does antitrust policy improve consumer welfare? assessing ...does antitrust policy improve consumer...

J O I N T C E N T E R AEI-BROOKINGS JOINT CENTER FOR REGULATORY STUDIES

Does Antitrust Policy Improve Consumer Welfare Assessing the Evidence

Robert W Crandall and Clifford Winston

Related Publication 04-07 April 2004

(Published in the Journal of Economic Perspectivesmdash Volume 17 Number 4mdashFall 2003

Reprinted with permission from the Journal of Economic Perspectives)

Does Antitrust Policy ImproveConsumer Welfare Assessing theEvidence


S hould the United States pursue a vigorous antitrust policy Soon after thepassage of the Sherman Antitrust Act of 1890 economists led by John BatesClark (1901) argued that the enforcement of such laws should be informed

by the prevailing economic theory on the merits of competition and the extent towhich rmsrsquo conduct can enhance or weaken competition However economictheory since then has proven remarkably fertile in pointing out how various actionsby rms may be interpreted as either procompetitive or anticompetitive Forexample when prices decline suf ciently so that no rm in an industry is earningeconomic pro ts and some rms exit this outcome may re ect a highly competitivemarket adjusting to a condition of temporary oversupply or it could indicate thata large competitor is employing a strategy of predatory pricing to drive out its rivalsSimilarly when a rm builds a large factory it may be engaged in vigorouscompetition and new entry or it may be creating excess capacity as an implicitthreat to potential competitors that it may raise output and cut price quickly ifcircumstances warrant Although economic theory can help organize analysis of theeconomic variables affected by antitrust policy it often offers little policy guidancebecause almost any action by a rm short of outright price xing can turn out tohave procompetitive or anticompetitive consequences

Given this range of theoretical possibilities the case for a tough and broadantitrust policy must rest on empirical evidence that shows that such policies haveworked in the broad social interest In this paper we argue that the currentempirical record of antitrust enforcement is weak We start with an overview of thebudgets and actions of the federal governmentrsquos antitrust authorities We thensynthesize the available research regarding the economic effects of three majorareas of antitrust policy and enforcement changing the structure or behavior of

y Robert W Crandall and Clifford Winston are both Senior Fellows The Brookings Institu-tion Washington DC

Journal of Economic PerspectivesmdashVolume 17 Number 4mdashFall 2003mdashPages 3ndash26

monopolies prosecuting rms that engage in anticompetitive practices namelyprice xing and other forms of collusion and reviewing proposed mergers We ndlittle empirical evidence that past interventions have provided much direct bene tto consumers or signi cantly deterred anticompetitive behavior1 We acknowledgethat the literature has not been able to utilize all potentially fruitful sources of dataand has rarely implemented recent empirical advances in industrial organization toanalyze the effects of speci c antitrust cases Thus the state of knowledge is not ata point where we are ready to make sweeping policy recommendations Nonethe-less the economics profession should conclude that until it can provide some hardevidence that identi es where the antitrust authorities are signi cantly improvingconsumer welfare and can explain why some enforcement actions and remedies arehelpful and others are not those authorities would be well advised to prosecuteonly the most egregious anticompetitive violations

The Scope of Antitrust Activity

US antitrust enforcement is primarily the responsibility of the Department ofJustice and the Federal Trade Commission (FTC) (There are also state antitrustlaws that are enforced by state attorneys general but the federal activity is far morepervasive) The Department of Justice enforces Section 1 of the Sherman Actprohibiting contracts combinations and conspiracies in restraint of trade and alsoenforces Section 2 of the Sherman Act prohibiting actions to monopolize orattempts to monopolize markets The Department of Justice and the FTC enforceSection 7 of the Clayton Antitrust Act of 1914 prohibiting mergers between rmsthat threaten to reduce competition substantially in any line of commerce TheClayton Act also prohibits anticompetitive practices like tying arrangements (whereconsumers are forced to purchase from a rm a product like razor blades whenthey buy the rmrsquos razors) and disallows competing rms from having overlappingboards of directors The FTC may also initiate cases under Section 5 of the FederalTrade Commission Act for ldquounfair methods of competitionrdquo thereby providing itwith the ability to combat abuses that DOJ attacks under Sections 1 and 2 of theSherman Act For example the FTC initially investigated Microsoft for possibleanticompetitive practices The Department of Justice subsequently brought itsSection 2 case after the FTC did not bring a complaint

Data on investigations and budgets for the Department of Justice and the FTCpublicly available for only the past 20 years are summarized in Table 1 Monopo-lization cases constitute a small share of antitrust investigations in a given year but

1 Our focus is on academic assessments of antitrust policy not studies conducted by federal agencies Infact there are very few government assessments of the economic effects of past antitrust decisions Whenthe government examined the outcome of mergers or divestiture orders its focus has typically not beenon competition or consumer welfare but on the viability of the proposed action For example theFederal Trade Commissionrsquos Bureau of Competition (1999) examined the viability of divestitures in 35merger cases between 1990 and 1994 in which divestiture orders were issued as a condition forapproving the merger

4 Journal of Economic Perspectives

they still absorb a moderate fraction of the Department of Justice antitrust budgetDOJ investigated a declining number of price xing allegations and other poten-tially collusive arrangements such as vertical market restraints during this periodbut still spent at least one-third of its budget on this activity Investigations ofproposed mergers currently account for the largest share of antitrust activity withthe FTC handling slightly more mergers than the Department of Justice Untilrecently the FTCrsquos budget for mergers was equal to the budget of the AntitrustDivision of the Department of Justice for all its investigations

Total resources consumed by antitrust enforcement however amount tomuch more than government antitrust agency expenditures shown in Table 1Firms involved in antitrust cases must pay for legal advice particularly in obtainingapprovals for mergers and acquisitions Fisher and Lande (1983) estimate that amerger case cost a rm as much as $15 million during the 1980s Firms that facea lawsuit must pay for their defense which could involve a lengthy trial andsubsequent appeals Antitrust cases also require the time and resources of manage-ment and critical staff to address issues of rm conduct to provide nancialinformation and so on We are not aware of estimates of the costs to rms causedby antitrust investigations and court proceedings but they undoubtedly run intothe billions of dollars per year Finally the largest cost of antitrust enforcement may

Table 1DOJ Antitrust Division and FTC Investigations and Budgets 1981 1991 2000(in millions of year 2000 ination-adjusted dollars)

Investigations

Agency Conduct 1981 1991 2000

Antitrust Division Monopolies 8 5 8Mergers 66 92 177Price Fixing 145 77 82

FTC Mergers 104 136 189

TOTAL 323 310 456

BudgetsAgency Conduct 1981 1991 2000

Antitrust Divisiona Monopolies and Mergers $311 $233 $572Price Fixing $222 $246 $307

FTCb Mergers $544 $455 $590

TOTAL $1077 $934 $1469

Sources US Budget 1982 1992 2002 Department of Justice Budget FY 1981 1991 2000 AntitrustDivision Workload Statistics 1981ndash1990 1991ndash2000 5th 14th and 23rd Annual Hart-Scott Rodino Report(FY 1981 1991 and 2000)a Antitrust Division budgetary information does not distinguish between expenditures on monopoly andmerger casesb Although its primary antitrust responsibility concerns mergers the FTC also occasionally brings casesrelated to tying arrangements price discrimination and unfair methods of competition under provisionsof the Clayton Act and the Federal Trade Commission Act

Robert W Crandall and Clifford Winston 5

be that rms are discouraged from pursuing potentially ef cient mergers takingcompetitive pricing actions developing new products or making new investmentsfor fear of being embroiled in an antitrust action especially if competitors use theantitrust authorities to block one another Of course the gains to consumers fromcurbing anticompetitive offenses could potentially outweigh these enforcementcosts

The ideal way to determine whether consumers have bene ted from antitrustpolicy and enforcement in the areas of monopolization collusion and mergerswould be to compare consumer welfare with and without antitrust policy all elseconstant2 However twentieth-century US history has offered only one example ofthis counterfactual During the Great Depression antitrust laws were suspended fordesignated industries for a time as a byproduct of the 1933 National IndustrialRecovery Act Bittlingmayer (1995) studied this episode and found that prices didnot rise an intriguing nding but dated and perhaps relevant only to the anom-alous experience of the Great Depression Other evidence is available from casesthat compare prices before and after antitrust interventions or across industriessubject to varying levels of antitrust enforcement

