can standard setting organizations address patent holdup ... · ways, such as decentralized...
TRANSCRIPT
1
Canstandardsettingorganizationsaddresspatentholdup?Commentsforthe
FederalTradeCommission
TimothySimcoe,BostonUniversitySchoolofManagement
Standardsettingorganizations(SSOs)areforumswherefirmsnegotiateovershared
designparameterssotheirproductscanworkwelltogether.Thisprocessproduces
manybenefits:usersmayshareinformation,or“mixandmatch”components;the
costofmarketentrydeclines;andthereisadivisionoflabor,enablingspecialization
incomponentproductionandinnovation.Whileinter‐operabilitycanbeachievedin
otherways,suchasdecentralizedtechnologyadoptionortheactionsofadominant
firm,SSOsareperhapsthemostcommonpathtocompatibility.1
AntitrustauthoritieshavegenerallyviewedSSOsasaformofpro‐competitive
horizontalcooperation.Forexample,theUnitedStates’StandardsDevelopment
OrganizationAdvancementActof2004(H.R.1086)providesthatSSOsare
evaluatedunderanantitrustruleofreason,andaresubjecttoactualratherthan
trebledamages.However,thereisconcernamongacademics,policy‐makersand
practitionersthatpatent‐holderscanexploitthecooperativestandardsetting
processtoextractexcessiveroyaltiesfromstandardsimplementers.
Whenapatentisnecessarilyinfringedbycompliantimplementationsofanindustry
standard,itiscalled“essential.”Essentialpatentscanbeveryvaluable,sincea
successfulstandardreducescompetitionfromsubstitutetechnologies.Thus,while
SSOsmayhavegoodreasonstoincludepatentedtechnologyinanindustry
standard,doingsoplacesessentialpatentownersinaverystrongbargaining
positionifthestandardsucceeds.Thisisaspecialcaseofwhateconomistscallthe
“holdup”problem.1 Biddle, White and Woods (2010) identified over 500 compatibility standards used in a prototypical laptop computer and found that 80 percent were developed by consortia and formal standards developing organizations. See Farrell and Simcoe (2011) for one discussion comparing alternative paths to compatibility.
2
Patenthold‐upproblemscouldbesolvedbydisclosingpatentsandnegotiating
pricesbeforechoosingastandard.Butuncertaintyovertheexistence,validityand
scopeofpotentiallyessentialpatentsmakesitdifficulttonegotiatealicensepriorto
implementation.Asanalternative,manySSOsaskpatent‐holderstocommitto
reasonableandnon‐discriminatory(RAND)licensingterms.However,many
observersworrythatitisnotclearhowfirmsorcourtsaremeanttointerpretthis
promise.
ProponentsofthecurrentsystemarguethatSSOsshouldfocusontechnicalissues;
intellectualpropertyownersshouldbecompensatedforthecostsandrisksof
technologydevelopment;andthatprospectivelicenseescanavoidthehold‐up
problembytakingaRANDlicenseorproposinganalternativetechnologywithinthe
SSO.Thesearevalidpoints,butthereisacounter‐argumenttoeach.Technical
decisionsshouldalwaysbeweighedagainstthelikelycostsofimplementation,
whichincludelicensing.AsdescribedintheFTC’sreporton“TheEvolvingIP
Marketplace”(FTC2011),overcompensatingpatentownershasrealsocialcosts.
Anditisunclearwhetherthepatentsystemorthedisclosurerulesadoptedbymany
SSOsprovidesufficientnoticeforprospectivestandardsimplementerstoengagein
exantenegotiation.
Thisshortessayarguesthatpatenthold‐upinstandardsettingisarealproblem,
andconsidersseveralstepsthatSSOs,courtsandpolicy‐makersmighttaketo
addresstheissue.
Standardsettingandpatentholdup
Standard setting organizations serve as a forum where industry participants
perform collaborative research anddiscuss themerits of alternative technologies.
Thegoalistoidentifythebestavailablesolutiontoagiventechnicalproblemwhen
therearegains fromcoordinatingona commondesign.Ultimately, SSOschoosea
particulartechnologyandissueaformalendorsement.Thiscertificationismeantto
3
signaltheendofdeliberationsandpromoteindustry‐wideinvestmentsinthenew
technology.2
Difficultiesarisewheninvestmentsinanewstandardaresunk(i.e.irreversibleor
technology‐specific),leadingtowhateconomistscalla“hold‐up”problem.3Sunk
investmentsmakeatechnologycheapertodeploy(onaforward‐lookingbasis)than
alternativesolutionsthatofferedapotentialsubstitutebeforestandardization.
