167 a summary of merger performance studies in banking ... · and linder, 1993, which is a case...

167 A Summary of Merger Performance Studies in Banking, 1980–93, and anAssessment of the ‘‘Operating Performance’’ and ‘‘Event Study’’ Methodologies

Stephen A. RhoadesStaff, Board of Governors

The staff members of the Board of Governors ofthe Federal Reserve System and of the FederalReserve Banks undertake studies that cover awide range of economic and financial subjects.From time to time the studies that are of generalinterest are published in the Staff Studies seriesand summarized in theFederal Reserve Bulletin.

The following paper, which is summarized inthe Bulletin for July 1994, was completed duringthe winter of 1993–94. The analyses and conclu-sions set forth are those of the author and do notnecessarily indicate concurrence by the Board ofGovernors, the Federal Reserve Banks, ormembers of their staffs.

Board of Governors of the Federal Reserve SystemWashington, DC 20551

July 1994

A Summary of Merger Performance Studies in Banking, 1980–93, and anAssessment of the ‘‘Operating Performance’’ and ‘‘Event Study’’ Methodologies

Mergers reached record levels in the bankingindustry as well as in the industrial sector in thesecond half of the 1980s. The general economicconditions of the period and changes in theenforcement of the antitrust laws regardingmergers may have eased the way for somecombinations, but there is good reason to believethat the increased merger activity is likely topersist on its own and result in a restructuringof the industry.1

The effect of mergers on firm performance isthe subject of ongoing debate, and studies of thequestion have been growing in number. To assessthe current state of knowledge, the present workexamines the thirty-nine studies I found to havebeen published from 1980 to 1993 on the effectsof bank mergers on efficiency, profitability, orstockholder wealth.2 The first of these studiesappeared in 1983; most of them have beenpublished since 1987. This recent burgeoning ofresearch is reminiscent of the period around 1970.At that time, passage of the 1970 amendments tothe Bank Holding Company Act and liberalizationof bank holding company laws by many states,particularly those with unit banking laws, set off asubstantial increase in bank holding companyformations, acquisitions, and expansion; thatactivity in turn stimulated many studies of theperformance effects of bank holding company

affiliations and acquisitions.3 By 1980, however,the holding company movement had slowed, andthrough the mid-1980s, bank mergers generatedlittle research interest. Then another combinationof legislative and marketplace developments led toa resurgence of interest in the performance effectsof bank mergers.4

This overview is intended to determine whether,in the aggregate, the research since 1980 permitsany general conclusions regarding the performanceeffects of bank mergers. It is not intended to be astudy-by-study critique of the research. However,about half of the thirty-nine studies published inthe 1980–93 period used a fundamentally differentmethodology than the other half: Nineteen usedthe ‘‘operating performance’’ (or ‘‘observedperformance’’) approach, which observes thefinancial performance of a firm following amerger; and twenty-one used the ‘‘event study’’approach, which measures the reaction of thestock price of acquirers and targets to a mergerannouncement (one study used both methods).Hence, after presenting what, on balance, appearto be the conclusions represented by the entirebody of studies in the period, the present workconcludes with a broad assessment of the twomethodological approaches. In the appendix, themethodological details and results of each studyare summarized in a table, which is followed bythe bibliography of the thirty-nine studies.

Operating Performance Studies

The use of the operating performance (OP)methodology was concentrated in the last fewyears of the 1980–93 survey period, and by thattime OP was the methodology of choice for bank

Note. I thank Dean Amel and Timothy Hannan for helpfulcomments. I also thank Gregg Forte and Sherrell Varner forproviding editorial assistance, and Gwen White-Dubose andTony Feuerstein for bibliographic assistance.

1. For data on the changes in the U.S. banking structurethrough 1993 and a discussion, see Donald T. Savage, ‘‘Inter-state Banking: A Status Report,’’Federal Reserve Bulletin,vol. 79 (December 1993), pp. 1075–89.

The various forces affecting antitrust policy toward mergersin the 1980s are discussed in some detail in Stephen A.Rhoades and Jim Burke, ‘‘Economic and Political Foundationsof Section 7 Enforcement in the 1980s,’’Antitrust Bulletin,vol. 35 (Summer 1990), pp. 373–446.

For a discussion of the factors underlying the mergertendency and a projection of U.S. banking structure, seeTimothy H. Hannan and Stephen A. Rhoades, ‘‘Future U.S.Banking Structure: 1990 to 2010,’’Antitrust Bulletin, vol. 37(Fall 1992), pp. 737–98.

2. A summary of studies of the effects of bank mergers onoperating performance from the 1960s and 1970s is in Rhoades(1986).

The merger performance studies summarized in the presentstaff study are cited in the text and footnotes in short (author,date) form; the bibliography for them appears at the end of theappendix.

3. Details on the rapid growth of multibank holding compa-nies around this time can be found in Gregory Boczar,TheGrowth of Multibank Holding Companies: 1956–1973, StaffStudies 85 (Board of Governors of the Federal ReserveSystem, 1975).

Many of the studies are reviewed by Timothy J. Curry,‘‘The Performance of Bank Holding Companies,’’ inThe BankHolding Company Movement to 1978: A Compendium(Board of Governors of the Federal Reserve System, 1978),pp. 95–120. Also see Peter S. Rose,The Changing Structureof American Banking (Columbia University Press, 1987),especially chap. 8.

4. The main contributing developments were the removal oflegal restrictions on geographic expansion (both inter- andintrastate) and a substantial increase in bank mergers.

merger performance studies.5 The substantialincrease in the total number of merger perfor-mance studies during this period and the increasedinterest in the OP methodology probably reflectthe interest in cost cutting and efficiency in thebanking industry, particularly through merger,beginning in the late 1980s.6 And because theOP methodology permits the researcher to focusspecifically on costs and efficiency, it may appearto be an attractive approach. The general method-ology of the OP studies is to analyze changesin accounting profit rates or cost ratios, or both,from before merger to after. All but two OPstudies compare the performance of merging bankswith a control group of nonmerging banks (Craneand Linder, 1993, which is a case study of onemerger, and Frieder and Apilado, 1983, are thetwo exceptions). Most of the OP studies analyzeboth cost ratios and profit rates, although a fewanalyze cost ratios only; one study (also unique inother respects) analyzes profit rates only (Friederand Apilado, 1983).

Variations among OP Studies

In spite of the more or less common methodologyemployed in the OP studies, a great deal ofvariation exists among studies in sample coverage,geographic coverage, size of mergers examined,statistical tests, and so on. For example, samplesize varies from 1, in a case study by Crane andLinder (1993), to 4,900 (Peristiani, 1993b);nonetheless, most OP studies use fairly largesamples. In addition, many samples achieve moreor less nationwide coverage (for example, Srini-vasan and Wall, 1992; Spindt and Tarhan, 1991),whereas a few focus on certain regions of thecountry (for example, Linder and Crane, 1993).Most studies analyze mergers over several years;the earliest year covered is 1968 (Rhoades, 1986),and the most recent year is 1991 (Crane andLinder, 1993). Two studies, however, analyzemergers from only one year (Spindt and Tarhan,

1991; and Crane and Linder, 1993, which was theone-merger study).

Some studies focus on large mergers (forexample, Berger and Humphrey, 1992; Rhoades,1990), and others (such as O’Keefe, 1992) onsmaller mergers, but most analyze a rather widerange of merger sizes. Most studies analyze themerger performance of the target and acquirer as acombined entity (although some do not combinethe target and acquirer for measuring pre-mergerperformance); a few studies focus on the acquiringbank or the target bank, or both, and they do sofor the period before and after acquisition (forexample, Rose, 1987b; Spong and Shoenhair,1992). Studies have focused on performanceduring varying periods of time both before andafter merger. For example, Rose (1987b) analyzesas many as eight postmerger years; Linder andCrane (1993) analyze one postmerger year.

The studies conduct various kinds of statisticaltests. For example, some employ univariate t teststo compare performance ratios before and aftermerger and between acquiring, target, and non-merging firms. Other studies use multiple regres-sion analysis to control for other factors in testingwhether performance levels or changes can beexplained by whether the bank was or was notinvolved in a merger. Most studies of efficiencyperformance are based strictly on expense ratios,but three studies estimate translog productionfunctions to measure X-efficiency, scale efficiency,and an efficiency frontier for evaluating expenseratios, or efficiency ranking, of merged firms(Berger and Humphrey, 1992; DeYoung, 1993;Peristiani, 1993b). The many studies usingexpense ratios as an indicator of efficiency varyconsiderably in terms of the ratios used, amongwhich are total expenses to assets, noninterestexpenses to assets, revenues to employees, andtotal expenses to total revenues.

Main Findings of the OP Studies

Findings of the OP studies are generally consis-tent. Almost all the studies that find no gain inefficiency also find no improvement in profitabilityif they include both measures. In contrast, the sixstudies that show at least some indication of aperformance improvement do not obtain consistentefficiency and profitability results, or they areunique in some respect, or both. For example,the Frieder and Apilado study (1983) analyzesa profitability measure but not an efficiency

5. Of the nineteen OP studies appearing between 1980 and1993, fifteen appeared in the 1990–93 period; and of the fifteenmerger performance studies in 1992 and 1993, thirteen used theOP methodology. Actually, only fourteen papers on mergerperformance appeared in 1992 and 1993, but the Cornett andTehranian study (1992) used both methodologies.

6. An indication of the emphasis on efficiency can be foundin Business Week, ‘‘ Banking Gets Leaner and Meaner’’(Oct. 16, 1989), pp. 106–7; ‘‘ Banks Will Learn to Salt ItAway’’ (Jan. 8, 1990), p. 112–13; ‘‘ If Mergers Were Simple,Banking’s Troubles Might be Over’’ (April 22, 1991),pp. 77–79; and ‘‘ Two Banks Out to Tie the Knot—Or aNoose’’ (Oct. 7, 1991), pp. 124, 126.

2

measure, and the profitability measure is based ondifferences between actual and hypothetical netincome. Both Spindt and Tarhan (1991) andCornett and Tehranian (1992) find some improve-ment in return on equity resulting from mergerbut no improvement in return on assets or costefficiency. Spong and Shoenhair (1992) findevidence of an improvement in overhead costefficiency as a result of merger but no significantimprovement in return on assets or equity. Peris-tiani (1993b) finds some improvement in return onassets following merger but generally no improve-ment in cost ratios and efficiency measures.Finally, the Crane and Linder (1993) study of theFleet–Bank of New England merger finds areduction in noninterest expenses relative to assets,but the findings are not compared with a controlgroup.

In summary, despite the substantial diversityamong the nineteen OP studies, the findings pointstrongly to a lack of improvement in efficiency orprofitability as a result of bank mergers. Thesefindings are robust within studies, across studies,and over time.

Other Findings

The major purpose of the OP studies, as noted,is to assess the performance effects of mergers,typically by comparing merging with nonmergingbanks. Some of the studies, however, present moredetailed comparisons. Probably the most importantof these other comparisons focuses on in-market(horizontal) mergers. A widespread view is thatin-market bank mergers have the greatest potentialfor yielding efficiency gains because they providethe opportunity for closing overlapping (directlycompeting) offices as well as permitting, like othermergers, the combining of back-office operations,computer systems, and administrative functions.