Monopolization

The Department of Justice typically investigates fewer than ten potentialmonopolization violations a year To prove monopolization the government mustdemonstrate that a rm has power over price and output in a market and that thispower derives from business decisions whose principal intent and effect was toexclude competition (Areeda 1988) Remedies in monopolization cases may becharacterized as structural behavioral or a reduction in the control of intellectualproperty Structural remedies involve court-ordered changes in a rmrsquos or indus-tryrsquos structure such as horizontal divestiture in which two or more directly com-peting companies are created from the assets of the defendant and verticaldivestiture where separate companies are created at different production stagesBehavioral remedies address some aspect of the rmrsquos behavior that the govern-ment identi ed as anticompetitive such as tying arrangements collusive agree-ments to exclude competitors predatory pricing and so on An enforcementagency must monitor those prohibitions and the courts are inevitably required toresolve issues that arise between the agency and the rm Finally relief may involveforcing the rm to give up or to license key intellectual property that is the sourceof the alleged monopoly power

Monopolization cases are impossible to analyze en masse because they involvedifferent market conditions and alleged misconduct over time We thereforeinvestigate the ef cacy of antitrust policy in curbing monopolization by focusing onsome landmark cases during the past century including Standard Oil American

2 Our assessment does not include cases involving allegations of price discrimination brought underSection 2 of the Clayton Act because such cases have been relatively rare during the past 20 years


Tobacco Alcoa Paramount United Shoe Machinery and ATampT A detailed discus-sion of these and other cases and their effects on consumer welfare can be foundin Crandall (2001) These cases are of particular interest here because the govern-ment prevailed in each of them and obtained substantial changes leading to theexpectation of consumer bene ts To be sure these cases are decades old butcurrent law and attitudes toward monopolization are based on precedents estab-lished by such cases We sketch each case and draw on the available empiricalevidence to assess whether the remedy improved consumer welfare

Standard OilDuring the late 1800s and early 1900s the Standard Oil Company re ned and

marketed crude oil produced in Pennsylvania Ohio Indiana and several surround-ing states and developed transportation and production facilities Complaintsabout its business practices took various forms Standard Oil was alleged to haveused ruthless tactics in negotiating contracts with railroads and in denying inde-pendent oil companies access to its pipelines and transportation facilities It wasalso alleged to have engaged in predatory pricing to drive rivals from the marketa claim disputed by McGee (1958) Public authorities feared that the Standard OilldquoTrustrdquo which pooled the companyrsquos pro ts was a source of market power andfacilitated price xing In 1911 the US Supreme Court upheld a 1909 lower courtdecision that Standard Oil had violated Sections 1 and 2 of the Sherman Act byattempting to monopolize the countryrsquos petroleum industry and using its NewJersey Trust to restrain trade (Standard Oil Company of New Jersey v United States 221US 1 [1911]) The courtrsquos decree required that the Trust be dissolved resultingin 38 separate and independent companies that were prohibited from beingcontrolled by a single entity

The government presumably expected the breakup of Standard Oil to reduceUS re ned petroleum product prices and perhaps also to reduce monopsonypower over crude oil prices Because of new oil discoveries real crude oil priceswere falling even before Standard Oil was brought to trial and actually rosesomewhat after the breakup as shown in Figure 1 Kerosene and gasoline prices uctuated after the decree was entered As a simple formal analysis we collectedannual time series data from 1889ndash1917 and we regressed real US crude oil priceson GNP total automobile registrations and total electricity production (whichcontrol for major in uences on petroleum demand) a time trend from 1889ndash1900that controls for the opening up of new western US elds that increased petro-leum supply and a dissolution dummy (de ned as 1 for 1912ndash1917 0 otherwise)The coef cient for the dissolution dummy was actually positive 050 but statisti-cally insigni cant with a t-statistic of 088 (The dummyrsquos sign and signi cance wasnot affected when we deleted some of the explanatory variables)

Earlier commentators have also concluded that the breakup of Standard Oilhad little effect on either consumers or on pro ts because Standardrsquos allegedmarket power had already declined substantially from its heyday For exampleStandard Oilrsquos market share of re nery capacity in the United States had fallenbefore the decree from 82 percent in 1899 to 64 percent in 1911 as oil-producing

Does Antitrust Policy Improve Consumer Welfare Assessing the Evidence 7

regions in the mid-Continent Gulf of Mexico and western regions developed andwell-capitalized independents such as Gulf Oil Union Oil the Texas Company SunOil Phillips and Cities Service provided competition By 1920 Standardrsquos share ofre ned petroleum products had fallen to 50 percent but this decline was simply anextension of an earlier trend (Comanor and Scherer 1995 Williamson et al1963) In addition the breakup of Standard into a large number of separatecompanies did not dilute the Rockefeller familyrsquos control over the new entitiesThus Burns (1977) concludes that the stock market interpreted the Standard Oildecree as ldquobenignrdquo The decree might have promoted competition had it beenimposed before 1900 but by 1911 the oil industry was much more competitive andthe decree had little effect

American TobaccoThe American Tobacco Company produced little and regular cigars plug

and smoking tobacco snuff and cigarettes By 1910 it accounted for at least75 percent of US sales of each product except for its smaller share of regularcigars Organized as a trust it obtained its market position by acquiring rmssuch as Union Tobacco Company and the Continental Tobacco Company andby aggressive pricing behavior which allegedly often resulted in prices belowmanufacturing costs (Tennant 1950) In 1908 the federal government ledand won a Sherman Act case against American Tobacco that sought to dissolvethe trust After the Supreme Court found that the trial courtrsquos initial dissolution

$0

$5

$10

Crude Oil

1889

1901

1903

1905

1907

1909

1911

1913

1915

1917

1919

1921

1923

1925

Decreeenacted

Kerosene

Gasoline

$15

$20

$25

$30

$35

Rea

l Pri

ce (

1967

$

bbl)

Notes Gasoline and kerosene prices are de ated by the Consumer Price Index for all urbanconsumers Crude oil prices are de ated by the GNP de atorSources Williamson et al (1963) US Bureau of the Census Historical Statistics of the United StatesColonial Times to 1970 Bicentennial Edition 224 593-594 (US Department of Commerce 1975)Bureau of Labor Statistics internet

Figure 1Real Petroleum Product Prices 1899ndash1925


remedy was extreme the court entered a decree in United States v AmericanTobacco (221 US 106 [1911]) that divided cigarette production into threeseparate parts American Tobacco kept assets that accounted for roughly37 percent of US production P Lorillard had 15 percent and a new companyLiggett and Myers was provided with assets to produce brands that accountedfor 28 percent of output Assets devoted to plug and smoking tobacco and cigarswere divided similarly

However the effect of restructuring the tobacco industry into a three- rmoligopoly was to unleash a battle for market share through advertising not price(Tennant 1950) Real cigarette prices were essentially stable in the few yearspreceding and following the decree and they rose several years later in response toincreases in tobacco excise taxes The breakup of American Tobacco also did notaffect the price paid to farmers for tobacco Absent price competition the three- rm oligopoly was highly pro table essentially earning the same pro t rate during1912ndash1949 as the Trust earned during 1898ndash1908 The stability of the industryrsquospro t rate and the absence of any clear decline in prices after 1911 suggest that theAmerican Tobacco case did little to spur meaningful competition in this industry

AlcoaThe Aluminum Company of America (ldquoAlcoardquo) formerly the Pittsburgh Re-

duction Company took its name in 1907 and by 1909 was integrated backward intomining ore and forward into fabricating products Alcoa also controlled AluminumLimited of Canada the largest source of aluminum imports into the United Statesat the time The production of aluminum consists of mining aluminum ore (usuallybauxite) re ning the ore to extract alumina reducing alumina into aluminumingot and fabricating the ingot into mill products like sheet tube and wire In 1912the Department of Justice charged Alcoa with restraining trade and monopolizingthe aluminum industry Alcoa signed a consent decree that required it to give upits interest in its Canadian subsidiary to terminate a contract with two chemical rms whose bauxite it had purchased not to participate in any collusive agreementsor mergers and not to discriminate against any competing fabricator in the sale ofingot

But the decree did not reduce Alcoarsquos dominance of a very small market thatwith economies of scale could probably support only one supplier By the late1930s Alcoarsquos primary production and imports still constituted 90 percent of thesupply of aluminum in the United States In 1937 the Department of Justice leda Sherman Act civil suit again charging Alcoa with monopolizing the aluminummarket and restraining trade The government appealed the District Courtrsquos ldquonotguiltyrdquo verdict to the Supreme Court which could not muster a quorum becausemany justices had previously worked on the case Legislation was enacted to allowthe three senior judges of the Circuit Court of Appeals with territorial jurisdictionto serve as the ultimate appellate court In United States v Aluminum Company ofAmerica 148 F2d 416 (2d Cir 1945) Judge Learned Hand reversed the lowercourtrsquos decision concluding that Alcoa had monopolized the market for primaryaluminum and had engaged in a price squeeze from 1925 to 1932 by selling some


aluminum sheet at prices that were too close to the price of primary aluminumingot to allow independent fabricators to achieve adequate margins on their salesof aluminum sheet Judge Hand did not rest his opinion on this violation butidenti ed it as a major problem to be dealt with in designing a remedy