Thus,whenthetechnologyinastandardispatented,theintellectual‐property
holdercanchargeroyaltiesuptothedifferenceinimplementationcostsbefore
vendorswillswitch–evenwhentheexantecoststructureoftwotechnologieswas
identical.4Inotherwords,apatentthatisworthlesspriortostandardization(given
theexistenceofaperfectsubstitute)maybecomequitevaluableifSSOendorsement
leadstosubstantialtechnology‐specificinvestments.5
Inprinciple,producersthatrecognizethepossibilityofhold‐upcouldprotect
themselvesbyconductingacomprehensivepatentsearchandsecuringthe
necessarylicensesbeforeimplementinganewstandard.Butfirmsinthe
informationandcommunicationstechnology(ICT)sectorrarelydothisgiventhe
largesearchandtransactionscostsassociatedwithup‐frontlicensing.Justfinding
therelevantpatentscanbehard,sinceamodernlaptoporsmart‐phonewill
implementhundredsofstandardsandinfringethousandsofpatents.Assuming
firmscanlocateallofthesepatents,theymightreasonablyhesitatetosignlicenses
whenthereisuncertaintyovereachpatent’sscopeandvalidity(FTC2011).Finally,
evenifallrelevantpatentscouldbelocatedandtheirscopeandvalidityaccurately
2 David and Greenstein (1990) and Shapiro and Varian (1998) describe the economic and business strategy issues associated with the formal standards process. Rysman and Simcoe (2008) provide some empirical evidence that SSO endorsements have an impact on the value of the underlying technology. 3 Williamson (1985) introduced the hold-up problem, which he called the “fundamental transformation.” Farrell et al (2004) provide a detailed discussion of hold-up problems in this context. 4 While policy debates often focus on “technically essential” patents, it is worth noting that the hold-up problem can also apply to so-called “commercially essential” patents, which are complementary to a standard, but not an inherent part of the specification. 5 Of course, this argument depends critically on the assumption that the patent is valid and enforceable. However, the threat of injunction may provide even “weak” patent owners with a very large stick in the bargaining process.
4
assessed,transactioncostconsiderationsleadmanyICTfirmstopreferabroad
company‐widecrosslicenseoveranarrowlicensecoveringjustthosepatents
neededtoimplementaparticularstandard.
Giventhelargesearchandtransactioncostsinmarketsforintellectualproperty,ICT
firmsoftendesign,buildandmarkettheirproductsbeforetryingtoclearallofthe
relevantpropertyrights,eventhoughtthisexposesthemtohold‐upbypatent
owners.Afewlargefirmsnegotiateup‐frontcross‐licenses,buteventhese
companiesmayfindthemselveslockedintoacounterparty’sproprietarytechnology
whenthoseagreementsarerenegotiated.6
Whileinefficientmarketsforintellectualpropertyareageneralproblem,hold‐up
problemsareespeciallyseverewhenfirmsholdpatentsinindustrystandards.Once
astandardhasbeenselected,SSOsworkhardtopromotecoordinatedinvestments
byabroadcommunityofimplementers,andtheseinvestmentsareoftenverycostly
toreverse.Thus,awidelydiffusedstandardprovidesalargepoolofpotential
licensors,reducescompetitionfromsubstitutetechnologiesandmakes
infringementeasytodetect.Simcoe,GrahamandFeldman(2009)providesome
evidenceonthispoint.TheirstudyfoundthatpatentsdeclaredtoSSOswere4to7
timesmorelikelytobelitigatedthanatypicalpatentwiththesameageand
technologyclass.Thesehighlitigationratesdonotimplywidespreadimplyhold‐up,
butareastrongindicationthatthemarketforstandards‐relatedintellectual
propertyisnotfunctioningwell.
Lookingforward,thereareseveralreasonstoworrythatconflictsoverstandards‐
relatedIPwillincreaseandintensify.First,thesecondarymarketforpatents
appearstobegrowing,withincreaseddemandcomingfrompatent‐assertion
entities(PAEs)whosemainbusinessistoacquireandassertpatents.Forexample,6 See Grindley and Teece (1997) for a discussion of broad cross-licensing practices. One example of a cross-license containing essential patents that fell apart at renegotiation is the deal between Motorola and Research in Motion (see Motorola, Inc vs. Research in Motion, Ltd et al., U.S District Court Northern District of Texas, 3:08-CV-0317-G).
5
afternegotiationsbetweenNokiaandRobertBoschGmbHfailed,Boschsoldseveral
standards‐relatedpatentstoaPAEcalledIPCom,whosuccessfullyassertedthose
patentsinGermancourts.Thereisalsoabroadshifttowards“openinnovation”
businessmodels,wherefirmscommercializetechnologythroughout‐licensingas
opposedtomanufacturing.FirmslikeQualcomm,whodevelopnewtechnologythat
becomescentraltoanindustrystandard,canearnsubstantialroyaltiesfrom
licensingtheirstandards‐relatedIP.ThestudybySimcoe,GrahamandFeldman
(2009)showsthatwhensmallfirms(whoappeartobeamixofPAEsand
specializedtechnologydevelopers)declareessentialpatentstoanSSO,the
probabilityofsubsequentlitigationincreasessignificantlycomparedtoessential
patentsdeclaredbylargerfirms,whoaremorelikelytobemanufacturers.