Seven of the OP studies analyze the effect onperformance when merging firms have directlycompeting offices (Berger and Humphrey, 1992;O’Keefe, 1992; Srinivasan and Wall, 1992;Srinivasan, 1992; Peristiani, 1993a,b; Rhoades,1993). None of these studies finds that the elimi-nation of deposit market overlap as a result ofin-market mergers yields efficiency improvementsrelative to other firms. This may mean that manyof the directly competing offices that are closedfollowing such mergers, and the offices into whichthey are merged, are operating at or close enoughto minimum efficient scale that cost reductionsfollowing merger are too small to significantly

reduce cost ratios, such as expenses to assets. Itmay also mean that although closing officesreduces costs, it nonetheless causes a loss ofcustomers, which lowers assets or revenues; as aconsequence, efficiency indicators such as cost-to-asset and cost-to-revenue ratios remain steady oractually deteriorate.

Three studies examine the performance effectsof mergers involving internally acquired banks(that is, the merging parties were already ownedby the same bank holding company before merger)as well as externally acquired banks. Results aremixed regarding the effects on performance ofinternal as compared with external mergers(Spindt and Tarhan, 1991; Linder and Crane,1993; Peristiani, 1993a). Rose (1992) focuses oninterstate bank mergers and, in a comparison ofpre-merger and postmerger indicators, generallyfinds no improvement in operating efficiency,profitability, or market share.

Several studies that analyze the target bankrelative to the acquirer in terms of pre-mergerefficiency or profitability, or both, generally findthe target to be inferior to the acquirer (Spindt andTarhan, 1991; Crane and Linder, 1993; DeYoung,1993; Rhoades, 1993). Rose (1987b) finds,however, that acquired firms are more profitablethan acquirers. Two studies that compare theperformance of acquired and nonmerging banksfind that acquired banks perform no differentlythan nonmerging banks before merger (Rhoades,1986; Rhoades, 1990). Rose (1987b) finds thatacquired firms have lower returns than nonmerg-ing firms.

The issue of corporate control is not the mainfocus of these OP studies. To the extent that thestudies are relevant to the corporate control issue,they suggest that target firms tend to be average(not poor) performers, and that acquiring firmstend to be better performers than the target firms.The results are mixed, however, and only a fewstudies provide relevant findings.

Weaknesses of Some OP Studies

One potential weakness of most of the OP studiesis that they measure efficiency with some kind ofnoninterest expense ratio, most frequently nonin-terest expenses to assets.7 A possible problem withsuch a measure is that it does not account for

7. The discussion is intended to suggest weaknesses in theway some of the OP studies are done rather than to critiqueproblems inherent in the underlying methodology. Problemswith the methodology will be addressed later.

3

merger-related changes in product mix that wouldshift expenses from noninterest to interestexpenses, or vice versa. Thus, for example, amerger-related reduction in retail offices mayresult in a reduction in noninterest expenses toassets, but the merged firm may substitute higher-interest money market deposits for the low-interestcore, or retail, deposits generated at retail offices.As a consequence, noninterest expenses woulddecline, suggesting a fundamental improvementin efficiency, whereas in fact the rise in interestexpenses resulting from the change in product mixmay have offset the gain.

The possibility of significant changes in productmix following merger is not an inherently fatalflaw of OP studies, for two reasons. First, suchchanges could be accounted for when using a costratio by including product-mix ratios—such aslarge deposits to total deposits, and loans toassets—within a multiple regression model.Second, that major changes in product mix resultfrequently and uniquely from merger seemsunlikely. Thus, most merging banks plan branchclosings in such a way that they minimize the lossof retail deposits, which does not suggest a changein strategy in that area. Moreover, if a bank haddecided to shift from retail deposits to moneymarket deposits, one would have little reason tosuspect that a merger would be particularly usefulin achieving this purpose and, therefore, littlereason to believe that this shift would likely bea frequent outcome of mergers.

In view of these considerations, it is reasonableto analyze bank mergers for their effect onnoninterest expense ratios so as to assess theargument that mergers reduce noninterestexpenses; nonetheless, analyzing interest expenses,which account for about 70 percent of total bankexpenses, is also useful. Arguably the mostreliable studies are those that account for interestexpenses (or total expenses) as well as noninterestexpenses. All five studies that analyze a totalexpense ratio report finding no efficiency gain(Cornett and Tehranian, 1992; O’Keefe, 1992;Berger and Humphrey, 1992; DeYoung, 1993;Rhoades, 1993).

Overall, the OP studies provide substantialevidence that (1) bank mergers do not generallyyield performance improvement in terms ofprofitability or cost efficiency and (2) in-marketmergers do not have performance effects differentfrom those of other mergers. Some mixed evi-dence suggests that acquired firms tend to beaverage performers rather than poor performers.

Event Studies

Sixteen of the twenty studies conducted during the1980s on the performance effects of bank mergersused the event study methodology. The eventstudy was used much less frequently during theearly 1990s to examine bank merger performance,and in 1992 and 1993, only two of fifteen studiesused that methodology. The reason for the drop-offis unclear, but at least one explanation suggestsitself. That is that event studies are designed toindicate the financial market’s expectation as tothe overall performance results of mergers,whereas recent interest has focused on theefficiency effects of mergers.

The basic event study methodology is morestandardized across studies than is the OP method-ology. Essentially, the event study analyzes thestock return (based on price changes and divi-dends) of acquiring banks or target banks, or both,relative to a portfolio of stocks representing themarket. Differences in returns of the acquiringfirm or target firm relative to market returns areusually calculated over a period ranging from oneday to many days or weeks leading up to andfollowing the ‘‘ event’’ of the merger announce-ment. In making the calculation, the investigatorsseek to determine whether the announcement ofthe merger causes the stock return of the acquiringor target bank to perform differently than thegeneral market return for stocks. In event studies,differences in the stock returns between acquiringbanks or target banks and the market are used asestimates of ‘‘ abnormal’’ or ‘‘ excess’’ returns,using the following model:

(1) ARit = Rit − (ai + biRmt),

where

ARit = abnormal returns to bank stock i at time t

Rit = actual returns to bank stock i at time t

ai = ordinary least squares (OLS) estimate ofthe intercept of the estimated marketmodel

bi = OLS estimate of the market model slopecoefficient reflecting change in themarket return relative to the return forbank i

Rmt = actual returns to a market portfolio ofbank stocks at time t, as proxied by, for

4

example, the value–weighted index ofbank stocks from the Center for Researchon Security Prices (CRSP).

In equation 1, Rit and Rmt can be obtained fromvarious sources such as CRSP tapes. However, theparameters of equation 1, ai and bi, must beestimated from a market model, as follows:

(2) Rit = ai + biRmt + eit ,

where Rit , Rmt , ai , and bi are as defined above,and eit is the residual. With actual data on Rit andRmt , usually for many days or weeks around themerger announcement, equation 2 is estimatedusing OLS. The resulting estimated values for ai

and bi are substituted into equation 1 with data forRit and Rmt to calculate the abnormal return (ARit),usually for a fairly limited number of days aroundthe announcement date. These calculations indicatewhether the stock return of the acquiring or targetbank is greater than, equal to, or less than thereturn to the market portfolio of bank stocks.

Variations among Event Studies

The underlying procedures for estimating theperformance effects of mergers are more standard-ized in event studies than in OP studies. Neverthe-less, the event studies exhibit a great deal ofvariation with respect to sample size, number ofmerger announcements studied, period of timeover which the market model is estimated, periodof time over which abnormal returns are calcu-lated, and so on. For example,

• Although sample sizes are generally small,they range from 11 (Pettway and Trifts, 1985) to138 (Hawawini and Swary, 1990).

• The period of time over which market modelsare estimated varies substantially, from a low of41 days (Wall and Gup, 1989) to highs of239 days (Desai and Stover, 1985) and 108 weeks(Neely, 1987).

• The period of time over which abnormalreturns are calculated varies widely. Cornett andTehranian (1992) and James and Wier (1987a)present and analyze returns for only the announce-ment date and the day before, whereas Trifts andScanlon (1987) present results for −40 to+20 weeks around announcement, Dubofsky andFraser (1989) present data for −50 to +20 days,and Pettway and Trifts (1985) present and discussabnormal returns data for −10 to +50 days.

• Most event studies estimate the market modelover a period preceding the merger announcement,but some studies estimate the model for a periodthat extends from before the announcement toafter it (James and Wier, 1987a,b; Trifts andScanlon, 1987; de Cossio, Trifts, and Scanlon,1988; Baradwaj, Dubofsky, and Fraser, 1992);Cornett and De (1991) estimate the market modelfor a period strictly after the announcement date.

• The vast majority of studies use a standardmarket model to provide the basis for calculatingabnormal returns, but a few use a variant of thestandard market model. Thus, Sushka and Bendeck(1988), Dubofsky and Fraser (1989), andHawawini and Swary (1990) use a ‘‘ market-adjusted returns’’ or ‘‘ mean-adjusted returns’’model.

• Most market models are estimated andabnormal returns calculated with daily data onstock returns, but some studies use weekly returndata (Neely, 1987; Trifts and Scanlon, 1987;de Cossio, Trifts, and Scanlon, 1988; Wall andGup, 1989; Hawawini and Swary, 1990).

• Most studies calculate abnormal returnsaround the announcement date of the merger, butone study focuses on the acquisition date (Lobue,1984) and another on the date that the FederalReserve approved the merger (Sushka andBendeck, 1988).

• Finally, all the event studies analyze the effectof an announcement on returns to the biddingfirm, but only about one-half of the studies ana-lyze the effect on the stock return of the targetfirm.

Main Findings of Event Studies

Findings are not consistent across event studies.For example, seven studies find that a mergerannouncement had a significantly negative influ-ence on the returns to stockholders of the biddingfirm. Seven other studies find no effect on bidderreturns, three studies find positive effects, and fourfind mixed effects. The differences in findings arenot readily explicable from the differences inapproach noted above or from differences in theyears covered by the analyses. In contrast to thefrequently negative or neutral returns to biddersfrom merger announcements, eight of nine studiesthat analyze the merger announcement effect onthe target bank find a positive return to targetstockholders, and one study finds no abnormalreturn (De and Duplichan, 1987).

5

Four studies calculate the net wealth effect ofthe returns to bidders and targets combined(de Cossio, Trifts, and Scanlon, 1988; Baradwaj,Fraser, and Furtado, 1990; Hannan and Wolken,1989; Hawawini and Swary, 1990). One finds asmall net wealth effect, one finds no wealth effect,and two find net gains to certain types of mergerannouncements but not to others.

In summary, the event studies generally findthat stockholders of target firms have gains.However, the evidence regarding returns tobidders, as well as that regarding the net returns tobidders and targets combined, is too inconsistentto permit any clear conclusion. On balance, then,evidence from the twenty-one event studies,especially the ones since 1989, undercut thehypothesis that the financial markets expectmergers to improve bank performance.

Other Findings of Event Studies

Although the main focus of the event studies is onthe general announcement effect of bank mergerson returns to bank stockholders, several eventstudies present more detailed findings. Forexample, some find a difference in the direction(that is, positive vs. negative) of abnormal returnsbefore and after announcement (Pettway andTrifts, 1985). Other studies find differences inreturns for different types of merger proposals.For example, Sushka and Bendeck (1988) obtainresults for external proposals that differ from thosefor internal proposals; Baradwaj, Fraser, andFurtado (1990) find differences between hostileand nonhostile takeover bids; and Dubofsky andFraser (1989) find that returns to bidders beforea key 1981 court decision regarding the FederalReserve’s antitrust standards differ from returnsto bidders after the court decision.8

Abnormal returns in response to interstate andintrastate merger announcements are investigatedby several studies, but generally no difference isfound (Hannan and Wolken, 1989; Hawawini andSwary, 1990; Baradwaj, Dubofsky, and Fraser,1992; Cornett and Tehranian, 1992). In addition toinvestigating differences in abnormal returnsfollowing the announcement of different types ofmergers, or for different time periods, nearly halfthe event studies use multiple regression analysis,with abnormal returns as the dependent variable,

to analyze the determinants of abnormal returns.The determinants most commonly investigated arethe number of bidders and targets (for example,James and Wier, 1987a,b) and the relative size ofthe target and bidder (for example, Kaen andTehranian, 1989). Results are mixed.