The nal decree was postponed until after World War II during which thegovernment had constructed plants for alumina reduction aluminum smelting andfabrication Crandall (2001) provides empirical evidence that the decree had noeffect on real aluminum prices and little effect on the margin between fabricatedaluminum products and primary aluminum After the war virtually all of thegovernmentrsquos aluminum properties were assigned to Reynolds Metals and Kaiser(then Permanente Metals Corporation) thus creating two viable competitors In1950 the District Court ruled against Alcoarsquos divestiture but the court retainedjurisdiction over the case for ve years in the event that the two new competitors didnot provide suf cient competition Three additional companies entered the pri-mary aluminum market between 1950 and 1955 again with government assistanceand in 1956 District Judge Cashin found suf cient evidence of competition andruled against another ve-year test3

The failure of the rst decree in 1912 to erode Alcoarsquos monopoly positionderived from the small and even declining market for aluminum that by theearly and mid-1930s amounted to fewer than 150000 tons per year In contrastthe second decree in 1945 required little of Alcoa because government pro-grams dispersed production facilities to new entrants When annual demand foraluminum grew in the 1940s and 1950s to more than 125 million tons it is quitelikely that more rms would have entered the market even without governmentassistance Given that Alcoa could not control the supply of the two mostimportant inputs to aluminum production bauxite and electricity it is dif cultto conclude that it could have blocked entry after World War II Moreover themarket was suf ciently large so that Alcoa did not exhibit the characteristics ofa natural monopoly By 1955 Alcoarsquos market share was less than half of what itwas when the government led its 1937 lawsuit yet its output was more thanfour times greater

ParamountThe motion picture industry is composed of movie studios lm distributors

and theatres During the 1930s some distributors owned theatre chains Thedefendants in the Paramount case initially brought in 1938 were ve majordistributors that owned theatres and three ldquominorrdquo distributors which togethercontrolled 95 percent of total lm rentals in the early 1940s (Conant 1960) In1946 a US District Court found that the distributors had engaged in severalpractices that violated the Sherman Act including xing admission prices andrestricting output to competing theatres through tying arrangements and ldquoformula

3 The court reporter numbers for the key decisions in the Alcoa case include United States v AluminumCompany of America 44 F Supp 97 (SDNY 1941) United States v Aluminum Company of America 148 F2d416 (2d Cir 1945) United States v Aluminum Company of America 91 F Supp 333 (SDNY 1950)


dealsrdquo The District Courtrsquos decree did not order divestiture but prohibited agree-ments to maintain uniform prices and required a system of competitive biddingamong theatres for each run of a feature lm The US Supreme Court howeverfound the bidding system unworkable and in United States v Paramount Pictures (334US 131 [1948]) it ordered the lower court to reconsider divestiture By the early1950s the ve major distributors had completely divested their theatre chains

The primary objective of the decree was to force distributors to compete fortheatre space by offering attractive terms for renting their lms Independentdistributors would presumably have better access to theatres and new distributorsmight even enter Under this scenario admission prices would fall and the numberof lm distributors and annual lm releases would increase In fact the average realprice of a movie ticket rose in the two decades following the Paramount decisionspeci cally the Consumer Price Index for indoor theatres rose 364 percentbetween 1948 and 1958 while the overall CPI rose just 201 percent The trendcontinued during 1958ndash1967 with the CPI for indoor theatres rising 689 percentwhile the overall CPI rose just 155 percent In addition little entry occurred intomotion picture distribution Twenty years after the Paramount litigation seven ofthe original eight defendants accounted for nearly three-fourths of all US theat-rical rentals (Crandall 1975)

Two interpretations are possible Either the defendantsrsquo original actions werenot raising ticket prices and restricting output in which case the antitrust suitshould not have been led or the decree failed to end collusive behavior Afundamental problem in analyzing the postdecree market is evaluating how theintroduction of television affected theatrical admissions which declined dramati-cally New entrants and independents may have fared poorly under these marketconditions and after decades of agreeing on clearances and lengths of runsthe Paramount defendants may have been able to coordinate a cartel agreementby reporting their weekly revenues from each theatre to the trade press Distribu-torsrsquo share of theatrical admission receipts rose from 304 percent in 1948 to458 percent in 1967 Thus distributors captured approximately two-thirds of the66 percent increase in real ticket prices during this period

United Shoe MachineryUnited Shoe Machinery manufactured a full line of machines used to

produce shoes By the 1940s USM offered more than 300 types of machines ofwhich a shoe manufacturer might need as many as 100 to produce a shoe(Masten and Snyder 1993) USM sold and leased its machines and providedrepair and advisory services In 1949 its market share of major machines was91 percent and its share of minor machines was 64 percent (Kaysen 1956) Thegovernment claimed that USM had monopolized the shoe machinery marketthrough leases that impeded the purchase or lease of its competitorsrsquo machinesand prevented the development of a secondhand market Exclusionary provi-sions of USMrsquos leases included ten-year terms and a ldquofull capacityrdquo clause thatrequired lessees to use each machine to the fullest extent possible (Masten andSnyder 1993) USM would charge shoe manufacturers with violating this clause


if they switched to a competitorrsquos machine but waived the penalties if thecancellation was caused by changes in demand conversion to manual opera-tions or replacement with another USM machine

In United States v United Shoe Machinery (110 F Supp 295 [DMass 1953] affrsquod347 US 521 [1954]) the US Supreme Court upheld a lower court decision thatUSM had illegally monopolized the shoe machinery market The trial court de-clined to order the dissolution of USM but structured a decree that prohibitedUSM from designing its lease and sales terms to make it substantially more advan-tageous to lease machines In addition the duration of all new leases had to bereduced to ve years or less with an option to return machines after one yearReturn charges or deferred payments were banned The decree was intended toincrease competition by encouraging the purchase of machines thus creating avibrant secondhand market and inducing shoe manufactures to be more receptiveto machines offered by USMrsquos competitors

The decree did succeed in establishing a secondhand market for machines andreducing USMrsquos market share from roughly 85 percent in 1953 to 62 percent in1963 (Parrish 1973) On the other hand USMrsquos revenue gains were more thantwice the sum of its four major competitorsrsquo gains and its return on equityremained relatively constant The heterogeneity of shoe machinery prevents adirect assessment of shoe machinery prices before and after the decree Howeverif the decree succeeded in reducing machinery prices it is highly likely that shoemanufacturers would have incurred lower machinery expenses relative to the valueof shoes produced But based on data from the Census of Manufacturers the ratio ofthe value of shoe machinery shipments to the value of shoe shipments remained at0012 between 1954 and 1967 (It is conceivable that the stability of relativeshipment values could have re ected lower machinery prices and a substitution ofmachinery for labor in shoe production technology during this period but noevidence exists to support this conjecture)

In any event the US Supreme Court was not satis ed that suf cient compe-tition had developed in the shoe machinery market because following a review ofthe decree it recommended in 1969 that the lower court consider ldquomore de nitivemeansrdquo to achieve competition As a result USM was forced to divest itself ofroughly one-third of its remaining shoe machinery operations Unfortunately thegovernment required structural relief only after the shoe industry had entered asteep decline because of the rise in imported shoes It has even been speculatedthat the USM decree accelerated the demise of US shoe manufacturing but weare not aware of evidence to support this conclusion

ATampTIn 1974 the US Department of Justice brought a monopolization case against

ATampT which eventually led to a 1982 consent decree that divested ATampT of itslocal operating companies creating in 1984 seven regional Bell companies thatprovide local phone service ATampT retained its long distance operations and atelephone equipment company that is now called Lucent Following the breakuplong distance telephone competition dramatically increased and rates fell so there


is at least some prima facie evidence that consumers bene ted from this monopo-lization case

But on closer examination the rise in competition and lower long distanceprices are attributable to just one aspect of the 1982 decree speci cally a require-ment that the Bell companies modify their switching facilities to provide equalaccess to all long distance carriers The Federal Communications Commission(FCC) could have promulgated such a requirement without the intervention of theantitrust authorities For example the Canadian regulatory commission imposedequal access on its vertically integrated carriers including Bell Canada in 1993 Asa result long distance competition developed much more rapidly in Canada thanit had in the United States (Crandall and Hazlett 2001) The FCC however wastrying to block MCI from competing in ordinary long distance services when theATampT case was led by the Department of Justice in 1974 In contrast to Canadianand more recent European experience a lengthy antitrust battle and a disruptivevertical dissolution were required in the US market to offset the FCCrsquos anticom-petitive policies Thus antitrust policy did not triumph in this case over restrictivepractices by a monopolist to block competition but instead it overcame anticom-petitive policies by a federal regulatory agency