Anothercauseforconcernistheincreasingimportanceofinter‐operability,and
particularlythepushforstandardstosupportthedeploymentofelectronichealth
recordsanda“smart”electricitygrid.Newstandardsareclearlyneededto
coordinatetheeffortsofdiversevendorsinrespondingtothesepotentiallylarge
marketopportunities.However,giventhehighlyinelasticconsumerdemandfor
healthandenergyservices,itisnotclearthatintermediarieslikehospitalsorpublic
utilitieshavetheabilityorincentivetobeanactiveSSOparticipantorengagein
protractedfightswithsophisticatedpatentlitigatorsiftheycansimplypassroyalty
costsontoconsumers.
Privatesolutions
MostSSOstakestepstomitigatethepatenthold‐upproblem,typicallybyadopting
intellectualpropertypoliciesthatimposecertainobligationsonmembers’behavior
duringandafterthestandardsettingprocess.Lemley(2002)surveyedthese
policiesatanumberofSSOs.Here,Idiscusesafewoftheirmainfeatures.
Disclosurerules
ManySSOshaverulesthatrequiremembersparticipatingintechnicaldeliberations
todiscloserelevantpatentsorpendingpatentapplications.Forinstance,thepatent
6
policyguidelinesendorsedbytheAmericanNationalStandardsInstitute(ANSI
2011)encourageSSOstoadopt“procedureswherebyoneormorerequestsare
madetoparticipantsforthedisclosureofpatentsthatmayberequiredforuseof
standardsinprocess.”ManySSOsalsomaintainapublicrepositoryofpatent
disclosuresorlettersofassurance,whichcanbesearchedbyprospective
implementerswhowishtoseekalicensebeforecommittingtothestandard.
Inprinciple,disclosureallowsSSOparticipantstoevaluatetrade‐offsbetween
technicalqualityandtheexpectedcostofimplementation.Inpractice,thereare
severalreasonswhythedisclosureprocessmaynotworkwell.Oneproblemisthat
disclosurerulesrarelyrequirefirmstosearchforordisclosespecificpiecesofIP.As
aconsequence,manyfirmsissue“generic”or“blanket”disclosuresindicatingthat
theyholdessentialpatents,butwithoutprovidinganypublicationnumbers.Clearly,
thesedisclosuresprovidelittleguidancetostandardsdeveloperswhoaretryingto
understandwhatpartsofaspecificationareactuallycoveredbypatents,orwhether
thosepatentsarevalidandenforceable.Simcoe(2005)suggeststhatonethirdofall
disclosuresmadetoagroupofnineSSOsprovidenoreferencetospecificIP,with
muchhigherratesofblanketdisclosureatcertainSSOs.
Thelogicofblanketdisclosureiscloselytiedtotheproblemofpatentnotice.Given
longpendencylags,andthewidespreaduseofcontinuationsanddivisional
applicationstoamendclaimsovertime,itisoftenunclearwhetherapending
applicationwillbeessentialatthetimeastandardisendorsed.Thecostsoflocating
essentialpatentscanalsobelargeforfirmswithsizablepatentportfoliosthat
participateinmanySSOs.Ofcourse,theswitchtoblanketdisclosuredoesnot
actuallyreducethesesearchcosts;itmerelyshiftsthemontoprospective
implementersandotherstandardsdevelopers,whoarearguablylessinformed.
Simplymandatingdisclosurespecificitywillnotnecessarilysolvetheproblemof
poornoticeunderblanketdisclosurepolicies,sincelargepatentholderscanadopta
highlyinclusiveapproach,e.g.claimingthatnearlyalloftheirpatentsarepotentially
7
essential.Forexample,ETSIrequiresexplicitdisclosureandsometimesgets
thousandsofdeclaredessentialpatentsforasingleproject.Ifonlyafewofthese
patentsaretrulyessential,implementersandstandardsdevelopersfaceessentially
thesamesearchanddiscoveryproblemasunderablanketpolicy.
AsecondproblemwithSSOs’disclosurerulesisthattheydonotguaranteetimely
informationprovision.Standardscommitteesmaytakeseveralyearstoagreeona
specification.Duringthattime,asconsensusbeginstoemergeonkeydesign
parameters,firmswilloftenbegintocommitdesignandmarketingresourcestothe
nascentspecification.WhileSSOsencourageparticipantstodisclosepatentsasearly
aspossible,thereisnothingtopreventanopportunisticpatent‐holderfromwaiting
untilthelastmoment.Atthatpoint,ifthereislittlechancethatthestandardwillbe
redesigned,thepatent‐holderhasmanagedtoputthemselvesinapositiontohold‐
upimplementerswithoutviolatingtheletteroftheSSOsdisclosurepolicy.
Moreover,toanoutsideobserver,theseopportunisticdelayswilloftenbe
indistinguishablefrom“late”disclosurescausedbylast‐minutechangestothe
standardthatmakeafirm’spreviouslyirrelevantIPessential.Effortstodefine
“timely”disclosurewouldlikelyencounterstrongresistance,giventhepractical
difficultiesofdistinguishingbenignfromopportunisticdelays.