Weaknesses of Some Event Studies

The event studies, like the OP studies, have someweaknesses.9 First, the period around theannouncement event for which abnormal returnsare analyzed varies greatly from study to study,and the results often appear to be sensitive to thetime period chosen. Significant positive abnormalreturns to the bidder during the day or twoimmediately around the announcement date arecommonly found. However, when abnormalreturns are cumulated for ten or twenty days afterthe announcement, the absence of significantabnormal returns is also common.

Second, because event studies require completestock price data, many studies analyze mergersinvolving only large, publicly traded banks asbidders or targets. Many merging banks (biddersor targets) are, of course, not publicly traded, soresults based on publicly traded stocks are notnecessarily representative of all bank mergers.

Third, the fact that in many studies the biddersare not bidding on the targets included in the samestudy or do not consummate a merger followingannouncement of a merger seems problematic.10

Overall, the results from the event studies aremixed. Stockholders in the target firms typicallygain, but stockholders in the bidding firms gener-ally experience negative abnormal returns or nosignificant abnormal returns. These results bythemselves do not provide much evidence ofefficiency gains from bank mergers. In addition,the fact that the stockholders of the typicallylarger bidding firms often suffer negative abnormalreturns around the time of announcement would initself seem to raise serious doubt about thelikelihood of any important efficiency gains frombank mergers. Or perhaps, because the abnormalreturns surrounding announcement do not, asnoted, focus strictly on efficiency effects, anyefficiency effects are being masked by otherfactors. For example, managerial hubris regarding

8. Mercantile Texas Corp. v. Board of Governors, 638 F.2d1255 (5th Cir. 1981), and Republic of Texas Corp. v. Board ofGovernors, 649 F.2d 1026 (5th Cir. 1981).

9. Problems inherent in the underlying methodology of thesestudies will be discussed later.

10. An exception is De and Duplichan (1987). That studyexcludes mergers that were not consummated afterannouncement.

6

the ability to improve the target after merger mayresult in overbidding, which affects the returns tobidders around announcement. Given their mixedresults and some of the weaknesses just noted, theevent studies appear to provide little support forthe hypothesis that bank mergers generally resultin efficiency improvements.

Overall Assessment of MergerPerformance Studies

The nineteen operating performance (OP) studiesprovide consistent evidence that bank mergershave not generally resulted in efficiency gains.The twenty-one event studies yield mixed resultsshowing generally positive abnormal returns tostockholders of targets and negative or no abnor-mal returns to stockholders of bidders followingannouncement of a merger. Even a simple weigh-ing of the two sets of results would lead onetoward an overall conclusion that bank mergersdo not generally tend to result in efficiency gains.One is led even more strongly in that directionupon taking into account the weakness of theevent study methodology relative to the OPmethodology as a means of studying the efficiencyeffects of bank mergers. This conclusion from afairly large body of empirical evidence is, interest-ingly, consistent with the views of a number ofbank analysts who were interviewed on this issuein 1991. They generally reported that in their ownexperience with bank mergers, most of them havenot resulted in efficiency gains.11

Shortcomings of Event Studies

The usefulness of event studies relative to OPstudies is seriously undermined by two factors.First, and least subject to debate, is that thefinancial market response to a merger announce-ment, in terms of abnormal returns to stockhold-ers, reflects expectations about all the elements(not only efficiency) that may influence the generalperformance results of a merger as well as differ-ences in expectations between investors andbidders. For example, abnormal returns shouldreflect the market expectations about market

power, or position, gains as well as efficiency.The returns should also reflect differences injudgment between bidders and investors regardingthe potential gains from a merger and the appro-priate purchase price. Because bidders are insidersto the deal and investors are outsiders, presumablywith less information, and because a bidder’smanagement may be overcome with hubriswhereas outside investors would not be so influ-enced, a great deal of room exists for mistakes tobe made or differences in judgment to arise. Allthese factors will be reflected in the stock returns.Thus, it is highly questionable to interpret mergerannouncement effects strictly with respect tooperating efficiency. In contrast, OP studies canfocus directly on the efficiency effects of mergersand thus yield results that are more directlyrelevant to the merger efficiency question.

Second, event studies are based on short-termmovements in stock prices. Short-term movementsin stock prices may reflect speculation by sophisti-cated investors who seek short-term trading gainsby outguessing other sophisticated market play-ers.12 If an investor trades in and out of a stock toachieve such short-run gains or for hedgingpurposes, the long-term performance of the firm isof little relevance to the investor. And because aninvestor may move quickly in and out of a stock,the investor’s performance and reputation (in thecase of a brokerage firm) is not necessarilyinfluenced by the long-term performance of astock. Consequently, to the extent that stock pricechanges surrounding a merger announcement eventreflect short-run trading, as opposed to long-terminvestments, abnormal returns would appear to beof limited use for assessing the performanceeffects of mergers.13

The possibility that a large volume of stockpurchases each day are not made primarily aslong-term investments is suggested by the substan-tial increase in trading volume and turnover ofshares during the past twenty years or so. Byalmost any measure, stock trading has increased in

11. See Stephen A. Rhoades, ‘‘ The Efficiency Effects ofBank Mergers: Rationale for a Case Study Approach andPreliminary Findings,’’ in Proceedings of the 29th AnnualConference on Bank Structure and Competition (FederalReserve Bank of Chicago, May 1993), pp. 377–99; the inter-views are reported on pp. 379–80.

12. Such speculative activity has been discussed at length inAnthony Bianco, ‘‘ Playing with Fire,’’ Business Week (Septem-ber 6, 1985), pp. 78–90. Concern about increased short-term,speculative trading has led to proposals for a securities transac-tion tax, as noted in Craig S. Hakkio, ‘‘ Should We Throw Sandin the Gears of Financial Markets?’’ Federal Reserve Bank ofKansas City, Economic Review, vol. 79 (Second Quarter 1994),pp. 17–30.

13. Even if investments in the stocks of merging firms werebased on expectations about firms’ future performance, giventhe notorious frailty of economic projections, it would notappear that expectations (as reflected in abnormal returns)provide a very reliable foundation for assessing performanceeffects of mergers.

7

terms of the turnover of stocks. For example,between 1970 and 1992, the average number ofshares traded daily on the New York StockExchange (NYSE) increased more than seventeen-fold, from 11.6 million to 202.3 million. Over thesame period, the average number of shares listedon the exchange rose only about sevenfold, from15,573.7 million to 107,730.5 million. Also, the‘‘ turnover rate’’ of stocks on the NYSE more thandoubled, from 0.19 to 0.48, between 1970 and1992.14 Finally, the increased transmission speedand reduced cost of information resulting fromtechnological change have provided the opportu-nity for the continued growth of short-termtrading.

Shortcomings of OP Studies

The OP studies have the advantage of focusing onactual observed operating results of a mergerrather than the expectations around the announce-ment date. The OP studies have, in other words,the advantage of ‘‘ hindsight over foresight,’’ asstated by Caves in a detailed analysis of theissue.15 Nevertheless, two possible problems areinherent in the OP methodology. First, the OPstudies typically analyze operating performance forperiods of one to six years after a merger occurs.During these years, many factors unique to themerged firm, other than the merger itself, mayaffect the firm’s efficiency or general performance.And, over a longer period, these other factors arepotentially a more serious problem, although itdoes not appear to be necessary to study thepostmerger performance of banks for an extendedperiod to gain an understanding of the perfor-mance effects.16 These other factors might includeproduct mix changes or other mergers. However,other factors can be explicitly taken into accountby sample design (for example, excluding mergersthat are followed by other mergers) or by includ-ing appropriate variables in the multiple regressionmodels often estimated in OP studies.

But not taking other factors into account is notnecessarily a serious problem. For example, if weaccept the proposition that bank mergers in factgenerally improve the efficiency or generaloperating performance of merged firms, the failureof OP studies to find the improvement in theseveral years after the merger would imply thatefficiency and general performance gains frommergers are somehow systematically squandered.Thus, there would be no lasting efficiency orgeneral performance benefits from mergers.However, if we have no reason to believe thatmerging banks would generally undertake actionsafter merger that would systematically negatewhatever general performance or efficiency gainsrelative to other banks they might have achievedvia merger, then the findings of OP studies aretelling even if they do not carefully account forproduct mix and other changes following merger.

A second possible problem inherent in OPstudies of efficiency (usually measured by expenseratios), and especially of general performance(often measured by return on assets), is that theseare accounting measures rather than economicmeasures.17 At a conceptual level, such criticismhas merit. At a practical level, however, it sug-gests that no readily available data exist withwhich to calculate the performance of Americanbusinesses and that the millions of pages ofaccounting data provided in myriad forms tomanagers, investors, and the government are oflittle use for measuring cost efficiency or profit-ability. Apparently, however, American businesses,investors, and government do find such data usefulfor making decisions and investments and allocat-ing resources. In fact, return on assets is oftenselected by market analysts as a good measure ofa bank’s overall performance.18 Bank regulators,who have a great interest in the financial sound-ness of banks, rely heavily on return on assets andother accounting measures to assess the perfor-mance of banks. Under the circumstances, usingaccounting measures to assess the efficiency andgeneral performance of merging banks appears tobe reasonable.

14. The turnover rate is the ratio of the volume of sharestraded in a year to the average number of shares listed duringthe year. See New York Stock Exchange, Fact Book: 1992Data (1993), pp. 82–3.

15. Richard E. Caves, ‘‘ Mergers, Takeovers, and EconomicEfficiency: Foresight vs. Hindsight,’’ International Journal ofIndustrial Organization, vol. 7 (March 1989), pp. 151–74.A brief assessment of these differences between the two typesof studies is in Rhoades (1986).

16. According to a number of bank analysts, all gains froma bank merger should generally be realized within three years,with 50 percent of any gains coming after one year (Rhoades,‘‘ The Efficiency Effects of Bank Mergers,’’ cited earlier).

17. For criticisms of the use of accounting data in measuringprofits, see Franklin M. Fisher and John J. McGowan, ‘‘ On theMisuse of Accounting Rates of Return to Infer MonopolyProfits,’’ American Economic Review, vol. 73 (March 1983),pp. 82–97; and George J. Benston, ‘‘ The Validity of Profits-Structure Studies with Particular Reference to the FTC’s Lineof Business Data,’’ American Economic Review, vol. 75(March 1985), pp. 37–67.

18. See, for example, the following issues of Business Week:April 18, 1977, p. 97; April 9, 1984, p. 83; and April 8, 1985,p. 106.