Overall Lessons and Recent Monopoly CasesThis brief overview of landmark monopolization cases suggests several rea-

sons why such cases have often failed to increase competition to the bene t ofconsumers

One problem is the protracted length of these cases which often take so longthat industry competition has changed before the remedy is implemented as inStandard Oil and Alcoa This problem has also arisen in modern monopolizationcases like those involving IBM and Microsoft The rst monopolization case againstIBM was brought in 1952 and settled by consent decree in 1956 but there islittle evidence that it had favorable effects on competition in the computer indus-try which was rapidly replacing tabulating machines with mainframe computers(Wilder 1975) IBM quickly vaulted to a dominant position in mainframes leadingthe Department of Justice to le another case in 1969 That case was droppedin 1982 in no small part because the market had changed once again (FisherMcGowan and Greenwood 1983) The ultimate merits of the Microsoft case are notyet clear but it has already required six years of litigation (excluding the FTCrsquosearlier investigation) and the courtrsquos nal judgment is still being appealed By thetime it is resolved the information technology market is likely to have changedsubstantially

Another major problem occurs when a monopolization case simply fails tobene t consumers because the remedy turns out to have a negligible practicalimpact as may have happened in American Tobacco Paramount and United ShoeMachinery Recently a number of monopoly cases like those led against Safewayand AampP were brought in an attempt to stop the replacement of small grocerystores by large national food chains but these cases have had little effect on market


concentration because they could not prevent more ef cient chains from replacingless ef cient small retailers (Crandall and Elzinga 2002)

Similarly airlines that dominate hub airports have been accused of havingmonopoly power and in some cases of engaging in predatory pricing behavior toprotect hub markets In 1999 the Department of Justice led a predatory pricingsuit against American Airlinesmdashbut lost on summary judgment Morrison andWinston (2000) cast doubt on the claim that airlines are successfully engaging inpredatory behavior They also show that fares may be higher on hub routes than onother routes because a hub carrier has market power or because low-cost SouthwestAirlines mainly serves nonhub routes and signi cantly depresses fares in thesemarkets In any case the cost to travelers from a hub ldquopremiumrdquo is clearly offset byhub bene ts including greater ight frequency and agglomeration economies inareas surrounding the airport

Challenging large rms in court is often politically popular but neitherpolicymakers nor economists have yet to offer compelling evidence of markedconsumer gains from antitrust policy toward monopolization

Collusion

Explicit agreements to x prices are often treated by the antitrust authoritiesand the courts as per se violations which means that evidence of an agreement issuf cient to prove guilt A wide variety of other restrictive practices are potentiallycollusivemdashincluding exclusive contracts exclusive territories and others Thecourts have generally adopted a ldquorule of reasonrdquo standard for these practices whichmeans that they are judged on a case-by-case basis with earlier precedents in mind4

The Department of Justice investigates about 100 allegations of price xing a yearand often proceeds with indictments

Retrospective assessments of some of these cases have failed to nd muchdirect bene t from curbing alleged instances of collusion (Besides price xingvery few empirical studies exist of cases involving collusive practices) For exampleNewmark (1988) found that an antitrust indictment of bakers in Seattle had noeffect on the price of bread and Morrison and Winston (1996) concluded that aconsent decree that prohibited airlines from announcing the ending dates of theirfare promotions had no effect on fares More systematically Sproul (1993) analyzeda sample of 25 price xing cases between 1973 and 1984 for which usable price datawere available He argued that if a cartel succeeds in raising prices then prosecu-tion should lower them However he found that controlling for other in uences

4 Under resale price maintenance agreements for example a producer of a product sets a price that theretailer may not undercut The procompetitive argument for such agreements is that they encourage theretailer to invest in knowledge and service about the product The likelihood of prosecution in suchcases was substantially reduced by a 1997 Supreme Court decision (State Oil Company v Khan 522 US3 [1997]) Also Ippolito and Overstreet (1996) provide some evidence that resale price maintenanceproduced ef ciency gains


prices rose an average of 7 percent four years after an indictment Sproul also foundthat prices rose on average even if one uses a starting point during the investiga-tion but before the indictment Even in the most successful cases prices fell only10 percent

One possible explanation for why these cases have not generally resulted inprice declines is that the Department of Justice may in some instances be prose-cuting rms that are engaging in activities that involve other goals besides raisingprices For example Sproul (1993) suggests that a cartel may reduce costs throughshared advertising and research which may tend to reduce prices rather than toincrease them Another possibility is that a cartel may be pursuing distributionalgoals For instance MIT and Ivy League colleges established a tradition of coordi-nating their need-based nancial aid decisions The schools claimed that theso-called Overlap process enabled them to concentrate their scarce nancial re-sources on needy students without affecting their total revenues The governmentsued claiming that the schools were conspiring on nancial aid policies to reduceaid and raise revenues Carlton Bamberger and Epstein (1995) found that theprocess did not have a statistically signi cant effect on the average ldquopricerdquo paid perstudent but that it prevented the ow of school resources from lower- to higher-income students Hoxby (2000) corroborates this nding

To be sure there are well known examples where rms have clearly colludedto raise prices including recent cases involving lysine citric acid and vitaminsHowever researchers have not shown that government prosecution of allegedcollusion has systematically led to signi cant nontransitory declines in consumerprices

Mergers

Department of Justice and Federal Trade Commission investigations of pro-posed mergers absorb more than half of federal antitrust resources The Hart-Scott-Rodino Antitrust Improvement Act of 1976 requires any rm valued over$100 million to le a premerger noti cation under various conditions the mostcommon of which is that it plans to merge with another rm valued at more than$50 million After ling the noti cation rms must wait 30 days before they canproceed with the merger During this period the FTC or the Department of Justicecan request additional time and information (known as a ldquosecond requestrdquo) beforedeciding whether to approve or oppose the merger

Mergers may harm or bene t consumers Mergers that enable rms to acquiremarket power may only raise consumer prices while mergers that enable rms torealize operational and managerial ef ciencies can reduce costs and thereby lowerprices Economists generally conclude that taken as a group mergers are notanticompetitive Andrade Mitchell and Stafford (2001) argue that mergersthrough the 1990s have produced ef ciency improvements leading to a modest1 percent gain in postmerger operating margins Carlton and Perloff (1994) claimthat the increase in shareholder value from a merger in the United States is not


typically due to the creation of market power But even if one accepts that theaverage merger results in an ef ciency gain antitrust enforcement could be goodor bad depending on how well the antitrust authorities distinguish procompetitivemergers from anticompetitive ones

How can a researcher sort out whether the mergers that are blocked or thathave conditions attached by the Department of Justice or the FTC are the ones thatwould have led to anticompetitive outcomes and welfare losses With a monopolyone can observe its impact on consumers before and after antitrust action But ablocked merger is never observed and thus its effects cannot be compared directlyto what would have happened if the merger had been allowed This dif culty helpsto explain why we could not nd any case studies that showed that the FTC orDepartment of Justice prevented signi cant welfare losses by blocking or attachingconditions to a proposed merger5

One approach to investigating whether the antitrust authorities can distin-guish good from bad mergers is to look at stock price data which is presumablyforward looking to test the hypothesis that horizontal mergers challenged by thegovernment would have created market power in the defendantsrsquo industries Thisis done by estimating whether proposed merger-induced changes in expectedfuture product and factor prices translate into positive abnormal stock returns to rms competing in the same industry as well as to the merging rms Eckborsquos(1992) conclusion from this literature is that the mergers that were challengedwere not anticompetitive and in all likelihood would have been ef cient had theybeen allowed to go through

Another approach is to consider whether the reporting requirements of theHart-Scott-Rodino Act of 1976 have enabled the antitrust agencies to judge amergerrsquos competitive impact better before ling a complaint Eckbo and Wier(1985) use stock price data to analyze merger cases led after 1978 and nd thatthe proposed mergers would not have harmed competition Thus they con-clude that the act has not helped the agencies improve their case selectionrecord

Still another approach is to look at mergers that were challenged or opposedby the antitrust regulators but were consummated anyway Such mergers haveoften worked well for consumers For example the FTC unsuccessfully challengedWeyerhaeuserrsquos acquisition of Menasha which led to a decline in corrugated boxprices (Schumann Reitzes and Rogers 1997) Similarly the Department of Justiceopposed airline mergers between TWA and Ozark and between Northwest andRepublic However the Department of Transportation allowed the two mergers

5 Pittman (1990) estimates that the Santa FeSouthern Paci c rail merger which was opposed by theDepartment of Justice and blocked by the Interstate Commerce Commission would have led to annualoperating cost savings by the carriers but deadweight losses of roughly $100 million Southern Paci chowever had failed to become ldquorevenue adequaterdquo and probably could only survive with a mergerIndeed it subsequently merged with Union Paci c which led to disastrous service disruptions in thesouthwest that cost shippers billions of dollars In any case many observers of the rail industry envisionthat the ldquo nal frontierrdquo of the industry is for the two remaining railroads in the East and the two in theWest to form two ef cient transcontinental railroads (Grimm and Winston 2000)