WhiletimingandspecificityposedifficultproblemsforSSOs,disclosurepolicies
remainausefultoolfortheirmembers.Inparticular,thedisclosureprocess
providesaplacewherefirmscanmakeunilaterallicensingcommitments(perhaps
withaneyetowardsgainingacceptanceoftheirtechnology).Thedisclosureprocess
alsoprovidesaforumwherepatent‐holderscanplaceprospectiveimplementerson
noticethattheyexpecttobepaid.Finally,whenSSOsprovideapublicrepositoryof
disclosuresorlettersofassurance,itsimplifiestheprocessofmatchingfuture
implementerstokeyintellectualpropertyowners.
Perhapsthebestwaytoaddresstheproblemsofdisclosuretimingandspecificity
wouldbethroughreformstotheunderlyingpatentsystem.Highqualitypatents
8
thatissuedmorequickly,andwhoseevolvingclaimswereeasiertopredict,would
makeitsimplerforSSOparticipantstoevaluatetrade‐offsamongsolutionswith
differentcost‐qualityprofilesonaforward‐lookingbasis.
LicensingCommitments
Inadditiontodisclosurerules,manySSOsrequireparticipantstomake
commitmentsregardingthetermsandconditionsofanylicensetheywillofferfor
patentsneededtoimplementastandard.AfewSSOsaskforveryspecificlicensing
commitments.Forinstance,theWorldWideWebConsortium(W3C)willonly
endorsetechnologiesthatcanbeimplementedonaroyalty‐freebasis,andtheHDMI
Consortiumasksimplementerstosignan“AdoptersAgreement”andpayanannual
feeof$10,000plusaroyaltyof$0.15perunitsold.7However,themostcommon
policyistorequireacommitmenttolicenseon“reasonableandnon‐discriminatory”
(RAND)or“fairreasonableandnon‐discriminatory”(FRAND)terms.
Whetherbecauseofantitrustfears,orconcernsthattheywillupsetcertain
members,SSOstypicallyshyawayfromprovidinganexplicitdefinitionofRAND,
leavingthemattertoindividualfirms,andultimatelycourts.Asaconsequence,
firmsoftentreatthiscommitmentasmerelyapromisetoenterlicensing
negotiations.Thereisnothingtopreventpatent‐holdersfromgoingbeyondaRAND
commitment,forinstancebypledgingtoofferaroyalty‐freelicense.Butasidefrom
unilateralcommitments,itisnotclearthataRANDpromiseplacesanyrestrictions
onprospectivepricesorlicensingterms,asidefromabanonexclusivity.8
Howthen,shouldweinterprettheRANDpromise?Economistshavesuggestedthat
a reasonable royalty will reflect the ex ante benefits of adopting a particular
technology as the industry standard, but not the additional switching and7 The W3C patent policy is published at http://www.w3.org/Consortium/Patent-Policy-20040205/ and the HDMI Adopters Agreement is published at http://www.hdmi.org/manufacturer/terms.aspx (both accessed July 8, 2011). 8 While there is considerable debate over the meaning of RAND, it is clear that firms cannot void their RAND pledge by selling the patent. See In the Matter of Negotiated Data Solutions LLC, FTC File No. 0510094, for a discussion of efforts to circumvent SSO licensing commitments through patent sales.
9
coordinationcostscreatedbystandard‐settingprocessandthetechnology‐specific
investmentsitencourages(e.g.SwansonandBaumol2005;Layne‐Farraretal2007;
Farrell et al 2007). But these proposals rarely specify how onemight calculate a
reasonablepriceafterinvestmentsaresunk,orinsettingswithsignificantnetwork
effects and complementarities among component technologies. As for non‐
discrimination, firms clearly do not interpret this to mean uniform pricing for a
stand‐alone license to essential patents. Perhaps non‐discrimination means that
essential patent holders cannot offer an exclusive license? But if licensors are
allowed to charge different rates to each licensee, and there is no cap on a
reasonableprice, itseemsthat licensorscouldofferdefactoexclusivity,orat least
substantialcompetitiveadvantage,withoutviolatingaFRANDcommitment.
Oneapproachtothe“reasonable”prongoftheRANDcommitmentistotreatitasa
promise to forgo lost profits as the approach to calculating damages in patent
infringementlitigation.Initsplace,courtswouldpresumablysetreasonableroyalty
damagesbasedonahypotheticalnegotiationbetweenawillinglicensorandwilling
licensee. The U.S Federal Trade Commission (FTC 2011) describes several
difficulties with such an analysis, and suggests that courts may systematically
overcompensatepatentholdersinapplyingthehypotheticalnegotiationframework.
Here,Isetasidethecontroversialquestionofexcessivedamagesandhighlightthe
economic importance of assuming that any hypothetical negotiation takes place
beforetheselectionanddiffusionofthestandard.
