 or 3 years. The data used for this
regression is aggregated for each adviser on a per-annum basis.
Numbers in parentheses are heteroscedasticity adjusted standard errors
(White 1980). As there are a priori expectations for the coefficients
one tailed cut-offs are used to test for significance.

Dependent Variable              1 Year Rank   2 Year Rank   3 Year Rank

Regression estimates              1             2             3
Intercept                        19.275 ***    20.420 ***    12.020
                                 (7.132)       (7.743)      (10.646)
Slope coefficients on:
  Ability to complete           -10.967 ***    -5.101         1.419
                                 (4.091)       (4.719)       (8.114)
  Lagged wealth rank              0.720 ***     0.432 **      0.039
                                 (0.278)       (0.244)       (0.180)
  Merger/Tender                   1.590         0.294        -0.015
                                 (4.080)       (3.705)       (5.141)
  Proportion are Acquisitions     1.509       -19.290 **     -0.477
                                 (9.561)       (9.780)      (11.002)
  Average deal size              -0.025 ***    -0.007        -0.004
                                 (0.007)       (0.006)       (0.006)
  Time to complete                0.037 ***     0.022 **      0.022
                                 (0.012)       (0.010)       (0.022)
  Attitude                       -8.323 **     -2.316        -3.267
                                 (3.551)       (3.736)       (5.238)
  SIC compatibility             -12.118 ***    -4.570         1.419
                                 (4.633)       (4.371)       (8.114)
Observations                     44            49            32
F-statistic                       3.192 ***     1.416         0.322
Adjusted [R.sup.2]               28.50          6.35         -0.2043

Note: *** Significant at the 1% level;

** Significant at the 5% level;

* Significant at the 10% level;

a. Ability to complete: total number of deals completed, divided by
total number of deals attempted;

Lagged wealth rank: is the rank achieved by an adviser in the previous
year after being ranked on the measure described in section 5.7.1;

Merger/Tender: a proportion which divides the number of mergers and
tenders attempted by the adviser, by the total number of deals
attempted;

Proportion are acquisitions: the number times the adviser worked for
bidders, divided by the sum of deals attempted for bidders and deals
attempted for targets;

Average deal size, is the sum of the value of all deals attempted
(net of debt) divided by the number of deals;

Time to complete, is the average time taken from announcement of the
takeover until the deal is completed;

Attitude is the proportion of all deals attempted which have been
hostile in nature; and SIC compatibility is the ratio of the number of
times a deal has involved bidders and targets with the same 3-digit
primary SIC code, divided by the number of deals attempted.

(1.) Data sourced from Thomson Financial and reported in The Australian Financial Review, 7 January 2004, p. 37.

(2.) Statistics sourced from the New York Stock Exchange website, www.nyse.com, the Reserve Bank of Australia financial markets' statistics website http://www.rba.gov.au/Statistics/Bulletin/F07hist.xls, the AGSM's share price, price relative (SPPR) database and Kenneth French's website http://mba.tuck.dartmuth.edu/pages/faculty/ken.french/data_library.html

(3.) McLaughlin (1990) finds 80% of fees paid to advisers in the US are contingent upon the deals being completed. Rau (2000) reports that about 66% of all fees paid to investment bank advisers in tender offers are contingent on deal completion. The percentage falls to about 38% in mergers. Fees need not be disclosed in Australia since, under the Corporations Law, only expenses that are considered 'very material' need be disclosed. Fees paid to advisers are virtually never judged to be 'very material', thus no data are available. However, there is no reason to expect that the payment structure differs from the US.

(4.) For instance, the Australian Financial Review uses Thomson Financial Securities SDC data when reporting on activity in the takeover market. One example being its report on 14 February 2001 that the investment bank UBS Warburg topped the adviser table in 2000 in terms of value of deals on which it advised. Rau (2000) cites other prominent media reference to SDC data.

(5.) Studies that have used SDC data include Rau (2000), Allen, Jagtiani, Persistiani and Saunders (2001), Rau and Rodgers (2002).

(6.) It is not just the investment bank clients that use M&A league tables. Financial reporter, Brett Clegg, has observed that 'like the companies they advise, investment banks feel the brunt of global pressures, particularly from overseas head offices, which commonly use M&A league tables to assess the performance of an Australian branch office' (Australian Financial Review, 14 February 2001, Special Report on Mergers and Acquisitions, p. 3, emphasis in italics added).

(7.) Section 615(7) of the Corporations Law deems that a person may be presumed to potentially control a company with any holding that equals or exceeds 20% of issued shares.

(8.) SDC/Thompson confirm that the "standard" league tables use all deals (including asset sales > $100m) and only exclude equity car,,e-outs, exchange offers & open market repurchase.

(9.) ASX Journal was replaced by Shares Magazine in December 1996, hence between December 1996 and December 1998 Shares Magazine provided the required information.