Morrison (1996) conducted a long-run analysis that improved upon previousairline merger assessments by considering fares well before and up to the mergerand fares immediately and several years after the merger He found that theTWA-Ozark merger led to a 15 percent decline in fares and that the Northwest-Republic merger led to a 2 percent increase in fares which may have been offset bybene ts from greater route coverage

We now turn to a broad assessment of recent merger policy based on price-costmargins across industries Although there are well known measurement concernswith using price-cost margins greater market power should increase them ceterisparibus We also recognize that using interindustry data to explain price-cost mar-gins can be problematic But this line of research has matured to the point whereit has produced a set of ldquostylized factsrdquo about industry competition (Schmalensee1989) Our hope is that the suggestive ndings from this exercise will be viewed incombination with other researchersrsquo ndings about the effects of antitrust mergerpolicy rather than dismissed on doctrinal grounds

For our dependent variable we use price-cost margins from 1984 to 1996 forthe 20 manufacturing industries that are de ned at the two-digit SIC level (usingthe pre-1997 classi cation system) We choose this time period and sample basedon data availability Outcomes of merger cases are available back to 1982 Howeverwe will specify merger enforcement variables with two-year lags (see below) thus wecan analyze price-cost margins only as far back as 1984 In addition case outcomesare publicly available only at the two-digit level of aggregation while consistentestimates of industry price-cost margins are available only for manufacturingindustries

In our regression price-cost margins are assumed to be in uenced by court-based outcomes second requests for information and industry characteristics Thecourt-based outcomes we include are the number of successful and unsuccessfulmerger challenges as well as the number of consent decrees reached by thegovernment and the rms proposing to merge In a given year the vast majority ofthese court-based outcomes are consent decrees during the period covered by oursample there were nine cases that went to a verdict and 88 cases settled by aconsent decree Our sample also contains 368 second requests for informationwhich may have discouraged some of the proposed mergers from moving forwardEach case is only counted once even if there were multiple decisions An industryis not likely to experience the effect of antitrust merger policy immediately thusthe estimation is based on two-year lags for the court-based outcomes and secondrequests Following previous speci cations like that of Salinger (1990) we includethe following industry characteristics the import-sales ratio to control for foreigncompetition the capital-sales ratio to control for technology and the growth of thenumber of rms in an industry with a ve-year lag (because this lag provided thebest statistical t) to control for entry6

6 Of course we experimented with this speci cation in various ways For example using one-year lagsand no lags had little effect on the main ndings Our ndings did not change when we speci edcourt-based outcomes and second requests as a percentage of the total mergers proposed in an industry


If antitrust interventions against mergers are bene ting consumers price-costmargins in an industry should fall from what they would have been when thegovernment successfully challenges a merger in court or negotiates a consentdecree Second requests for information may also lower prices by discouraginganticompetitive mergers from moving forward If antitrust investigations are focus-ing on mergers that primarily have ef ciency effects price-cost margins should risefrom what they would have been when the government successfully challenges amerger in court or negotiates a consent decree because the merger as proposedwould have reduced rmsrsquo costs7

Our results are presented in Table 2 The parameter estimates of theindustry characteristics are plausible A higher import-sales ratio and rmgrowth reduces an industryrsquos price-cost margin as does an increase in anindustryrsquos capital-sales ratio Salinger (1990) found that the capital-sales ratiohad a positive effect on price-cost margins during the 1970s but that its effectbecame negative during the early 1980s This negative coef cient persistedduring the 1980s downturn and expansion the negative coef cient in Table 2is consistent with this nding

The coef cients of the court-based outcomes are of central interest andsuggest that merger enforcement policy is primarily undermining mergers thatwould enhance ef ciency rather than protecting competition We nd that asuccessful merger challenge does have a negative effect on the price-cost marginbut that the effect is not statistically signi cant In contrast an unsuccessful chal-lenge in which a court eventually allows the proposed merger is associated with adecline in price-cost margins and the effect is statistically signi cant The mostoptimistic interpretation to place on these ndings is that potential challengesfrom antitrust authorities succeed in blocking or discouraging mergers that wouldreduce welfare and that the courts do not allow the regulators to block mergersthat improve economic welfare However we believe that a more plausibleinterpretation consistent with the ndings reported earlier in the section and thestatistically insigni cant effect of second requests is that the mergers blocked byantitrust authorities have no signi cant effect on price-cost margins in those

in a given year They were also not affected when we speci ed separate coef cients for interventions bythe Department of Justice and the FTC We tried using industry xed effects to control for unmeasuredindustry characteristics but the parameters for the court-based outcomes and second requests were notaffected if the xed effects were excluded from the speci cation thus they are not included here It ispossible that merger policy could in uence the rate of entry thus we estimated a model that droppedthis variable but found that this speci cation did not affect the parameters for the merger policyvariables so we kept it in the speci cation We also estimated models that controlled for several otherpotential in uences on the price-cost margin including macroeconomic variables (unemploymentinterest rates GDP growth) year xed effects industry output growth selected commodity dummiesand a time trend but these variables were statistically insigni cant7 If antitrust enforcement were fully optimal and complete then all the enforcement variables shouldbe statistically insigni cant because the Department of Justice and FTC would have thwarted allanticompetitive attempts to raise price-cost margins and not thwarted mergers that would have loweredprice-cost margins The preceding summary of evidence suggests that it is extremely unlikely that mergerpolicy has been optimal


industries because the regulators are not sorting out good mergers from bad oneswith much accuracy Further the negative and statistically signi cant coef cient ofunsuccessful court challenges suggests that the antitrust authorities overreach andattempt to block productive mergers although only a handful of merger casesactually reach a court verdict

When the government and the potential merger partners reach a consentdecree to gain regulatory approval for the merger price-cost margins in theindustry subsequently increase In our data the FTC and DOJ negotiated45 percent of their consent decrees with companies that at that time were intwo-digit industries located in the upper quintile of price-cost margins This ndingcan be interpreted either as an argument that the antitrust authorities should have

Table 2Price-Cost Margin Parameter Estimates(robust standard errors in parentheses)

Variable Coefcient

Court-Based OutcomesMergers successfully blocked by FTC or DOJ (2-year lag) 20040 (0032)Mergers unsuccessfully challenged by FTC or DOJ (2-year lag) 20038a (0011)Consent decrees (2-year lag) 0017a (0004)

Other OutcomesSecond request for information made by FTC or DOJ (2-year lag) 20001 (0002)

Industry CharacteristicsImport-sales ratio 20071a (0020)Log of the growth of the number of rms (5-year lag) 20721a (0188)Capital-sales ratio 20105a (0008)Constant 0518a (0018)R2 045Number of observations 260

a Statistically signi cant at the 1 percent levelNotes The price-cost margin variable is constructed following standard practice as (value added 1D inventories 2 payroll)(value of shipments 1 D inventories) Data for each of the components wereobtained from the Annual Survey of Manufacturers published by the Bureau of the Census for 1984 to1996For the import-sales ratio from 1984 to 1996 total imports were obtained from Robert Feenstra whoassembled data from the US Department of Commerce and the US Industry and Trade Outlook Salesdata reported as shipments were from the Annual Survey of ManufacturersGrowth of the number of rms was obtained from the Economic Census published every ve years by theBureau of the Census and from the annual County Business Patterns (CBP) also published by the CensusThe Economic Census contains rm data while the CBP contains plant data that were used to estimate thenumber of rms The ratio of plants to rms in the ldquobenchmarkrdquo years of 1977 1982 1987 and 1992 wasused to generate an estimate for the growth of the number of rms on an annual basisFor the capital-sales ratio capital is measured as the historical cost of the net stock of xed private capitaland is from Fixed Reproducible Tangible Wealth published by the Bureau of Economic Analysis within theDepartment of Commerce For sales see aboveData on the number of mergers successfully challenged in court mergers unsuccessfully challenged incourt consent decrees and second requests are from the Hart-Scott-Rodino Annual Reports which areannual reports to Congress prepared jointly by the FTC and the Antitrust Division of the DOJ Courtoutcomes were described in each report and the SIC codes for the companies involved in the cases weredetermined by consulting FTC and DOJ case histories


negotiated stronger conditions to address potential anticompetitive problems orthat the consent decrees allowed mergers to go forward only when the rms weresaddled with conditions that compromised production ef ciencies Neither inter-pretation is complimentary to the antitrust authorities8