First, as emphasized by the economists cited above, an ex ante negotiation will
compensate licensors for thebenefits createdby selecting their technology as the
standard, but not the switching and coordination costs caused by implementers’
sunkinvestments.Second,inahypotheticalexantenegotiation,thereshouldbeno
presumption that the patents were known to be valid and infringed. Third, and
perhapsmostimportantly,theexistenceofalternativestothepatentedtechnology
placeanupperboundona licensees’willingnesstopay.Thelastpoint issubtle in
thestandardscontext.Itmaybedifficulttoviewtechnologiesthatlooklousyexpost
10
as post as attractive ex ante substitutes, or to distinguish between a technology’s
innate benefits and the cumulative effect of sustained investment by a broad
communityofimplementers.Moreover,ahypotheticallicensee’sexantewillingness
topaywilldependnotonlyonthemeritsofthebestavailablesubstitute(asinthe
case of a bilateral negotiation), but on the anticipated actions of other firmswith
respect to the nascent standard. Thus, courts may need to consider the price at
which the substitute technology becomes a suitable choice for the pivotal SSO
participant (i.e. the SSO member who casts the deciding vote) and not just the
differenceinexantewillingnesstopayofthedefendantinaparticularcase.Despite
these complexities, it is important to recognize thatwhen there are closeex ante
substitutesforapatentedtechnology,areasonableroyaltymaybeverysmall.
Assumingthe“reasonable”prongofRANDdoesrefertothemethodofcomputing
infringementdamages,SSOscouldtakeseveralstepstoimprovethelikelihoodthat
courtsunderstandthesepoints,andreacheconomicallysoundconclusionswhen
calculatingreasonableroyaltydamages.Onesuchstepwouldbetoclarifythe
meaningofRANDbyexplicitlydefining“reasonable”asthepricereachedthroughex
antenegotiationbetweenawillinglicensorandwillinglicensee.Anotherusefulstep
wouldbetoclearlydocumentcaseswheretherewasaconsciouschoicebetween
competingsolutions,andprovidingsomeindicationofhowtherelativemeritsofthe
differentapproacheswereviewedatthetime.
SomeobserverstaketheviewthatRANDcommitmentsnotonlycommitpatent
holderstoaparticulardamagesrule,butalsowaivetheirrighttoseekinjunctive
reliefagainststandardsimplements.Forinstance,Miller(2007)writesthat,“by
adoptingaRANDpolicy…SSOparticipants…contractoutofaninjunction‐backed
propertyruleintoareasonable‐royaltyliabilityrule.”Whilethisisaninteresting
proposal,patentholdersmightreasonablyobjectthatwithoutthethreatof
injunctionitisverydifficulttobringanobstinateinfringertothenegotiatingtable.
ThematterisripeforclarificationfromSSOs.Butregardlessoftherelationship
betweenRANDandinjunctiverelief,properapplicationofthehypothetical
11
negotiationframeworkshouldnotrelyon“comparable”licensesthatwere
negotiatedintheshadowofaninjunction,sincethatthreatwouldnotbeavailableto
thepatentholderinanexanteworldwheretheessentiality,validityand
infringementofthepatentareunknown.
Whilemuchhasbeenwrittenonthedeterminationofreasonableroyalties,thenon‐
discriminationprongofaRANDcommitmenthasreceivedrelativelylittleattention.
OneinterpretationofNDmightbethatlicenseswillbemadeavailableatauniform
price.Butoutsideofroyalty‐freelicensing,itishardtoseehowthiswouldbe
practiced,sinceessentialpatentsareoftencoveredbybroadcross‐licenseswhose
termsvaryfromonedealtothenext.Anotherinterpretationofnon‐discrimination
isthatpatenteesmaynotadopt“divideandconquer”adoption‐forcingstrategies,
suchasanescalatingroyaltyratethatguaranteesearlylicenseesadecisivecost
advantageoverlaterones.AfinalinterpretationoftheNDrequirementisthatit
prohibitsexclusivelicensing,thoughthiswouldberedundantifthereasonable
prongofRANDimpliesawillinglicensor.PerhapsNDsimplyassuresaneasily
pronouncedacronym.
Sofar,courtshavemanagedtoavoidprovidingaclearinterpretationoftheRAND
standard.TheappealscourtinBroadcomv.Qualcommsuggestedthatthefifteen
GeorgiaPacificfactorsprovideausefulstartingpoint.9However,thoseguidelines
havelittletosayabouthowonecoulddetermineareasonablepriceinsettingswith
hundredsorthousandsofcomplementarypatentsperproduct,andwherefirms
makesubstantialirreversibleinvestmentstobringproductstomarketbefore
enteringintolicensingnegotiations.Moreover,astheFTC’sreportontheevolving
IPmarketplace(FTC2011)highlights,theGeorgiaPacificfactorsareonlyalistof
evidentiarycategories,andnotacoherentanalyticalframeworkthatcanbeusedto
calculateareasonableroyalty.Inmyview,thecurrentambiguitysurroundingRAND
commitmentscontributestothehighlitigationrateofdeclaredessentialpatents,
9 See Broadcom Corp. v. Qualcomm, Inc., 501 F.3d 297; 2007.
12
makingitreasonableforcourtsandpolicy‐makerstopushSSOsformoreclarity
regardingthiskeydimensionoftheirintellectualpropertypolicies.