8

On balance, the problems inherent in the eventstudy methodology for studying the effects of bankmergers appear to be substantially more trouble-some than those inherent in the OP methodology.One is therefore justified in giving greater weightto the findings of the OP studies, which are thatbank mergers generally do not deliver gains inefficiency or general operating performance.19

Summary and Conclusion

The thirty-nine studies of the effects of bankmergers on efficiency, profitability, or stockholderwealth that appeared from 1980 to 1993 providelittle support for the view that bank mergers resultin improvements in performance. The nineteenoperating performance studies indicate, with veryfew exceptions, that bank mergers do not yieldimprovements in efficiency or profitability and thatin-market mergers do not have more favorableeffects on performance than other mergers. Thetwenty-one event studies (one study included bothmethodologies) generally indicate that net gainsaccrue to stockholders of target firms after theannouncement of a merger. Evidence on returns tobidders, and net returns to bidders and targetscombined, is mixed and thus does not provide anyclear evidence of performance improvements frombank mergers. In view of the shortcomings of the

event study methodology in this context, moreweight should be accorded the operating perfor-mance studies in evaluating the overall evidence.

Two caveats are to be noted. First, findingsfrom cross-sectional statistical studies such asthose examined here allow us to conclude that,in general, bank mergers do not lead to improve-ments in efficiency or profitability. However, thisdoes not mean that mergers never yield suchimprovements. Second, almost all the mergersanalyzed in these studies were undertaken before1989, and mergers after that time might yielddifferent results.

Finally, the findings by economists that bankmergers do not generally result in efficiency gainsare not necessarily inconsistent with the argumentby some bankers that mergers achieve significantcost cutting. Both groups may be right. However,they are usually talking about different concepts.The difference may arise from the fact thateconomists focus on the efficiency effects ofmergers, which are typically measured by anexpense ratio such as total expenses to totalassets.20 In contrast, bankers typically focus on thedollar volume, or percentage, of costs that will becut. If dollar costs are cut following a merger butassets or revenues decline more or less proportion-ately (as a firm shrinks after merger, which is notuncommon), the banker would rightly claim thatcosts were cut and the economist would rightlyclaim that efficiency, in terms of costs to revenuesor assets, did not improve. Nonetheless, from thestandpoint of public policy considerations and thereal long-term performance of the industry, anefficiency measure is the relevant benchmark.

19. One might argue that even if performance does notgenerally improve following a merger, mergers neverthelesshave positive effects on efficiency and performance, and theperformance of one or both firms involved in the merger mighthave deteriorated without the beneficial effects of the merger.First, what would have happened had a merger not occurred isimpossible to know. Second, the argument proposes theunlikely generalization that merging firms would experiencedeteriorating efficiency and general performance relative toother firms were it not for the merger. This argument mayapply to a few mergers, but it provides a highly improbableexplanation for the overall findings of the merger performancestudies.

20. Moreover, a meaningful assessment of the cost, effi-ciency, or general performance results of a merger mustanalyze changes in relation to a control group to account forgeneral industry trends in performance.

9

Appendix: Summary and Bibliography of Studies

Studies of the Effects of Bank Mergers, 1980–93

Study Years studied

Number ofmergers orannounce-

ments1 Characteristics of sample Methodology2

Frieder andApilado (1983)

1973–77 106 106 affiliates of 4 large BHCs in alarge southern state. Lead banksof the BHCs are excluded. Uniquesample and data that cannot bereplicated

Unique methodology involvinghypothetical measures of keyvariables. Compares actual earningsper share of a BHC that made anacquisition with hypotheticalearnings per share of the BHCconstructed on the assumption thatan acquisition did not take place.Affiliates of the BHCs have beenaffiliated for 1 year or more,providing varying postacquisitionperiods. Uses differences betweenhypothetical and actual adjusted netincome measures to estimatemagnitude of any merger effect.Analyzes effect on combined entity.Presumably the affiliates analyzedwere established banks acquired bythe BHC rather than de novo banksformed and ‘‘acquired’’ by theBHC, but this is not clear

Lobue (1984) Not noted 37 Stock of acquiring firms is tradedon the NYSE or OTC. Six of theacquisitions are of nonbank firms.Sample includes only acquiringfirms

Event study. Analyzes abnormalreturns on the stock of the acquiringfirm relative to the market return.Calculation of abnormal returns isbased on a log-transformed standardmarket model estimated over anunspecified period. Abnormalreturns relative to the market areanalyzed for 24 months before, to18 months after, the effectiveacquisition date. Returns areanalyzed for different groupings offirms based on size of mergingfirms, state branching laws, etc.

Desai andStover (1985)

1976–82 18 Acquiring firms are identifiedfrom Moody’s Bank and FinanceManuals and other sources andare listed on the CRSP dailyreturn files. They have made noacquisitions for at least 2 yearsbefore 1976. Analysis focusesonly on the acquiring firm

Event study. Analyzes cumulativeabnormal returns, relative to themarket, for a 2-day period includingthe event day and 1 day after.Returns are measured only for theacquiring firm. Analysis is based onstock market returns to the marketand the acquiring firm using astandard market model estimatedover the period from 270 days to31 days before announcement date.Analyzes stock market reaction toboth the announcement and FederalReserve approval

Note. In this table, the studies appear in chronological orderby date of publication. Two works not covered by the presentstudy analyze the efficiency effects of simulated rather thanactual mergers or merger announcements. See Donald T.Savage, ‘‘Mergers, Branch Closings, and Cost Savings’’(Board of Governors of the Federal Reserve System, 1991);and Sherrill Shaffer, ‘‘Can Megamergers Improve BankEfficiency?’’ Journal of Banking and Finance, vol. 17(April 1993), pp. 423–36. Savage finds declines in efficiencybased on large in-market mergers; Shaffer finds that aboutone-half of simulated large mergers would reduce efficiencyand the other half would increase efficiency.

AMEX American Stock ExchangeBHC Bank holding companyCRSP Center for Research on Security PricesFDIC Federal Deposit Insurance CorporationNASDAQ National Association of Securities Dealers

Automated QuotesNYSE New York Stock ExchangeOLS Ordinary least squaresOTC Over the counterS&P Standard and Poor’sWSJ Wall Street Journal

10

Hypothesis or issueaddressed

Performancemeasures 3 Type of test 4 Findings

Hypothetical effect ofmergers on earnings pershare

Hypothetical measuresof earnings per shareand adjusted netincome, with andwithout acquisition

Univariate t tests BHC bank affiliates have a positive effect onhypothetical BHC profitability each year afteracquisition, from 1 year to 8 or more years

Wealth effects Abnormal returns onstock

Residual analysis Acquiring firms, on average, have positivecumulative abnormal returns


Residual analysis Acquiring firms have positive significant cumula-tive abnormal returns on the announcement dayand 1 day after

The size of the target relative to the acquirer doesnot affect the return to the acquirer

Acquiring firms have positive significant cumula-tive abnormal returns during the 2-day eventwindow associated with approval of the acquisi-tion by the Federal Reserve Board

1. For event studies, may include merger announcements formergers never consummated.

2. In event studies, a standard market model refers to theregression model, Rit = ai + bi Rmt, used to estimate thecoefficients, a and b, needed to compute the abnormal return,ARit = Rit − (ai + bi Rmt). See text p. 4 for details.

3. Measures that reflect performance in terms of profitability,efficiency, or abnormal returns to shareholders. Does notinclude measures such as loans/assets or capital/assets.

4. In event studies, residual analysis and prediction erroranalysis are treated as synonymous.

11

Studies of the Effects of Bank Mergers, 1980–93—Continued

Study Years studied


ments 1 Characteristics of sample Methodology 2

Pettway andTrifts (1985)

1972–81 11 Merged firms are all failing banksthat the FDIC assisted in apurchase and assumption merger.Acquiring firms are listed on amajor exchange or OTC, are onCRSP tapes, and are frequentlytraded. Sample includes only theacquiring firms

Event study. Analyzes abnormalreturns on the stock of the acquiringfirm relative to the market. Analysiscovers from 10 days before mergerto 50 days after. Prediction errors orabnormal returns for individualfirms are calculated using a standardmarket model estimated over theperiod from 60 days to 10 daysbefore merger. (The merger and itsannouncement occur simultaneouslyin FDIC-assisted mergers.) Theresiduals are not cumulated by theusual cumulative abnormal returnsmethod. Instead, an averagegeometric residual return iscalculated

Rhoades (1986) 1968–78,even years

413 Banks acquired by BHCs. Alsoincludes about 3,600 nonacquiredbanks located in the same marketsas acquired banks and withoperations in no other markets

Analyzes operating performancechanges. Compares average perfor-mance during 3 years before with4th–6th years after acquisition ofacquiring and nonacquiring firms.Analysis focuses on the acquiredbank rather than the combinedentity both before and afteracquisition

De andDuplichan(1987)

1982–85 28(22 targetsand 24bidders)

Strictly interstate mergers thatwere ultimately consummated.Bidders and targets have stocktraded on a major exchange orOTC and have more than$100 million in assets. Acquisitionannouncements by grandfatheredbanks or BHC are excluded

Event study. Analyzes abnormal andcumulative abnormal returns,relative to the market, of acquiringand acquired firms for up to30 days before and after theannouncement date. Analysis isbased on a standard market modelestimated with weekly return datafor 52 weeks before the announce-ment date. Return data are fromCRSP, the S&P Stock Price Record,and the S&P 500 composite index

12



Wealth effects Average geometricresidual return

Residual analysis Acquiring firms have positive significant averageabnormal returns during 10 days before merger

Negative significant average abnormal returns areexperienced during 50 days after merger for 11FDIC-assisted purchase and assumption mergers;positive returns occur for 3 days after merger

Performance changes Net income/totalassets

Noninterest expenses/total assets

Multiple regression(staff study)

Probit analysis(chapter)

Efficiency does not improve after merger com-pared to that of nonacquired firms

Profit rates do not improve after merger relative tothose of nonacquired firms

Before merger the profitability and efficiency ofacquired firms are no different from those ofnonmerging firms

Growth of market share of merged firm does notimprove after merger when compared with that ofnonacquired firms


Residual analysis Neither acquiring nor acquired have significantabnormal returns during the 30 days beforeannouncement or after

13


Study Years studied



James and Wier(1987a)

1973–83 79 19 mergers resulting from FDICfailed-bank auctions in which thewinning bidder has actively tradedcommon stock and 60 otherrandomly selected unassistedacquisitions in which the bidderhas actively traded stock

Event study. Analyzes abnormalreturns of the acquiring firm,relative to the market, for differentperiods, with emphasis on 2days—the announcement day andday before. In the case of the19 failed-bank auctions, analysisfocuses on stock returns of winningbidders. Analysis is based on stockmarket returns to the market andacquiring firms using a standardmarket model that accounts for risk.The model is estimated for 80 daysto 11 days before announcementand 11 days to 80 days after. Alsouses OLS to analyze differences inmarket reaction to assisted andunassisted mergers and the effect ofthe number of bidders and potentialbidders on abnormal returns

James and Wier(1987b)

1972–83 60 Random sample of acquiringbanks with actively traded stocksthat are in the Compustat Bankingand Finance File. Acquiring firmsthat made other acquisitions in the6 months before acquisition areexcluded

Event study. Analysis of abnormalreturns on the stock, relative to themarket, of acquiring banks coversthe 2-day and 5-day periods up tothe announcement date. Analysis ofabnormal returns is based on astandard market model estimatedover the period from 80 days to15 days before announcement and15 days to 80 days after. Also usesOLS to examine the effect, onreturns to the acquirer, of thenumber of bidders, number oftargets, and the ratio of size ofacquired firm to the size of thetarget. Finally, examines relationbetween abnormal gains andattainment of concentration andmarket share