We do not want to overstate our con dence in the speci c estimated coef -cients from Table 2 It would clearly be preferable to have more disaggregated datafor more industries As we have noted the ndings can be interpreted in variousways There are of course counterexamples of individual mergers that have raisedprices (for example Barton and Sherman 1984) But the regression results are notbiased in any particular direction and are broadly consistent with the other empir-ical evidence that we have surveyed We can only conclude that efforts by antitrustauthorities to block particular mergers or affect a mergerrsquos outcome by allowing itonly if certain conditions are met under a consent decree have not been found toincrease consumer welfare in any systematic way and in some instances the inter-vention may even have reduced consumer welfare

Deterring Anticompetitive Behavior

Given the lack of direct evidence that antitrust actions on monopolizationcollusion and mergers have promoted competition and bene ted consumerssupporters of an activist antitrust policy are left with the argument that such policydeters rms from anticompetitive behavior If the authorities had not prosecutedIBM ATampT Microsoft and others who knows what abuses would have occurredAdmittedly providing evidence on what has been deterred and therefore did nothappen is a dif cult task In any event we have not found any evidence thatantitrust enforcement has deterred rms from engaging in actions that would haveseriously harmed consumers

Historically it has been suggested that government victories in Standard Oiland American Tobacco deterred other companies such as US Steel from pursuingsimilar paths to monopoly power However Comanor and Scherer (1995) concludethat US Steelrsquos failure to maintain its large share of the countryrsquos steel output inthe rst half of the twentieth century was due to its high costs not to a concertedeffort to avoid antitrust prosecution

International evidence has been used to assess the deterrence effect of theantitrust laws Stigler (1966) compared concentration in speci c industries inEngland which at the time did not have a public policy against concentrationof control with the same industries in the United States and concluded that the

8 It is possible that the mergers may have involved antitrust markets within a given two-digit industry thathad price-cost margins that were quite different from a two-digit industryrsquos average price-cost margin forexample a merger may have occurred within a relatively concentrated subindustry of a relativelyunconcentrated industry Because we control for other systematic in uences on two-digit industryprice-cost margins our methodology should uncover the impact of merger policy albeit with asomewhat diluted effect The extent of this dilution however is not clear after all we do nd that twoof the four merger policy variables had statistically signi cant effects


Sherman Act has had a very modest effect in reducing US concentrationEckbo (1992) explored whether the antitrust laws deter potentially anticom-petitive mergers by estimating whether the probability that a horizontal mergeris anticompetitive was higher in Canada where until 1985 mergers were essen-tially unconstrained than in the United States His analysis compared estimatedparameters in cross-section models that explained announcement stock returnsto merging rms and their nonmerging industry rivals as a function of industryconcentration in the two countries Based on this comparison he rejected thehypothesis that the US antitrust laws are deterring anticompetitive mergers

Although we have not found any evidence that the antitrust laws have hadbene cial deterrence effects we suspect that such effects exist However anydeterrent effect of the antitrust laws may be relatively small compared with the welldemonstrated ability of competitive markets to deter anticompetitive monopoliescollusion and mergers We have identi ed a few of the many instances whereerstwhile monopolies have seen their market shares eroded by new competitorsStandard Oil US Steel Alcoa and IBM for example Moreover collusion among rms is more dif cult than it may appear Stigler (1964) pointed out that evenwhen few rms compete in a market it may be dif cult for them to reach aconsensus on price and market shares and even if they do they may not be able todiscourage cheating

Empirical evidence from the rail airline ready-to-eat cereal and brewingindustries illustrates some of the ways that markets prevent rms from success-fully colluding Beginning in the mid-1980s electric utilities that received coalshipments from the Powder River Basin in Wyoming were served by only tworailroads Many economists would expect that the two carriers would be able tocome to some arrangement that elevates rates above competitive levels How-ever Gaskins (2001) found that rail rates in the Powder River Basin approachedlong-run marginal costs suggesting that carriers were not colluding on pricesIt seems that shippers are able to play one railroad off against another whennegotiating long-term contracts to reduce their rates because if a carrier doesnot compete ercely for a shipperrsquos traf c it may have to wait several yearsbefore it has an opportunity to recapture any traf c that it loses (Grimm andWinston 2000)

In April 1992 the president of American Airlines Robert L Crandallattempted to introduce some discipline in airline pricing by urging othercarriers to adopt Americanrsquos pricing regimen of four basic fares and reducedfull-fare coach and rst-class fares But Americanrsquos in uence was too limited toget other carriers to follow its lead (Morrison and Winston 1995) By October1992 Crandall abandoned the strategy bemoaning ldquoWe tried to provide someprice leadership but it didnrsquot work so we are back into the death by a thousandcutsrdquo (Lollar 1992)

In contrast to railroads and airlines the ready-to-eat cereal and brewingindustries are characterized by persistently high price-cost margins Economistshave explored whether market power in these industries is attributable to collusivepricing behavior but have rejected this explanation Cereal rms (Nevo 2001) and


brewers (Baker and Bresnahan 1985) have engaged in nonprice competitionparticularly through advertising to in uence the perceived quality of their prod-ucts and to elevate price-cost margins Indeed rms that produce differentiatedproducts face less incentive to engage in and nd it more dif cult to maintaincollusive agreements than rms that produce homogeneous products

There is a widespread belief that the antitrust laws deter collusion more thanthey deter attempts to monopolize Firms and individuals convicted of price xingare subject to federal criminal penalties and also vulnerable to private suits fortreble damages Block Nold and Sidak (1981) provide evidence that such classactions are the strongest deterrence against collusion It is possible that the De-partment of Justice has succeeded in deterring the most serious instances of price xing and has therefore been increasingly prosecuting marginal cases but thissurmise has not been documented Recently the Antitrust Division of the Depart-ment of Justice has attempted to strengthen deterrence by imposing higher neson corporations for price xing and expanding the use of corporate leniency for rms that disclose their role in a conspiracy and cooperate with the governmentHowever Kobayashi (2002) develops a model of optimal deterrence and cautionsthat these actions may lead to overdeterrence which would induce excessiveinvestments in monitoring and prevention raise production costs and result inhigher consumer prices

Finally the surrounding climate of market competition is also an importantreason why most mergers are not anticompetitive Indeed Paulterrsquos (2001) surveyof the literature on mergers concludes that they ldquofailrdquo 35 percent to 75 percent ofthe time where failure is determined by survival pro tability retention of assetsand so on Because of internal and external market forces mergers have much lesspredictable outcomes than do most other business investments It is also notewor-thy that although the US economy experienced major waves of large mergersduring the 1980s and 1990s aggregate concentration has not increased over thepast two decades (White 2002)

Most of US industry is structurally competitive For example Pashigian(2000) used a government task forcersquos de nition of an imperfectly competitivemarket as one with a four- rm concentration ratio above 70 percent and found thatin 1992 only 46 out of 398 four-digit US manufacturing industries met thisthreshold9 In a competitive climate monopolies will tend to be eroded collusive

9 This theme that the market is largely competitive is compatible with the common nding that the USeconomy has experienced only a small deadweight loss from noncompetitive pricing Harbergerrsquos(1954) initial nding of a deadweight loss of roughly 01 percent of GDP has been revisited by severalauthors Cowling and Mueller (1978) found a much larger deadweight loss than other researchersbecause they included advertising expenditures as part of welfare losses More recent estimates sum-marized by Ferguson (1988) indicate a deadweight loss of about 1 percent of GDP These estimates ofdeadweight loss are not fully appropriate for our purposes however Our focus is on consumer bene tswhich would involve transfers from consumers to rms not just on deadweight loss Moreover theestimates of losses from imperfect competition include distortions caused by government interventionssuch as regulations and trade protection but do not include possible offsetting dynamic bene ts ofimperfect competition such as greater investments in RampD that lead to enhanced product quality anddesign


agreements will fall apart and mergers will either provide ef ciency bene ts or failAny additional deterrence antitrust policy provides should be evaluated in thiscontext1 0

Conclusion

The apparent ineffectiveness of antitrust policy stems from several causes1) the excessive duration of monopolization cases which portends that the partic-ular issue being addressed will evolve into something differentmdashoften of lessimportancemdash by the time it is resolved 2) the dif culties in formulating effectiveremedies for monopolization and effective consent decrees for proposed mergers3) the dif culties in sorting out which mergers or instances of potentially anticom-petitive behavior threaten consumer welfare 4) the substantial and growing chal-lenges of formulating and implementing effective antitrust policies in a neweconomy characterized by dynamic competition rapid technological change andimportant intellectual property (Carlton and Gertner 2002) 5) political forces thatin uence which antitrust cases are initiated settled or dropped (Weingast andMoran 1983 Coate Higgins and McChesney 1995) including situations where rms try to exploit the antitrust process to gain a competitive advantage over theirrivals (Baumol and Ordover 1985) 6) the power of the market as an effective forcefor spurring competition and curbing anticompetitive abuses which leaves antitrustpolicy with relatively little to do