ExAnteNegotiations
OnewidelydiscussedalternativetoRANDcommitmentsisforpatent‐ownersto
actuallynegotiatespecificlicensingterms–includingprices–beforeirreversible
investmentsandcoordinationonaparticularsolutionmakesubstitutetechnologies
lessattractive.ThisisessentiallythemechanismproposedbySwansonandBaumol
(2005),Layne‐Farraretal(2007)andFarrelletal(2007)fordetermininga
reasonableprice.
Inprinciple (andperhaps inpractice), firmsmay conductbilateralexante license
negotiationsoutsideofanSSO (seeHerman,2010).Butwhen theoutcomeofany
bilateral negotiation is confidential, it provides no information to other
implementersaboutthetermsthataparticularpatent‐holderviewsasreasonable,
and therefore induces less ex ante price competition between substitute
technologies. For this reason, the recent policy discussion has focused on ex ante
policieswherepatentholderspubliclydiscloseamaximumroyaltyrateandsetof
mostrestrictiveterms.
Themainappealofexante disclosurepolicies is that theymimic thehypothetical
negotiation between a willing licensor and willing licensee that might take place
beforeastandardisadopted.Inparticular,inexantenegotiationsarationallicensor
would be unwilling to pay an unreasonable price, i.e. a price that exceeds the
difference in surplus between adopting a preferred technology and the next best
alternative (which could be an existing or as‐yet undeveloped substitute or a
decisiontonarrowthescopeofthestandard).
However, ex ante disclosure policies also have someweaknesses. The first one is
widely recognized: price setting and group negotiations among product market
competitorswithinanSSOraisesthespecterofantitrustlaw,particularlychargesof
13
collusion. A second concern draws less attention: considering pricing requires a
changeofcultureandorganizationofSSOsthatmaydistractfromtheirprimarygoal
ofdeveloping technical standards. Finally, someargue that shifting toexantewill
reducetheincentivetoparticipate,orthatfirmswillbeabletogamethepolicy.
Withregardstothefirstconcern,aretheSSOs’antitrustfearsjustified?Thereisin
fact gathering support to have SSOs negotiate licensing fees simultaneously with
determining a standard (e.g. Majoras, 2005). The US Department of Justice has
issuedBusinessReviewLettersthatendorseapolicyofexantepricedisclosureat
VITA(anSSOthatpromotestheVMEbuscomputerarchitecture)andtheIEEE.10The
VITApolicyrequiresIPholderstocommittoa“pricecap”(i.e.amaximumroyalty
rateandmostrestrictivesetoflicensingterms),whichcanbeamendeddownwards,
whiletheIEEEpolicyallowsfirmstodisclosetheirmostrestrictivelicensingterms
onavoluntarybasis.Bothpoliciescontinuetoforbiddirectnegotiationsoverprices
withintheSSO.
So why do SSOs remain hesitant to allow ex ante negotiations? While Business
Review Letters insulate SSOs against public lawsuits, there would no doubt be
privateantitrustsuitsinthiscontext,anditremainstobeseenwhethercourtswill
view ex ante rate setting as legal. For example, Golden Bridge sued the 3GPP for
facilitatingcollusionafterthefirmdisclosedapatentandbegannegotiatinglicense
agreements, only to find that 3GPP members rewrote the standard so as not to
infringeGoldenBridge’spatents.Similarclaimsofmonopsonyandgroupboycotts
bySSOmemberswere litigated inSonyElectronicsvs.SoundviewTechnologiesand
Addamaxvs.OpenSourceSoftwareFoundation.11Farrellatal(2007)suggestthatthe
economicmeritsoftheseclaimsareweak:thedamagetoinnovationincentivesfrom
collectivenegotiationsarenogreaterthanwhentheinnovatormustnegotiatewith
10 These letters are available at http://www.usdoj.gov/atr/public/busreview/219380.htm and http://www.justice.gov/atr/public/busreview/222978.htm (accessed on October 19, 2010). 11 Golden Bridge, Inc. vs, Nokia, Inc., 416 F. Supp. 2d 525, 528 (E.D. Tex. 2006). Sony Electronics Inc. vs. Soundview Technologies, Inc., 157 F. Supp. 2d 180, 185 (D. Conn 2001). Addamax Corp. vs. Open Source Software Found., 888 F. Supp. 274, 278 (1995).
14
asinglelargeuser.Nevertheless,SSOsareunderstandablyreluctanttoincurtherisk
oflitigation.