Neely (1987) 1979–85 26 Stock of acquiring and acquiredbanks is traded on NYSE, AMEX,or OTC. Acquirers are BHCs, andno regulatory mergers areincluded

Event study. Analyzes abnormalreturns, relative to the market, foracquiring and acquired firms duringthe period from 10 weeks before to30 weeks after the acquisitionannouncement. Analysis is based ona standard market model estimatedover the period from 119 weeks to11 weeks before merger. S&P dataare used to construct the marketreturn index for banks. Attempts todistinguish between acquisitions ofbanks and of BHCs in terms ofstock price effect

14




Residual analysis

Multiple regression

Acquiring firms in FDIC-assisted and unassistedmergers have positive significant abnormal returnson the day before announcement and the day ofannouncement

No pattern for returns to acquirers is apparent forthe 50 days after merger

The number of bidders has a significant negativerelation with abnormal returns to acquirers

For unassisted mergers (but not for assistedmergers), the number of potential bidders isnegatively related, and the number of alternatetargets is positively related, to abnormal returns toacquirers


Residual analysis

Multiple regression

Acquiring firms have small but significant positivereturns during the 2-day and 5-day periods up tothe announcement date. Returns after announce-ment are not examined

The number of targets is positively related toabnormal returns to acquirers, and the number ofpotential bidders is negatively related to acquirers’returns

Returns to acquirers are positively related to ratioof target to bidder size

Attainment of market share and concentration byacquirers are minimal, so achievement of marketpower is not an important source of observedgains


Residual analysis Acquiring firms generally have normal returns

Acquired firms have positive significant cumula-tive abnormal returns before announcement andafter

15


Study Years studied



Rose (1987a) 1970–85 106 106 merged banks plus 106nonmerging banks that werematched, in terms of size andgeographic market, with themerged banks. Wide variation insize and geographic location

Analyzes operating performancechanges. Compares average perfor-mance of merging and nonmergingfirms both before and after merger.Analysis covers the 1-year and3-year periods before merger andvarious periods from 1 year to8 years after merger. Acquiring andacquired firms are treated as acombined entity. Also reportsfindings of a survey of mergingfirms that asked respondents tospecify motives for mergers

Rose (1987b) 1970–80 40 acquiringbanks,138 acquiredbanks

Acquiring banks are each pairedwith a nonmerging bank ofsimilar size and location. Mergershave wide geographic representa-tion. All acquirers are nationalbanks. The acquired banks areindependently selected in thesame manner as the acquiringbanks and are not necessarilyacquired by the acquiring banks inthe sample

Comparison of operating perfor-mance differences. All comparisonscover each of the 5 years beforeand after merger. Analyzes differ-ences between merging and non-merging banks after merger. Alsoanalyzes differences betweenmerging and nonmerging banks inchanges from pre- to postmergerperiod

Trifts andScanlon (1987)

1982–85 17 acquiredand14 acquiring

All cases involve interstatemerger. Each acquiring andacquired bank has stock traded onNYSE, AMEX, or OTC. Sampleincludes both acquiring andacquired firms from 11 mergersand a total of 17 acquired banksand 14 acquiring banks

Event study. Analyzes abnormalreturns on the stock price plusdividends, relative to the market, ofacquiring and acquired firms for40 weeks before announcement and20 weeks after. Analysis is based ona standard market model estimatedover the period from 61 weeks to41 weeks before announcement and21 weeks to 41 weeks after.Alternative market indexes of bankswere used in the market models(S&P 500 and an index of bankstocks based on S&P data) andyielded the same results

16



Performance changes Return on assets

Return on equity

Univariate t tests Returns on assets and equity do not improve aftermerger when compared with those of nonmergingfirms

Returns on assets and equity of combined mergingfirms and nonmerging firms do not differ beforemerger

Growth in assets and deposits do not improveafter merger when compared with that of non-merging firms

Market share of merged firms in local marketsdoes not improve after merger when comparedwith that of nonmerging firms

Growth in net earnings after tax does not improveafter merger when compared with that of non-merging firms

Performance effects Return on assets

Return on equity

Operating revenue/operating expenses

Revenues/employees

Assets/employees

Paired comparisonwith t tests

Multiple regression

Canonical analysis

Operating efficiency (operating revenue/operatingexpenses) of acquiring firms does not improveafter merger when compared with that of non-merging firms

Employee productivity (assets/employees) ofacquiring firms does not improve after mergerwhen compared with that of nonmerging firms

Profitability of acquiring firms does not improveafter merger when compared with that of non-merging firms

Multiple mergers do not effect performance

Acquired firms have lower rates of return thannonmerging firms before merger

Acquired firms are more profitable than acquirersbefore merger

Acquiring firms have lower operating efficiencyand employee productivity than nonmergeringfirms


Residual analysis Acquiring firms have negative significant cumula-tive abnormal returns during the 20 weeks afterannouncement but no cumulative abnormal returnsfor the 40 weeks before announcement or for theentire 60-week period

Acquired firms have positive cumulative abnormalreturns for the 40 weeks before announcement andthe entire 60-week period

No abnormal returns to acquired firms for the20 weeks after announcement

Returns differ substantially among mergerproposals

Returns tend to be larger for merger proposals inwhich the size difference between acquiring andacquired firms is relatively small

17


Study Years studied



de Cossio,Trifts, andScanlon (1988)

1982–85 41 intrastate,21 interstate

Stock of acquiring and acquiredbanks is traded on NYSE, AMEX,or OTC. Excludes banks makingmultiple mergers. 18 of 41intrastate mergers include bothbidder and target firms. 11 of 21interstate mergers include bothbidder and target firms

Event study. Analyzes the abnormalreturn on the stock, relative to themarket, of acquiring and acquiredfirms for 30 weeks beforeannouncement and 20 weeks after.Analysis is based on a standardmarket model estimated over theperiod from 50 weeks to 31 weeksbefore announcement and 21 weeksto 40 weeks after. Analyzes bothinter- and intrastate announcements

Sushka andBendeck (1988)

1972–85 41 Stock of acquring banks is listedon NYSE or AMEX. Sampleincludes only acquiring firms

Event study. Analyzes abnormalreturns on the stock of the acquiringfirm relative to the market return.Calculation of abnormal returns isbased on a mean-adjusted returnsmodel and on a standard marketmodel, and each yields the sameresults. The adjusted returns modelis based on the period from340 days to 120 days before FederalReserve approval. The analysiscovers 5 days to 14 days beforemerger approval by the FederalReserve and 1 day to 14 days aftermerger approval. Time periodassessed seems to vary. Does notinvestigate effects on acquiredfirms. Analysis examines four typesof merger: emergency, nonemer-gency, external, and internal

Baradwaj,Fraser, andFurtado (1990)

1980–87 23 hostile,30 nonhostile

Acquiring or target bank ispublicly traded. One-third of thehostile offers were successful.Hostile offers based on reports ofunsolicited bids in the WSJ from1980 to 1987. The 30 nonhostilemergers serve as a control groupand were drawn from anotherstudy

Event study. Analyzes abnormal andcumulative abnormal returns ofacquiring and acquired firms forvarious periods from 5 days beforeto 5 days after announcement.Compares abnormal returns forhostile bids with those for non-hostile bids. Analysis is based on astandard market model estimated for150 days (from day 210 to day 61)before the announcement. Alsocompares characteristics of bankssubject to hostile bid with thosesubject to nonhostile bid

18




Residual analysis Acquired firms experience positive significantabnormal returns over the entire period for bothinter- and intrastate announcements

Acquiring firms do not have abnormal returnsover the entire period for either inter- or intrastatemerger proposals

Combining the dollar value of effects on acquiringand acquired firms shows net gains for intrastatemerger proposals and no net gain or loss forinterstate proposals

Merger proposals involving large targets (relativeto bidders) show greater gains than proposalsinvolving relatively small targets


Prediction erroranalysis

Acquiring firms have negative significant abnor-mal returns in the 23 external merger proposalsbut generally normal returns for other kinds ofproposals during the period surrounding FederalReserve approval of a merger proposal

Merger proposals involving banks internal to aBHC have no significant abnormal returns

Merger proposals involving banks external to theBHC have negative significant abnormal returns


Residual analysis Acquired firms have positive significant abnormalreturns for various periods from 5 days beforeannouncement to 5 days after

Acquiring firms have negative significant abnor-mal returns during the period from 5 days beforeannouncement to 5 days after

Hostile bidders have a smaller negative returnthan nonhostile bidders

Targets of hostile bids have significantly higherreturns than targets of nonhostile bids

Generally little difference in returns betweensuccessful and unsuccessful hostile bidders andbetween successful and unsuccessful hostiletargets

Estimates of net dollar value of wealth effects forbidders and targets combined indicate a positivesignificant net wealth effect from hostile takeoverannouncements and no significant net wealtheffect from nonhostile announcements

19


Study Years studied



Bertin,Ghazanfari,and Torabzadeh(1989)

1982–87 33 Acquiring firms are all involvedin an FDIC-assisted purchase andassumption of a failing bank.Analysis focuses only on acquir-ing firms. Stock of all firms issubject to continuous daily tradingon an exchange

Event study. Analyzes abnormalreturns on the stock of acquiringfirms relative to the market. Analy-sis of abnormal returns covers20 days before announcement to20 days after, using stock price anddividend data from the WSJ. Alsoconducts OLS analysis of determi-nants of abnormal returns. Analysisis based on a standard marketmodel estimated over the periodfrom 121 days to 21 days beforeannouncement

Dubofsky andFraser (1989)

1973–83 101 Acquiring banks are selected onthe basis of announcementsappearing in the WSJ, and theirstock must be actively traded(on an exchange or OTC) on theannouncement date. Merger mustinvolve a target bank that is atleast one-tenth the size of theacquiring bank in terms of assets.Excludes combinations of equals

Event study. Analyzes abnormalreturns on the stock of the acquiringfirms during the 2 days covered bythe announcement day and the daybefore announcement. Analysis isbased on the market-adjustedreturns method rather than astandard market model. Focusesonly on acquiring firms. Focuses ondifferent announcement effectsbefore and after various dates. Alsoexamines excess returns for variousperiods up to 50 days beforeannouncement and 20 days after formergers before and after the date(July 1, 1981) when two DistrictCourt decisions concluded that theFederal Reserve cannot applystricter standards to mergers thandoes antitrust law. Conducts OLStests to determine effects of severalvariables on abnormal returns

Hannan andWolken (1989)

1982–87 69 targets,43 bidders

Stock of acquiring and acquiredfirms is traded on NYSE, AMEX,or OTC, the announcementappeared in the WSJ, and thefirms were not involved in amerger during preceding 6 monthsor the month after announcement.The 43 bidders are included onlyif they bid on 1 of the 69 targetfirms in the sample

Event study. Analyzes abnormal andcumulative abnormal returns,relative to the market, of theacquiring and acquired firms for upto 15 days before merger and15 days after. Analysis is based ona standard market model estimatedover the period from 90 days to15 days before the announcement.Stock price data are from Interna-tional Data Service’s Daily StockPrice tape. The market return dataare based on the Wilshire Index

20




Residual analysis

Multiple regression

Acquiring firms generally do not have significantabnormal or cumulative abnormal returns beforemerger or after

There are a couple of significant positive returnsduring the 4 days before merger

Higher outside board membership and location ina statewide branching state are associated withhigher cumulative abnormal returns