We recognize that antitrust doctrines have changed and continue to changeover time (Baker 2002) Our concern is that these changes have not been moti-vated and guided by empirical assessments that identify which policies have andhave not succeeded in increasing consumer welfare

We also believe however that the evidence presented here would be moreextensive and persuasive if researchers had greater access to potentially informativesources of data and employed the latest empirical developments in industrialorganization The Department of Justice and the FTC could help advance ourknowledge of the effects of antitrust policy by making more data generated by casesavailable to researchers Indeed we were restricted to using two-digit industryclassi cations for court-based and other outcomes in mergers even though theantitrust authorities have this information at a more disaggregated level Baker andRubinfeld (1999) survey models that economists have developed to analyze price xing mergers and oligopoly conduct but fail to identify a single instance where

10 In his response to this paper Baker tries to advance the argument that antitrust policy has signi cantdeterrence effects But he fails to acknowledge that the in ux of foreign competition deregulation theentry of new rms and the emergence of new technologies has created an extremely competitiveenvironment for contemporary US industry Indeed Bakerrsquos evidence regarding deterrence is mainlydrawn from episodes that predate the current intensity of industry competition Moreover the antitrustauthorities may deter rms from actions that either increase or decrease social welfare Baker fails toprovide a balanced quantitative assessment of the effects of deterrence so we have no feel for the impactor even the sign of this component of antitrust policy


any of these models has been used to assess the welfare effects of antitrust policyClearly economists should make greater efforts to use such methodological tools toaid our understanding of antitrust

The present state of and gaps in our knowledge suggest a short-term andlong-term course of action Until economists have hard evidence that the currentantitrust statutes and the institutions that administer them are generating socialbene ts the Federal Trade Commission and the Department of Justice shouldfocus on the most signi cant and egregious violations such as blatant price xingand merger-to-monopoly and treat most other apparent threats to competition withbenign neglect As the antitrust research agenda evolves we envision that econo-mists may identify cases where antitrust policy has improved consumer welfare Ifthey do the long-term task will be to explain why certain policies have beencounterproductive and others helpful and to provide guidance for how antitrustresources can be con ned to bene cial activities11 A research agenda has emergedfor those who are truly interested in improving the consumer welfare effects ofantitrust policy and enforcement

y Crandall has been employed as a consultant for Microsoft and various telecommunicationcompanies A long list of people provided us with helpful comments on previous drafts We aregrateful to them and the editors for their help and to David Zipper for research assistance

References

Andrade Gregor Mark Mitchell and Eric Staf-ford 2001 ldquoNew Evidence and Perspectives onMergersrdquo Journal of Economic Perspectives Spring15 pp 103ndash20

Areeda Philip1988 Antitrust Analysis BostonMass Little Brown and Company

Baker Jonathan B 2002 ldquoA Preface toPost-Chicago Antitrustrdquo in Post-Chicago De-velopments in Antitrust Law Roger van den BerghRoberto Pardolesi and Antonio Cucinotta edsCheltenham UK Edward Elgar chapter 1

Baker Jonathan B and Timothy F Bresna-han 1985 ldquoThe Gains from Merger or Collu-sion in Product-Differentiated IndustriesrdquoJournal of Industrial Economics June 33 pp427ndash 44

Baker Jonathan B and Daniel L Rubinfeld1999 ldquoEmpirical Methods in Antitrust Litiga-tion Review and Critiquerdquo American Law andEconomics Review 11ndash2 pp 386ndash 435

Barton David M and Roger Sherman 1984ldquoThe Price and Pro t Effects of HorizontalMerger A Case Studyrdquo Journal of Industrial Eco-nomics December 33 pp 165ndash77

Baumol William J and Janusz A Ordover1985 ldquoUse of Antitrust to Subvert CompetitionrdquoJournal of Law and Economics May 28 pp247ndash 65

Bittlingmayer George 1995 ldquoOutput andStock Prices When Antitrust is Suspended TheEffects of the NIRArdquo in The Causes and Conse-quences of Antitrust Fred S McChesney and Wil-

11 Bakerrsquos response initiates an all-purpose defense of antitrust policy rather than distinguishing goodantitrust policy from that which is ineffective irrelevant or harmful


liam F Shugart II eds Chicago University ofChicago Press pp 287ndash318

Block Michael Kent Frederick Carl Nold andJoseph Gregory Sidak 1981 ldquoThe Deterrent Ef-fect of Antitrust Enforcementrdquo Journal of PoliticalEconomy June 89 pp 429 ndash45

Burns Malcolm R 1977 ldquoThe CompetitiveEffects of Trust-Busting A Portfolio AnalysisrdquoJournal of Political Economy August 85 pp717ndash39

Carlton Dennis W and Robert H Gertner2002 ldquoIntellectual Property Antitrust andStrategic Behaviorrdquo NBER Working Paper8978 June

Carlton Dennis W and Jeffrey M Perloff1994 Modern Industrial Organization Second Edi-tion New York Harper Collins

Carlton Dennis W Gustavo E Bambergerand Roy J Epstein 1995 ldquoAntitrust and HigherEducation Was There a Conspiracy to RestrictFinancial Aidrdquo Rand Journal of Economics Spring26 pp 131ndash47

Clark John B 1901 The Control of Trusts NewYork Macmillan

Coate Malcolm B Richard S Higgins andFred S McChesney 1995 ldquoBureaucracy andPolitics in FTC Merger Challengesrdquo in TheCauses and Consequences of Antitrust Fred SMcChesney and William F Shugart II edsChicago University of Chicago Press pp 213ndash30

Comanor William S and F M Scherer 1995ldquoRewriting History The Early Sherman Act Mo-nopolization Casesrdquo International Journal of Eco-nomics and Business 22 pp 263ndash89

Conant Michael 1960 Antitrust in the MotionPicture Industry Berkeley Calif University ofCalifornia Press

Cowling Keith and Dennis C Mueller 1978ldquoThe Social Costs of Monopoly Powerrdquo EconomicJournal December 88 pp 727ndash 48

Crandall Robert W 1975 ldquoThe Postwar Per-formance of the Motion-Picture Industryrdquo Anti-trust Bulletin Spring 2 pp 49ndash 88

Crandall Robert W 2001 ldquoThe Failure ofStructural Remedies in Sherman Act Monopoli-zation Casesrdquo Oregon Law Review Spring 80 pp109ndash98

Crandall Robert W and Kenneth G Elzinga2002 ldquoInjunctive Relief in Sherman Act Monop-olization Casesrdquo Brookings Institution workingpaper

Crandall Robert W and Thomas W Ha-zlett 2001 ldquoTelecommunications Policy Re-form in the United States and Canadardquo inTelecommunications Liberalization on Two Sidesof the Atlantic Martin Cave and RobertW Crandall eds Washington DC AEI-

Brookings Joint Center for Regulatory Studiespp 8 ndash38

Eckbo B Espen 1992 ldquoMergers and theValue of Antitrust Deterrencerdquo Journal of Fi-nance July 47 pp 1005ndash 029

Eckbo B Espen and Peggy Wier 1985 ldquoAn-timerger Policy Under the Hart-Scott-RodinoAct A Reexamination of the Market Power Hy-pothesisrdquo Journal of Law and Economics April 28pp 119ndash 49

Federal Trade Commission 1999 A Study ofthe Commissionrsquos Divestiture Process WashingtonDC Bureau of Competition Federal TradeCommission

Ferguson Paul R 1988 Industrial EconomicsIssues and Perspectives London Macmillan

Fisher Alan A and Robert H Lande 1983ldquoEf ciency Considerations in Merger Enforce-mentrdquo California Law Review December 71 pp1580ndash 673

Fisher Franklin M John J McGowan andJoen E Greenwood 1983 Folded Spindled andMutilated Economic Analysis and US v IBMCambridge Mass MIT Press

Gaskins Darius 2001 ldquoDuopoly Pricing ofCoal Transportationrdquo Unpublished paperAugust

Grimm Curtis and Clifford Winston 2000ldquoCompetition in the Deregulated Railroad In-dustry Sources Effects and Policy Issuesrdquo inDeregulation of Network Industries Whatrsquos NextSam Peltzman and Clifford Winston edsWashington DC Brookings Institution pp41ndash71

Harberger Arnold 1954 ldquoMonopoly and Re-source Allocationrdquo American Economic ReviewMay 44 pp 77ndash87

Hoxby Caroline M 2000 ldquoBenevolent Col-luders The Effects of Antitrust Action on Col-lege Financial Aid and Tuitionrdquo NBER WorkingPaper No 7754 June