Withregardstothesecondconcern,standardsaretypicallydevelopedbyengineers
wholackthetrainingandauthoritytoconsiderprices.Businessexecutivesare
certainlyinvolvedinSSOsfromtimetotime.However,combiningtechnical
deliberationswithroyaltynegotiationswouldpresumablyleadtosignificant
changesinthewaysthatSSOsworkandinwhoattends.Suchchangeswouldbe
costlyanddifficult,andnotparticularlywelcomebymanySSOparticipants.Whileit
isdifficultsayhowmuchweightshouldplacedontheseconcerns,theyareclearly
important.
Athirdconcernwithexantepoliciesisthattheymightforcepatent‐holderstoprice
aproductthatnolicenseeactuallywants.Inparticular,mostICTlicensescovera
firm’sentirepatentportfolio,therebyensuringalicensee’sfreedomtooperatewith
respecttoaparticularcounterparty.Firmsrarelyofferanarrowlicensethatcovers
onlytheessentialpatentsforasinglestandard.12However,intheabsenceofa
stand‐alonelicensingoption,itishardtoknowwhethercurrentpracticereflectsthe
preferenceoflicenseesorlicensors.
Afinalconcernwithexantepoliciesisthattheywillbeineffective,orevenharmful.
Inparticular,firmsmightdeclareveryhighroyaltyratestopreserveflexibilityin
subsequentnegotiations(essentiallypreservingthestatusquo)orstop
participatinginanSSOratherthancommittopricesinadvance.Thesearguments
arenotespeciallystrong.Whilefirmscouldcertainlydeclarehighmaximum
royaltiesunderanexantepolicy,thiswouldstillremovesomeelementof
uncertainty.Whencombinedwithincentivestomaketimelydisclosures,itishardto
seehowmandatoryexantedisclosurewouldnotleadtosomepricecompetition
betweensubstitutetechnologies.
12 [[Note on technically vesus commercially essential patents.]]
15
Concerningparticipation,themainreasonforafirmtoquitanSSOthatadoptsanex
antepolicyisthatitexpectstoearnlessroyaltyincome.Butifanexanteprocess
mimicsthehypotheticalnegotiationthatwouldleadtoareasonableprice,this
impliesthatthedepartingfirmexpectedtocollectunreasonableroyaltiesunderthe
RANDpolicy.Inthatcase,whatharmcomesfromhavingthemoutsidetheSSO?
Arguably,theexantepolicyscreensoutfirmswhoviewRANDasanemptypromise,
andreducesthechancethattheywillendupholdingtrulyessentialpatents.
Ultimately,theimpactofadoptinganexantedisclosurepolicyonSSOparticipation
isanempiricalquestion.Whilethereareonlyafewcasestostudy,theW3Cand
VITAaretwoexamplesofSSOsthathavestrengthenedtheirlicensingrequirements
withoutsufferingacatastrophicdeclineinmembership.
OtherPolicies
WhilerecentdebatesonSSOintellectualpropertyruleshavefocusedonRAND
commitmentsandexantedisclosurerules,therearemanyotherpossibilities.One
alternativetoholdingup‐frontnegotiationswithinSSOsistoencouragemoreprice
competitionbetweenthem.Forexample,anumberofstandardsconsortiausea
“Promoter‐Adopter”modelwhereasmallgroupoffoundingmembers(Promoters)
contributethekeytechnologiesandagreetoanexplicitlicensingregimebefore
signingupimplementers(Adopters).Standardsdevelopedusingthisapproach
includetheUSB2.0andHDMIspecifications.Incomparingexantedisclosure
policiestothePromoter‐Adoptermodel,onecandiscernatrade‐offbetweenthe
proceduralcostsofcombiningtechnicalandlicensingnegotiationsinsideasingle
body,andthepossibilityofincreasedfragmentationandcoordinationfailurewhen
substitutetechnologiesarepromotedbycompetingSSOs.
TherearealsoseveralalternativepoliciesthatcouldbecarriedourwithinanSSO.
Forinstance,MarkLemley(2007)proposedseveralinterestingextensionstoex
antenegotiation.Oneoptionistocoupleexantedisclosurerequirementswithlow
“penaltydefault”royaltyrates.Underthispolicy,theSSOwouldsetaverylow
16
defaultroyaltyrate(perhapsfree)asthemaximumallowableroyaltyforfirmsthat
failedtoprovideexplicitpricingcommitmentsormaketimelydisclosures.
OneweaknessofaRANDorexantepolicywithpenaltydefaultsisthatitdoesnot
addresstheroyaltystackingproblem:evenwithalowdefaultrate,totalroyalties
mayaddupwhentherearehundredsofdeclaredessentialpatentsandhundredsof
standardsperproduct.Toaddressthisissue,LemleyproposedthatSSOscould
adoptastep‐downroyaltyprocedure.Underthispolicy,anSSOcouldcapthetotal
royaltiesassociatedwithastandardandawardasharetoeachessentialpatentor
firm.Thesesharescoulddeclineaspatentsareaddedtothepool,andmightbe
higherforthosewhodiscloseearlier.Whilethisstep‐downproceduremightnot
awardthemostroyaltiestothehighestvaluepatents,itwouldcertainlyprovidean
incentiveforearlydisclosure,whichwouldallowcommitteestomakebetter‐
informeddecisionsaboutanytrade‐offbetweenqualityandimplementationcost.