Residual analysis

Multiple regression

Acquiring firms have positive significant abnormalreturns during the 2-day announcement period forannouncements before July 1, 1981, and negativesignificant returns during the 2-day announcementperiod for announcements after July 1, 1981

Cumulative excess returns over the period from50 days before announcement to 20 days aftersuggest some gain from announcement toacquirers before July 1, 1981, and some lossdue to announcement after July 1, 1981

OLS tests show that no variables affect abnormalreturns before July 1, 1981. After that date, acouple of variables affect returns, but they maynot be conclusive


Residual analysis Acquiring firms have negative significant cumula-tive abnormal returns both before announcementand after

Acquired firms have positive significant cumula-tive abnormal returns both before announcementand after

A calculation of the dollar value of the change instock prices indicates no net change in value ofthe stock of the 2 firms combined

Results generally do not differ significantlybetween large and small organizations for biddersor targets

Net combined dollar value of wealth effects donot differ between inter- and intrastate acquisitionproposals

Net combined dollar value of wealth effects aresignificantly positive for acquisitions involvingless capitalized targets and significantly negativefor those involving more capitalized targets

21


Study Years studied



Kaen andTehranian(1989)

1979–87 31 in NewHampshire(8 biddersmade all theproposals)

Mergers by banks in New Hamp-shire after a statewide branchinglaw was passed in 1979.Announcement dates weredetermined primarily from theUnion Leader newspaper.Excludes 2 mergers because othermerger announcements were madewithin 2 days of the announce-ments of these 2 mergers. Stockreturns for the 8 bidders are basedon data in the Union Leader, andthe market return data are fromthe NASDAQ Bank Index

Event study. Analyzes abnormalreturns on the stock of the acquiringfirms during various periods from10 days before announcement to10 days after. Analysis is based ona standard market model estimatedover the period from 136 days to16 days before the announcement.Also analyzes returns for 2 differentsubperiods and uses OLS to analyzedifferences in returns due to methodof payment and due to relative sizeof target and acquirer

Wall and Gup(1989)

1981–83 23 Acquiring banks are taken fromthose on the Automated DataProcessing FASTOCK stock pricefile and with stock prices reportedin the American Banker. Acquisi-tions are intrastate and involveBHCs, and acquired firm is atleast 10 percent of the size of theacquirer

Event study. Analyzes abnormalreturns on the stock, relative to themarket, of acquiring firms for theperiod from 2 weeks beforeannouncement to 4 weeks after.Analysis is based on a standardmarket model estimated over theperiod from 44 weeks to 3 weeksbefore announcement. Does notinvestigate effect on acquired firms.Additional tests seek to determinewhat variables influence thecumulative abnormal return

Hawawini andSwary (1990)

1971–86 123(78 bidders,123 targets)

Stock of acquiring and acquiredbanks is publicly traded, withstock price data on CRSP, OTCdaily tapes, or the S&P StockPrice Record

Event study. Analyzes abnormalreturns on the stock of acquiringand acquired banks during variousperiods from 5 weeks before, to5 weeks after, the week of theannouncement of a merger andother announcements. Analysis isbased on a standard market model,as well as a mean-adjusted returnsapproach, estimated over the periodfrom 57 weeks to 6 weeks beforethe announcement week. Alsoanalyzes abnormal returns due toannouncement of regulator’sdecision and announcement ofinterstate legislation. OLS testsconducted to determine what factorsexplain abnormal returns

Rhoades (1990) 1981–87 68 Acquiring and acquired firms havemore than $1 billion in assets.Covers 322 nonmerging firmswith more than $1 billion inassets and located in same statesas acquiring firms for pre-acquisition tests. Postacquisitiontests include 13 acquired and97 nonacquired firms

Analyzes operating performancechanges. Compares performance ofmerging and nonmerging firms bothbefore and after merger. Analysis isbased on average performance overthe period from 3 years beforemerger to 3 years after. Focuses onperformance of the acquired firmand treats it as a separate entityfrom the acquirer both before andafter merger

22




Residual analysis

Multiple regression

Acquiring firms have a negative significantabnormal return on the day before announcementand a negative significant cumulative abnormalreturn on the announcement date and the day after

The 5 mergers based on cash payment tend toshow a positive return while the 26 mergers basedon stock payment tend to show a negative return

OLS tests confirm other findings and also showthat a higher ratio of target size to acquirer size isassociated with higher abnormal returns


Residual analysis Acquiring firms generally have negative signifi-cant cumulative abnormal returns during 4 weeksafter announcement


Residual analysis

Multiple regression

Acquired firms have positive significant abnormalreturns over the 11 weeks centered on theannouncement week

Acquiring firms have negative significant abnor-mal returns over the 11 weeks centered on theannouncement week

Crude estimates of net dollar value of wealtheffects for bidders and targets combined indicatea possible small net increase in wealth forshareholders

Intrastate merger proposals tend to have a morepositive effect on wealth than interstate proposals

Efficiency and profit-ability effects

Return on assets

Noninterest expenses/assets

Logit analysis

Multiple regression

Profit rates of acquired firms do not improvefollowing merger when compared to those ofnonacquired firms

Noninterest cost efficiency does not improverelative to that of nonacquired firms

Pre-acquisition profitability and efficiency ofacquired firms are no different than for non-acquired firms

23


Study Years studied



Allen andCebenoyan(1991)

1979–86 138 Only acquiring firms; they areBHCs with stock listed on NYSEor AMEX and stock prices onCRSP.

Event study. Analyzes abnormalreturns on the stock of the acquiringfirm relative to the market. Analysiscovers 11 days, apparently beforethe merger announcement. Abnor-mal returns are calculated using astandard market model estimatedover the period from 136 days to16 days before the announcement.Analyzes various differences,including stock price reaction tomergers, between 4 sets of firmswith different degrees of manage-ment ownership and concentrationof stock ownership. Emphasis is ondifferences in stock returns due tothe degree of management owner-ship and concentration of stockownership

Cornett and De(1991)

1982–86 37 targetsinvolving 152bids by 59bidders

Includes only interstate mergers.Acquiring or acquired bank mustbe traded on a major exchange orover the counter. 36 of the 37merger proposals occurred.Acquiring banks were excluded ifa second acquisition was madewithin 15 days of initial bid.Acquiring banks, or bidders, didnot necessarily bid on any of the37 target banks in the sample

Event study. Analyzes abnormalreturns of bidding and target banksfor 2 days—the announcement dateand the day before, with datapresented for 15 days before and15 days after announcement.Analysis is based on a standardmarket model estimated over theperiod from 16 days to 75 daysafter the announcement event andrelies on CRSP data. Also analyzesannouncement effect of statelegislation allowing entry byout-of-state banks and uses OLSto analyze the effect of thenumber of potential bidders onabnormal returns around mergerannouncement

Spindt andTarhan (1991)

1986 297 (61intra-BHC,236 newlyacquired)

Broad sample of mergers includesintra-BHC and newly acquiredfirms. A control group of matchednonmerging firms (similar in sizeand state) is used to account forindustry trends

Analyzes operating performancechanges. Analysis is based on theperformance of firms in each of the2 years before and after merger.Analysis treats acquiring firms asseparate entities before merger andas a combined entity with thesuccessor after merger. Differencesin performance between themerging and nonmerging firms arethe basis for analysis. Also com-pares performance of mergers (ofaffiliates) internal to the BHC withmergers of banks not previouslyowned

24




Residual analysis Acquiring firms do not have abnormal returnsduring the 11 days before announcement

Firms with both higher management ownershipand concentration of stock ownership have highercumulative abnormal returns than firms that werelower in either category

Wealth effects ofinterstate mergers

Abnormal returns onstock

Residual analysis

Multiple regression

Acquiring and target firms have positive signifi-cant returns on the day of announcement butgenerally negative returns for the next 15 days

The negative returns to the bidder during 15 daysafter announcement negate the gains on announce-ment day, although the cumulative abnormalreturn to the targets remain positive over the15 days in spite of negative abnormal returns afterannouncement

Acquiring banks are not affected by legislationallowing out-of-state entry, but acquired bankshave a positive return on announcement day

The existence of more potential bidders does notreduce returns to acquirers

Performance effects Net income/equity

Net income/assets

Employee expenses/revenue

Univariate sign tests Before merger, externally merged firms generallyexhibit poorer performance in terms of rates ofreturn and the ratio of employee expenses toassets than successor firms. Internally mergedfirms show no difference in performance whencompared with the successor firms

Return on assets of newly acquired (external)firms generally does not improve after merger

Return on equity of newly acquired (external)firms generally tends to improve after merger

Ratio of employee expenses to assets of newlyacquired (external) firms generally does notimprove after merger

Internally merged firms generally do not showperformance improvements after merger

25


Study Years studied



Baradwaj,Dubofsky, andFraser (1992)

1981–87 108 Stock of acquiring firms is listedon CRSP daily file or CRSPNASDAQ historical file. Thetarget firm is at least 10 percentof the size of the acquirer.Analysis focuses only on theacquiring firm

Event study. Analyzes abnormalreturns to the stock of the acquiringfirm relative to the market asmeasured by a CRSP value-weighted portfolio. Abnormalreturns are calculated on the basisof a standard market model esti-mated for the period from 60 daysto 11 days before announcementand 11 days to 60 days after.Analysis covers various periodsfrom 5 days before to 5 days afterthe announcement. Comparesabnormal returns of inter- andintrastate mergers. Also comparesstock effects of merger announce-ment before and after the June 10,1985, Supreme Court decisionvalidating constitutionality of statelaws permitting interstate banking

Berger andHumphrey(1992)

1981–89 57 (involves69 mergers,but multiplemergers in thesame orcontiguousyears arecombined)

Acquiring and acquired firms havemore than $1 billion in assets.Includes all nonmerging firms inU.S. with more than $1 billion inassets, for an average sample sizeof about 300 banks

Analyzes operating performancechanges. Analysis comparesperformance rank from beforemerger to after. Acquiring andacquired firms are treated as acombined entity before merger andafter. A translog total cost functionis estimated for each year (1980–90) for merging and nonmergingbanks, and each bank’s costfunction residual over time iscompared with the average residualsfor each of the members of the peergroup (large merging and nonmerg-ing firms). The residual, whichrepresents X-efficiency performance,and other performance measures areused to rank merging firms relativeto their peers before and aftermerger. Pre- and postmergerperformance is measured overvarying time periods. Controls fordeposit overlap and difference inperformance between acquiring andacquired firms before merger

26




Residual analysis

Multiple regression

Acquiring firms have negative significant abnor-mal returns 5 days before announcement and5 days after for both inter- and intrastate mergers

Negative abnormal returns to inter- and intrastateacquirers exist before and after the June 10, 1985,court decision on state interstate banking laws


X-efficiency rank

Total efficiency rank

Return on assets

Average total costs/assets


Multiple regression Cost efficiency, on average, does not improvefollowing merger

Some mergers improve efficiency and someworsen efficiency

Profit rates, on average, do not improve followingmerger

Some mergers improve profit rates and othersworsen profit rates

Firms making in-market mergers do not haveefficiency improvements when compared withother mergers or firms

Mergers in which the acquiring firms are moreefficient than the acquired do not lead to effi-ciency improvements when compared with othermergers or firms

27


Study Years studied



Cornett andTehranian(1992)

1982–87 30 Acquiring and acquired firms arepublicly traded with stock returndata from CRSP tapes andMoody’s Bank Manuals. Alsoincludes, as a proxy for theindustry, all banks with stocktraded on the NYSE or AMEX(30 to 36 banks, depending uponyear)