(10) This number was calculated by downloading all takeover attempts contained in the Australian Mergers and Acquisitions database in SDC Platinum. This database covers the period from 1984 to current. The bidding firms were first ranked alphabetically, then ranked by announcement date. This was repeated for all bidders contained in the SDC database. Deals were included independent of their outcome, this is to ensure all attempted takeovers are included. This is important as the greatest amount of work initially is to price the offer correctly and sell it to the target. It is also in the initial stages where the decision to use an adviser is made.

(11.) If the bidder had no previous experience, then Log(Previous Acquisitions) is set to zero. For the bidders with experience, 1 is added to the logged value. Otherwise, bidders with no experience and bidders who had performed one takeover would have the same value (0).

(12.) Revision dates are not disclosed in SDC hence the number of revisions made or the return to the target from the revisions cannot be estimated.

(13.) This may also be a reflection of the database, as SDC confirmed the database was more extensive after 1995.

(14.) This, of course, assumes that investors' reaction to the takeover bid is highly correlated with managers' estimates of the NPV of the bid at the time they decide on whether to hire an adviser. As noted in earlier discussion, evidence on the long-run sharemarket performance of acquiring firms in the post-acquisition period arguably indicates that initial market reaction may not be an accurate or unbiased estimate of takeover NPV, however, for our purpose it is sufficient that the market reaction to announcement and managers' estimate of NPV at the time of hiring an adviser coincide.

(15.) Results from extending the event-window in regression 5 to day -10 to +10 are essentially the same.

(16.) These results are not reported but are available from the authors upon request.

(17.) It is possible that deals involving no advisers may underreport withdrawal rates. If this is the case, then the percentage is also overstated.

(18.) The analysis was repeated using a window of -5 to +5 to indicate whether a deal was considered to be 'good' or 'bad'. Similar results were obtained.

(19.) In the sample of 801 takeovers, 60.7% of all bidders were listed, accounting for 66.5% of cumulative deal values.

References

Allen, L., Jagtiani, J., Peristiani, S. & Saunders, A. 2001, 'The role of bank advisers in mergers and acquisitions', Working Paper, Zicklin School of Business, Baruch College, City University of New York.

Australian Financial Review, 2001, 'Special Report on Mergers and Acquisitions', Australian Financial Review, 14 February, p. 3

Australian Financial Review, 2004, 'Thomson Financial', The Australian Financial Review, 7 January, p. 37.

Benston, G. & Smith, C. 1976, 'A transactions costs approach to the theory of financial intermediation', Journal of Finance, vol. 31, pp. 215-31.

Bowers, H. & Miller, R. 1990, 'Choice of investment bankers and shareholders' wealth of firms involved in acquisitions', Financial Management, vol 19, pp. 34-44.

Brown, P. & da Silva Rosa, R. 1998, 'Research method and the long-run performance of acquiring firms', Australian Journal of Management, vol. 23, pp. 23-38.

Chemmanur, T. & Fulghieri, P. 1994, 'Investment bank reputation, information production and financial intermediation', Journal of Finance, vol. 49, pp. 57-79.

Maher, P. & Cooper, R. 1996, 'The new bulge bracket', The Investment Dealers" Digest, vol. 62, pp. 14-25.

McLaughlin, R. 1990, 'Investment-banking contracts in tender offers: An empirical analysis', Journal of Financial Economics, vol. 28, pp. 209-32.

Michel, A., Shaked, I. & Lee, Y. 1991, 'An evaluation of investment banker acquisition advice: The shareholders' perspective', Financial Management, vol. 20, pp.40-9.

Rau, P.R. 2000, 'Investment bank market share, contingent fee payments, and the performance of acquiring firms', Journal of Financial Economics, vol. 56, pp. 293-324.

Rau, P.R. & Rodgers, K.J. 2002, 'Do bidders hire top-tier investment banks to certify value?' Working Paper, January 2002, Krannert Graduate School of Management, Purdue University.

Servaes, H. & Zenner, M. 1996, 'The role of investment banks in acquisitions', Review of Financial Studies, vol. 9, pp. 787-815.

White, H. 1980, 'A heteroscedasticity consistent covariance matrix estimator and a direct test of heteroscedasticity', Econometrica, vol. 48, pp. 817-18.

Raymond da Silva Rosa ([dagger]) Philip Lee ([section]) Michael Skott ([double dagger]) Terry Walter ([dagger] [dagger])

([dagger]) UWA Business School, The University of Western Australia, 35 Stirling Highway, Crawley, WA 6009. Email: [e-mail]

([section]) School of Business, University of Sydney, Camperdown, NSW, 2006. Boston Consulting Group, Level 28, Chifley Tower, 2 Chifley Square Sydney NSW 2000.

([dagger] [dagger]) School of Banking and Finance, The University of New South Wales, Sydney, 2052 and Capital Markets Cooperative Research Centre Limited, Level 2, 9 Castlereagh Street, Sydney NSW 2000. Email: [e-mail]