Ippolito Pauline M and Thomas R Over-street Jr 1996 ldquoResale Price Maintenance AnEconomic Assessment of the Federal TradeCommissionrsquos Case Against the Corning GlassWorksrdquo Journal of Law and Economics April 39pp 285ndash328

Kaysen Carl 1956 United States v United ShoeMachinery Corporation Cambridge Mass Har-vard University Press

Kobayashi Bruce 2002 ldquoAntitrust Agencyand Amnesty An Economic Analysis of theCriminal Enforcement of the Antitrust LawsAgainst Corporationsrdquo George Mason UniversitySchool of Law Law and Economics WorkingPaper Series No 02-04

Lollar Coleman 1992 ldquoBack to the Bad OldDaysrdquo Frequent Flyer December


Masten Scott E and Edward A Snyder 1993ldquoUnited States versus United Shoe MachineryCorporation On the Meritsrdquo Journal of Law andEconomics April 36 pp 33ndash70

McGee John S 1958 ldquoPredatory PriceCutting The Standard Oil (NJ) Caserdquo Jour-nal of Law and Economics October 1 pp 137ndash68

Morrison Steven A 1996 ldquoAirline Mergers ALonger Viewrdquo Journal of Transport Economics andPolicy September 30 pp 237ndash50

Morrison Steven A and Clifford Winston1995 The Evolution of the Airline Industry Wash-ington DC Brookings Institution

Morrison Steven A and Clifford Winston1996 ldquoCauses and Consequences of Airline FareWarsrdquo Brookings Papers on Economic Activity Mi-croeconomics pp 85ndash123

Morrison Steven A and Clifford Winston2000 ldquoThe Remaining Role for GovernmentPolicy in the Deregulated Airline Industryrdquo inDeregulation of Network Industries Whatrsquos NextSam Peltzman and Clifford Winston edsWashington DC Brookings Institution pp1ndash 40

Nevo Aviv 2001 ldquoMeasuring Market Powerin the Ready-to-Eat Cereal Industryrdquo Economet-rica March 69 pp 307ndash42

Newmark Craig M 1988 ldquoDoes HorizontalPrice Fixing Raise Price A Look at the Bakers ofWashington Caserdquo Journal of Law and EconomicsOctober 31 pp 469ndash 84

Parrish Gordon 1973 The Experience with An-titrust Relief in Shoe Machinery Unpublished PhDdissertation Department of Economics Wash-ington State University

Pashigian B Peter 2000 ldquoTeaching Micro-economics in Wonderlandrdquo Working Paper 161George J Stigler Center for the Study of theEconomy and the State University of ChicagoJuly

Paulter Paul A 2001 ldquoEvidence on Mergersand Acquisitionsrdquo Bureau of Economics Fed-eral Trade Commission Washington DCSeptember

Pittman Russell W 1990 ldquoRailroads andCompetition The Santa FeSouthern Paci c

Merger Proposalrdquo Journal of Industrial EconomicsSeptember 39 pp 25ndash 46

Salinger Michael 1990 ldquoThe Concentration-Margins Relationship Reconsideredrdquo BrookingsPapers on Economic Activity Microeconomics pp287ndash335

Schmalensee Richard 1989 ldquoInter-IndustryStudies of Structure and Performancerdquoin Handbook of Industrial Organization Volume2 Richard Schmalensee and Robert Wil-lig eds Amsterdam North-Holland pp 951ndash1009

Schumann Lawrence James D Reitzes andRobert P Rogers 1997 ldquoIn the Matter ofWeyerhaeuser Company The Use of a Hold-Separate Order in a Merger with Horizontal andVertical Effectsrdquo Journal of Regulatory Economics113 pp 271ndash89

Sproul Michael F 1993 ldquoAntitrust and PricesrdquoJournal of Political Economy August 101 pp 741ndash54

Stigler George J 1964 ldquoA Theory of Oligop-olyrdquo Journal of Political Economy February 72 pp44 ndash61

Stigler George J 1966 ldquoThe Economic Ef-fects of the Antitrust Lawsrdquo Journal of Law andEconomics October 9 pp 225ndash58

Tennant Richard B 1950 The American Ciga-rette Industry A Study in Economic Analysis andPublic Policy New Haven Conn Yale UniversityPress

Weingast Barry R and Mark J Moran 1983ldquoBureaucratic Discretion or Congressional Con-trol Regulatory Policymaking by the FederalTrade Commissionrdquo Journal of Political EconomyOctober 91 pp 765ndash 800

White Lawrence J 2002 ldquoTrends in Aggre-gate Concentration in the United Statesrdquo Journalof Economic Perspectives Fall 16 pp 137ndash 60

Wilder Ronald P 1975 ldquoThe Electronic DataProcessing Industry Market Structure andPolicy Issuesrdquo Antitrust Bulletin Spring 2 pp25ndash47

Williamson Harold F Ralph L Andreano Ar-nold R Daum and Gilbert C Klose 1963 TheAmerican Petroleum Industry The Age of Energy1899ndash1959 Evanston Ill Northwestern Univer-sity Press


Does Antitrust Policy ImproveConsumer Welfare Assessing theEvidence


S hould the United States pursue a vigorous antitrust policy Soon after thepassage of the Sherman Antitrust Act of 1890 economists led by John BatesClark (1901) argued that the enforcement of such laws should be informed

by the prevailing economic theory on the merits of competition and the extent towhich rmsrsquo conduct can enhance or weaken competition However economictheory since then has proven remarkably fertile in pointing out how various actionsby rms may be interpreted as either procompetitive or anticompetitive Forexample when prices decline suf ciently so that no rm in an industry is earningeconomic pro ts and some rms exit this outcome may re ect a highly competitivemarket adjusting to a condition of temporary oversupply or it could indicate thata large competitor is employing a strategy of predatory pricing to drive out its rivalsSimilarly when a rm builds a large factory it may be engaged in vigorouscompetition and new entry or it may be creating excess capacity as an implicitthreat to potential competitors that it may raise output and cut price quickly ifcircumstances warrant Although economic theory can help organize analysis of theeconomic variables affected by antitrust policy it often offers little policy guidancebecause almost any action by a rm short of outright price xing can turn out tohave procompetitive or anticompetitive consequences

Given this range of theoretical possibilities the case for a tough and broadantitrust policy must rest on empirical evidence that shows that such policies haveworked in the broad social interest In this paper we argue that the currentempirical record of antitrust enforcement is weak We start with an overview of thebudgets and actions of the federal governmentrsquos antitrust authorities We thensynthesize the available research regarding the economic effects of three majorareas of antitrust policy and enforcement changing the structure or behavior of

y Robert W Crandall and Clifford Winston are both Senior Fellows The Brookings Institu-tion Washington DC

Journal of Economic PerspectivesmdashVolume 17 Number 4mdashFall 2003mdashPages 3ndash26










Investigations




TOTAL 323 310 456




TOTAL $1077 $934 $1469





Monopolization














$0

$5

$10

Crude Oil

1889

1901

1903

1905

1907

1909

1911

1913

1915

1917

1919

1921

1923

1925

Decreeenacted

Kerosene

Gasoline

$15

$20

$25

$30

$35

Rea

l Pri

ce (

1967

$

bbl)








































Collusion









Mergers

























Variable Coefcient



























Conclusion









References












































































Investigations




TOTAL 323 310 456




TOTAL $1077 $934 $1469





Monopolization














$0

$5

$10

Crude Oil

1889

1901

1903

1905

1907

1909

1911

1913

1915

1917

1919

1921

1923

1925

Decreeenacted

Kerosene

Gasoline

$15

$20

$25

$30

$35

Rea

l Pri

ce (

1967

$

bbl)








































Collusion









Mergers

























Variable Coefcient



























Conclusion









References





































































Monopolization














$0

$5

$10

Crude Oil

1889

1901

1903

1905

1907

1909

1911

1913

1915

1917

1919

1921

1923

1925

Decreeenacted

Kerosene

Gasoline

$15

$20

$25

$30

$35

Rea

l Pri

ce (

1967

$

bbl)








































Collusion









Mergers

























Variable Coefcient



























Conclusion









References












































































$0

$5

$10

Crude Oil

1889

1901

1903

1905

1907

1909

1911

1913

1915

1917

1919

1921

1923

1925

Decreeenacted

Kerosene

Gasoline

$15

$20

$25

$30

$35

Rea

l Pri

ce (

1967

$

bbl)








































Collusion









Mergers

























Variable Coefcient



























Conclusion









References







































































































Collusion









Mergers

























Variable Coefcient



























Conclusion









References






































































Mergers

























Variable Coefcient



























Conclusion









References
























































































Variable Coefcient



























Conclusion









References
























































































Conclusion









References






































































References



































































does antitrust policy improve consumer welfare? assessing ...does antitrust policy improve consumer...

Documents