Uses both the event study andoperating performance methodolo-gies. Event study analyzes stockprices for the day before and theday of announcement using astandard market model estimatedover the period from 136 days to16 days before announcement.Analysis of operating performancecovers 3 years before and 3 yearsafter merger. Also comparesperformance and abnormal returnsof inter- and intrastate mergers.Operating performance of combinedacquirer and acquired is comparedto the mean performance of allbanks with stock traded on theNYSE or AMEX

Linder andCrane (1993)

1982–87 47 (25intra-BHC,22 newlyacquired)

Acquiring and acquired banks arelocated in New England states. Acontrol group of all nonmergingbanks in the same state is used toaccount for industry trends. Onaverage, the acquired firm is 35percent as large as the acquirer,although this is not clear in thedata. Includes intra-BHC mergersas well as newly acquired firms

Analyzes operating performancechanges. Analysis is based on theperformance of firms during 1 yearbefore merger and 1 year and2 years after, as well as changesover the period. Performance ofmerging firms is compared with thatof a comparable industry group.Analysis treats acquiring andacquired firms as a single entityboth before merger and after.Efforts are made to adjust data formergers occurring after the mergerunder analysis. Compares findingsbetween newly acquired banks andintra-BHC mergers

O’Keefe (1992) 1984–90 469 (123assisted, 346nonassisted)

Covers acquiring banks that madeFDIC-assisted acquisitions and theacquired bank was absorbed intothe acquirer. A group of non-assisted acquirers is included forcomparison. Acquiring firms aregenerally small. The majority ofthese mergers involve firms in thesame market. Acquisitions areexcluded if another acquisition bythe acquirer is made within aperiod of 3 years before acquisi-tion or after

Analyzes performance changes fromthe 2 years before merger to the2 years after. Emphasis is oncomparison of assisted with unas-sisted mergers. Only noninterestexpenses to assets are comparedwith an industry control group.Acquiring and acquired firms aretreated as a combined entity afteracquisition

28



Performance and wealtheffects

Abnormal returns onstock

Expenses/revenues

Cash flow/marketvalue of assets

Return on equity

Return on assets

Various univariatetests

Residual analysis

Efficiency overall (total expenses/revenues) doesnot improve relative to that of the industry,although employee productivity improves

Ratio of cash flow to market value of assetsimproves relative to that of the industry

Return on equity improves relative to that of theindustry

Return on assets does not improve relative to thatof the industry

Acquiring firms have negative abnormal returns inthe 2 days covered by the announcement day andthe day before

Acquired firms have positive abnormal returns inthe 2 days covered by the announcement day andthe day before, and the combined firm haspositive aggregate returns

Inter- and intrastate mergers experience similarabnormal returns

Performance effects Operating income(and components ofoperating income)/assets


Growth in income andassets

Univariate t tests

Multiple regression

Operating income/assets and growth in operatingincome of merging banks, as a group, do notimprove after merger relative to those of non-merging firms

Cost efficiency (noninterest) of merging banks, asa whole, declines relative to that of nonmergingfirms

Intra-BHC mergers generally have greater perfor-mance improvement, in terms of operatinginome/assets and growth in operating inome, thannewly acquired (not de novo) firms

Operating income/assets of intra-BHC mergersimproves relative to that of nonmerging firms

Cost efficiency (noninterest) and growth ofintra-BHC mergers decline relative to those ofnonmerging firms

Growth in assets of merging firms, as a group, isslower than that of nonmerging firms

Performance effects Noninterest expenses/assets

Interest expenses/assets

Return on assets

Return on equity

Graphical trendanalysis

Efficiency for the assisted and unassisted mergedfirms, in terms of both interest and noninterestexpenses to assets, does not change, a resultsimilar to industry trends in noninterest expenses

Assisted and unassisted mergers generally yieldsimilar results

No efficiency changes for those merged firmsoperating in the same county before merger

Return on assets for the assisted and unassistedmerged firm does not change

Return on equity for the assisted and unassistedmerged firm does not change

29


Study Years studied



Rose (1992) 1980–89 279 Limited to interstate acquisitions.A control group of banks appar-ently not involved in an interstateaccquisition, of similar size andlocated in the same county ormetropolitan statistical area as theacquired bank, is included forcomparison. The sample shouldhave a rather wide variation ingeographic location and size ofinstitution

Analyzes operating performancechanges. Studies performance ofacquired and acquiring banks bothbefore acquisition and after. Analy-sis usually covers each of 5 yearsbefore acquisition and after. Perfor-mance of acquiring banks isanalyzed separately. However, forsome analyses all acquired banksowned by the same BHC areconsolidated as a single entity andcomparable hypothetical sets of peerbanks are constructed for compari-son. The detailed tabular compari-sons of performance before mergerwith performance after, by year, donot appear to account for peers, sofindings may be difficult to inter-pret. A multiple regression modelthat analyzes a smaller number ofperformance measures does controlfor performance of peers

Spong andShoenhair(1992)

1985–87 179 Limited to interstate acquisitions,a selection that should involvefairly large firms operating indifferent geographic areas. Alsoincludes a control group of banksnot under interstate ownership thatare similar in terms of size andlocation

Analyzes observed performancechanges. Analysis is based on datafor 1 year before merger and 1 yearto 3 years after. Only the acquiredfirms are studied. A peer group isused for comparison to control forindustry trends. Analysis is basedon comparisons of means andmedians of the acquired firms withthe control group for variousperformance measures

Srinivasan andWall (1992)

1982–86 240 Acquiring and acquired firms eachhave more than $100 million inassets. An industrywide sample ofnonmerging firms is used tocontrol for industry trends inunivariate tests but not in multipleregressions, where time dummyvariables are used instead. Onaverage, the acquired firm is32 percent as large as the acquirer

Analyzes operating efficiency anddollar value of expense changes.Analysis is based on averageefficiency and expenses for the2 years before merger and 1 year to4 years after merger. The analysistreats the acquiring and acquiredfirms as a single entity both beforemerger and after. The analysisaccounts for the degree of overlapof deposits in local markets.Multiple acquisitions in 1 year by afirm are treated as a single merger.20 outliers, in terms of expenseratios, are excluded

Srinivasan(1992)

1982–86 Two samples:77 in south-east, 240 inentire U.S

Acquiring and acquired bankshave more than $100 million inassets. Nonmerging banks are alsoincluded

Analyzes operating efficiencychanges. Focuses on efficiencychanges from 2 years before mergerto each of 4 years after merger.Acquiring and acquired firms aretreated as a combined entity beforemerger and after. Changes inefficiency of merged banks arecompared with nonmerging banks.Accounts for degree of marketoverlap of merger partners

30




Operating expenses/total revenue

Noninterest operatingexpenses/totalexpenses

Noninterest expensesper employee

Return on assets

Return on equity

Multiple regression Multiple regression results, which include acontrol group of banks, indicate the following:Changes in operating efficiency (operatingexpenses/total revenue), return on equity, andmarket share for various types of loans anddeposits do not differ from those of nonmergingfirms

Tabular results suggest that return on equity ofmerging firms, but not return on assets or effi-ciency, improves somewhat relative to that of acontrol group

Performance effects Return on assets

Return on equity

Overhead costs/assets

Personnel expenses/assets

Tabular comparisons Cost efficiency (overhead) tends to improvesomewhat relative to that of peers

Returns on assets and equity generally declinerelative to those of peers based on mean and donot decline or increase relative to those of peersbased on median

Efficiency effects Noninterest expenses/total assets

Noninterest expensesin terms of dollarvolume

Wilcoxian signed ranktests

Multiple regression

Univariate t tests

Cost efficiency does not improve relative to thatof nonmerging firms

Greater office overlap tends to be associated withlower costs after merger but apparently not withefficiency gains

Efficiency effects Noninterest expenses/operating income

Also analyzes compo-nents of noninterestexpenses, i.e., salaries,premises, and otherexpenses

Univariate t tests

Multiple regression

Efficiency of merged firms does not improverelative to that of nonmerging firms

Changes in efficiency of firms making in-marketmergers do not differ from those of firms in othermergers

Firms in mergers of equals tend to have efficiencygains when compared with firms in other mergers

Overall efficiency findings are similar for banks insoutheast and U.S. as a whole

31


Study Years studied



Crane andLinder (1993)

1991 1 Fleet–Bank of New Englandmerger

Case study analyzing operatingperformance changes. The studyexamines the firm’s projections forcost cutting (both real reductionsand financial reductions) and usespublicly available data to analyzeperformance before merger andafter. Examines expenses for 1 yearafter merger. Analysis is based onthe combined banks in the BHCsrather than the BHC. Changes arenot compared with a control group

DeYoung(1993)

1987–88 348 348 mergers approved by theComptroller of the Currency andcovering a wide range of banksizes. About 31 percent ofmergers involve purchase andassumptions of failed banks and43 percent involve banks alreadyaffiliated within the same BHC.Sample includes all other com-mercial banks for comparison

Analyzes operating changes inefficiency. Compares pre- andpostmerger costs of merging banksto efficiency benchmarks estimatedfrom a ‘‘ thick’’ cost frontier thatwas estimated for all U.S. commer-cial banks using a multiproducttranslog production function. Costefficiency is measured for 1 yearbefore merger and the 4th year aftermerger. Efficiency is assessed interms of total expenses to totalassets. Analyzes (1) pre-merger costefficiency by comparing costs ofthe target and acquirer, separately,with the pre-merger thick frontierbenchmark and (2) the postmergerefficiency of the merged firm withthe postmerger benchmark. Alsocompares the postmerger position ofthe merged firm relative to thepostmerger benchmark with thepre-merger position of the acquirerrelative to the pre-merger bench-mark to determine whether effi-ciency improved

Peristiani(1993a)

1981–88 Two samples,large numberin each

One sample, for tabular analysis,and one, for statistical analysis,covering all banks involved inmultiple mergers. Includes about2,000 mergers and a control groupof nonmerging banks

Analyzes operating performancechanges. Examines performanceduring 1 year before merger and2 years to 4 years after. Comparesperformance of internal BHCmergers, intrastate mergers, andFDIC-assisted mergers. Alsoseparately examines mergers before1986 and the top 200 mergers basedon target size. Analysis focuses onthe target and acquirer separatelybefore merger and on the survivorafter merger

32



Efficiency effects Noninterest expenses None Noninterest expenses declined about asprojected—30 percent over the year followingmerger, compared with a 10 percent decline inassets

No data reported on total expense results

The acquiring bank was more efficient than theacquired at the time of merger

Efficiency effects Improvement inX-efficiency rankbased on totalcosts/total assets

Univariate t tests

Multiple regression

Efficiency gains do not generally result frommerger

Efficiency gains are most likely when both thetarget and acquirer are inefficient before mergerrather than when the acquirer is more efficientthan the target

Acquiring banks, on average, are somewhat moreefficient than targets

Both acquiring and target firms exhibit substantialinefficiency relative to ‘‘ best-practice’’ fi rms


Net income/totalassets

Noninterest expenseratio. Denominatornot clear but probablyassets

Univariate t tests Efficiency does not improve in internal BHCmergers nor for all mergers when compared withthat of a control group of nonmerging banks

In-market mergers produce no improvement

Profit rates for overall sample do not changesignificantly relative to those of a control group

Results vary somewhat over different groupscompared, but overall findings hold

33


Study Years studied



Peristiani(1993b)

1981–88 4,900 (2,000intra-BHCmergers,2,900 externalmergers)

All bank mergers, including about1,000 FDIC-assisted mergers.Multiple mergers by a bank in1 year are treated as singlemerger. Includes a control groupof nonmerging banks of similarsize and in the same states asmerging banks

Analyzes operating performancechanges in terms of expense/incomeratio and efficiency. Compareschanges in profit rates and expenseratios from 1 year before merger tothe average of the 4 years after formerged firms and a control group.Also estimates a multiple producttranslog cost function to derivemeasures of scale and X-efficiency.Compares changes in scale andX-efficiency from 1 year beforemerger to the average of 2 to4 years after for merged firms and acontrol group. Distinguishesbetween intra-BHC, intrastate, andFDIC-assisted mergers. Focusesprimarily on the combined entitybefore merger and after. Also usesOLS to analyze determinants ofmergers including a deposit overlapvariable

Rhoades (1993) 1981–86 898 Includes banks or BHCs engagedin horizontal, or in-market,mergers. Also includes about10,000 nonmerging banks

Analyzes operating performancechanges. Two methodologies areused. One analyzes changes inexpense ratios from the 3 yearsbefore merger to the 4th through6th years after. The other analyzeschanges in efficiency quartile fromthe 3 years before merger to the 4ththrough 6th years after. In acquisi-tions by BHCs, the tests compareefficiency of the combined entity(before merger and after) withnonmerging firms. The tests accountfor the degree of office overlap andare conducted for each of the years1981–86

34




Return on assets

Noninterest expenseratio. Denominatornot clear but probablyassets

Change in scaleefficiency

Change inX-efficiency

Univariate t tests

Multiple regression

Noninterest expense ratios do not improve relativeto those of a control group for firms in multiplemergers. Expense ratios of firms in single mergersincrease relative to those of the control group

Return on assets of the combined merged firmimproves, from a low pre-merger level, relative tothat of the control group, although the return onassets for the acquirer alone declines

X-efficiency does not improve, and may evendecline, relative to that of the control group

Scale efficiency improves for firms in multiplemergers, and declines for firms in single mergers,relative to that of the control group

Branch, or deposit, overlap between the acquiringand acquired firms is not related to changes inefficiency or profitability

Efficiency effects Total expenses/assets


Multiple regression

Logit analysis

Ratio of total expenses to assets does not changefor firms doing in-market mergers when comparedwith that of nonmerging firms

Ratio of noninterest expenses to assets does notchange for firms doing in-market mergers whencompared with that of nonmerging firms

A greater degree of office overlap of mergingfirms is not associated with an improvement inefficiency relative to that of nonmerging firms

Acquiring banks are, on average, more efficientthan acquired banks

35

Bibliography of Studies Summarized

Allen, Linda, and A. Sinan Cebenoyan. ‘‘BankAcquisitions and Ownership Structure: Theoryand Evidence,’’Journal of Banking andFinance, vol. 15 (April 1991), pp. 425–48.

Baradwaj, Babu G., David A. Dubofsky, andDonald R. Fraser. ‘‘Bidder Returns in Interstateand Intrastate Bank Acquisitions,’’Journal ofFinancial Services Research, vol. 5 (February1992), pp. 261–73.

, Donald R. Fraser, and Eugene P.H.Furtado. ‘‘Hostile Bank Takeover Offers:Analysis and Implications,’’Journal of Bankingand Finance, vol. 14 (December 1990),pp. 1229–42.

Berger, Allen N., and David B. Humphrey.‘‘Megamergers in Banking and the Use of CostEfficiency as an Antitrust Defense,’’AntitrustBulletin, vol. 37 (Fall 1992), pp. 541–600.

Bertin, William J., Farrokh Ghazanfari, and KhalilM. Torabzadeh. ‘‘Failed Bank Acquisitions andSuccessful Bidders’ Returns,’’FinancialManagement, vol. 18 (Summer 1989),pp. 93–100.

Cornett, Marcia Millon, and Hassan Tehranian.‘‘Changes in Corporate Performance Associatedwith Bank Acquisitions,’’Journal of FinancialEconomics, vol. 31 (April 1992), pp. 211–34.

, and Sankar De. ‘‘Common StockReturns in Corporate Takeover Bids: Evidencefrom Interstate Bank Mergers,’’Journal ofBanking and Finance, vol. 15 (April 1991),pp. 273–95.

Crane, Dwight B., and Jane C. Linder. ‘‘TheStrategy of a Merger: Fleet and Bank of NewEngland,’’ in Proceedings of the 29th AnnualConference on Bank Structure and Competition(Federal Reserve Bank of Chicago, May 1993),pp. 371–76.

de Cossio, Francisco, Jack W. Trifts, and Kevin P.Scanlon. ‘‘Bank Equity Returns: The Differencebetween Intrastate and Interstate BankMergers.’’ Working Paper. Columbia: Universityof South Carolina, College of Business Admin-istration, 1988.

De, Sankar, and Diane Duplichan. ‘‘Effects ofInterstate Bank Mergers on Shareholder Wealth:Theory and Evidence,’’Geld, Banken undVersicherungen (Band II, 1987), pp. 751–73.

Desai, Anand S., and Roger D. Stover. ‘‘BankHolding Company Acquisitions, StockholderReturns, and Regulatory Uncertainty,’’Journalof Financial Research, vol. 8 (Summer 1985),pp. 145–56.

DeYoung, Robert. ‘‘Determinants of Cost Efficien-cies in Bank Mergers,’’ Economic and PolicyAnalysis Working Paper 93–1. Washington:Office of the Comptroller of the Currency,August 1993.

Dubofsky, David A., and Donald R. Fraser.‘‘Regulatory Change and the Market for BankControl,’’ in Benton E. Gup, ed.,Bank Mergers:Current Issues and Perspectives. Boston:Kluwer, 1989, pp. 121–39.

Frieder, Larry A., and Vincent P. Apilado. ‘‘BankHolding Company Expansion: A Refocus on ItsFinancial Rationale,’’Journal of FinancialResearch, vol. 6 (Spring 1983), pp. 67–81.

Hannan, Timothy H., and John D. Wolken.‘‘Returns to Bidders and Targets in the Acquisi-tion Process: Evidence from the BankingIndustry,’’ Journal of Financial ServicesResearch, vol. 3 (October 1989), pp. 5–16.

Hawawini, Gabriel A., and Itzhak Swary.Mergersand Acquisitions in the U.S. Banking Industry:Evidence from the Capital Markets. New York:North Holland, 1990.

James, Christopher, and Peggy Wier. ‘‘An Analysisof FDIC Failed Bank Auctions,’’Journal ofMonetary Economics, vol. 20 (July 1987a),pp. 141–53.

, and . ‘‘Returns toAcquirers and Competition in the AcquisitionMarket: The Case of Banking,’’Journal ofPolitical Economy, vol. 95 (April 1987b),pp. 355–70.

Kaen, Fred R., and Hassan Tehranian. ‘‘NewHampshire Bank Mergers: The Returns toAcquiring Bank Shareholders,’’ in Benton E.Gup, ed.,Bank Mergers: Current Issues andPerspectives. Boston: Kluwer, 1989,pp. 141–55.

36

Linder, Jane C., and Dwight B. Crane. ‘‘ BankMergers: Integration and Profitability,’’ Journalof Financial Services Research, vol. 7 (January1993), pp. 35–55.

Lobue, Marie. ‘‘ Categorical Bank Acquisitions,’’Journal of Bank Research, vol. 14 (Winter1984), pp. 274–82.

Neely, Walter P. ‘‘ Banking Acquisitions: Acquirerand Target Shareholder Returns,’’ FinancialManagement, vol. 16 (Winter 1987), pp. 66–74.

O’Keefe, John. ‘‘ Bank Failure Resolutions:Implications for Banking Industry Structure,Conduct and Performance,’’ FDIC BankingReview, vol. 5 (Spring/Summer 1992),pp. 17–35.

Peristiani, Stavros. ‘‘ The Effect of Mergers onBank Performance,’’ in Studies on ExcessCapacity in the Financial Sector. FederalReserve Bank of New York, June 1993a,pp. 90–120.

. ‘‘ Evaluating the PostmergerX-Efficiency and Scale Efficiency of U.S.Banks.’’ Federal Reserve Bank of New York,August 1993b.

Pettway, Richard H., and Jack W. Trifts. ‘‘ DoBanks Overbid When Acquiring Failed Banks?’’Financial Management, vol. 14 (Summer 1985),pp. 5–15.

Rhoades, Stephen A. ‘‘ Billion Dollar BankAcquisitions: A Note on the PerformanceEffects.’’ Board of Governors of the FederalReserve System, 1990.

. ‘‘ Efficiency Effects of Horizontal(In-Market) Bank Mergers,’’ Journal of Bankingand Finance, vol. 17 (April 1993), pp. 411–22.

. The Operating Performance ofAcquired Firms in Banking before and afterAcquisition. Staff Studies 149. Washington:Board of Governors of the Federal ReserveSystem, 1986. A version with alternativestatistical tests is, ‘‘ The Operating Performanceof Acquired Firms in Banking,’’ in Robert L.Wills, Julie A. Caswell, and John D. Culbert-son, eds., Issues after a Century of Federal

Competition Policy. Lexington, Mass.: Lexing-ton Books, 1987, pp. 277–92.

Rose, Peter S. ‘‘ Improving Regulatory Policy forMergers: An Assessment of Bank MergerMotivations and Performance Effects,’’ Issues inBank Regulation (Winter 1987a), pp. 32–39.

. ‘‘ The Impact of Mergers in Banking:Evidence from a Nationwide Sample of Feder-ally Chartered Banks,’’ Journal of Economicsand Business, vol. 39 (November 1987b),pp. 289–312.

. ‘‘ Interstate Banking: Performance,Market Share, and Market ConcentrationIssues,’’ Antitrust Bulletin, vol. 37 (Fall 1992),pp. 601–30.

Spindt, Paul A., and Vefa Tarhan. ‘‘Are ThereSynergies in Bank Mergers?’’ New Orleans:Tulane University, November 11, 1991.

Spong, Kenneth, and John D. Shoenhair. ‘‘ Perfor-mance of Banks Acquired on an InterstateBasis,’’ Federal Reserve Bank of Kansas City,Financial Industry Perspectives (December1992), pp. 15–32.

Srinivasan, Aruna. ‘‘Are There Cost Savings fromBank Mergers?’’ Federal Reserve Bank ofAtlanta, Economic Review, vol. 77 (March–April 1992), pp. 17–28.

, and Larry D. Wall. ‘‘ Cost SavingsAssociated with Bank Mergers,’’ WorkingPaper 92–2. Federal Reserve Bank of Atlanta,February 1992.

Sushka, Marie E., and Yvette Bendeck. ‘‘ BankAcquisitions and Stockholders’ Wealth,’’Journal of Banking and Finance, vol. 12(December 1988), pp. 551–62.

Trifts, Jack W., and Kevin P. Scanlon. ‘‘ InterstateBank Mergers: The Early Evidence,’’ Journal ofFinancial Research, vol. 10 (Winter 1987),pp. 305–11.

Wall, Larry D., and Benton E. Gup. ‘‘ MarketValuation Effects of Bank Acquisitions,’’ inBenton E. Gup, ed., Bank Mergers: CurrentIssues and Perspectives. Boston: Kluwer, 1989,pp. 107–20.

37

167 a summary of merger performance studies in banking ... · and linder, 1993, which is a case...

Documents