Finally,RysmanandSimcoe(2011)suggestanalternativepatentpolicythat
dispenseswithbothRANDcommitmentsandexantenegotiations.Theirproposalof
Non‐AssertionAfterSpecifiedTime(NAAST)wouldmimictheincentivesofthe
underlyingpatentsystembyallowingpatent‐holdersaperiodofunrestricted
licensing(i.e.theywouldnotbeboundbyRANDoranyothercommitment),but
forcingthemtogiveupallrightstoassertessentialpatentsaftersomepre‐
determinedperiodoftime.Thispolicywouldallowpatent‐holderstorecouptheir
investmentsininnovationduringtheassertionphase,butprovidefreedomofaccess
toimplementersandinnovatorswhowishtobuildonthestandardafterthenon‐
assertphasebegins.AmajorvirtueoftheNAASTpolicyisthatitwouldbeeasyto
adjudicate,sinceitdoesnotrelyoncomplexformulationsfordetermininga
reasonableprice.Itmightalsobringtheeffectivepatenttermforessentialpatents
intolinewiththelengthofICTproductlifecycles.Ontheotherhand,itishardto
knowtheappropriatelengthoftheNAASTassertionperiod,orwhenitshould
commence.
17
Penaltydefaults,step‐downroyaltiesandNAASTarenotperfectsolutionstothe
hold‐upproblem.RathertheyillustratethewiderangeoffeasibleoptionsforSSO
intellectualpropertypolicies.Giventhemanypossiblepolicies,onemightaskwhy
wecurrentlyobserverelativelylittlevariation,andwhetherweshouldexpect
competitionbetweenSSOstoleadtowardsthebestpossiblesetofIPRpolicies?Ido
notthinkweknowtheanswertothesequestions.
Tobesure,manySSOshavetweakedandtightenedtheirIPpoliciesinthewakeof
Rambusandothermajorantitrustcases.Therehavebeenexperimentswithexante
negotiationatVITAandtheIEEE;theW3Cswitchedtoaroyalty‐freepolicy,and
OASISallowsindividualtechnicalcommitteestodecideonthemostappropriateIPR
policy.However,manySSOscontinuetorelyonRANDpoliciesthatprovidelittle
clarityaboutdisclosuretimingandspecificity,orhowcourtscoulddeterminea
reasonableprice.ForsomeSSOsthatrunintofewpatentdisputes,stickingwith
RANDmaybeapragmaticapproachthatreflectsthecostsofcreatinganewIP
policyaswellasantitrustconcerns.Ontheotherhand,evenatSSOslikeETSI,which
havebeenembroiledinanumberofmajorpatentdisputes,effortstomoveaway
fromtheRANDregime(e.g.bydefining“timely”disclosure)haveencountered
strongresistance.Ultimately,itisnotobviouswhethercompetitionamongSSOs
takestheformofdevelopingmoreefficientrulesandprocedures,orofproviding
alternativevenuesthatfavortheinterestsofdifferentconstituencies.Theseissues
deservemoreresearch.
Publicpolicy
Howcanpublicpolicysupporttheemergenceofprivatesolutionstotheproblemof
patenthold‐upinindustrystandards?Oneroleforpublicagenciesistoprovidea
crediblethreatofdisciplineforfirmsthatviolateSSOpolicies.Standardsetting
organizationstypicallyhaveverylimitedenforcementpowers:whiletheycan
threatentowithdrawsupportforastandard,thishappensinfrequently,andwill
havelittleeffectwhenthespecificationalreadyhasalargeinstalledbase.However,
antitrustagencieshavebroughtactionsagainstDell,Unocal,Rambusandothersfor
18
seekingtolicensepatentsthatwerenotdisclosedduringthestandard‐setting
process.13Simcoe(2005)showsthattherewasasubstantialuptickinthenumberof
patentdisclosuresfollowingtheDellactionin1996.Thus,evenifthetermsof
individualpatentdisclosuresremainvague,theseantitrustactionshavearguably
increasedtheinformationavailabletostandardsdevelopersandimplementersby
makingtheenforcementofSSOpolicescredible.
AnotherproductivesteptakenbyantitrustagenciesintheUSandEuropehasbeen
toofferBusinessReviewLettersandpublicstatementsencouragingSSOstoallow
moreexantepricenegotiation.Inparticular,allowingtheVITAandIEEEexante
disclosurepoliciestomoveforwardclearlysignalsthatregulatorsunderstandthe
pro‐competitivepotentialofpoliciesthatleadtopublicpricecommitments.SSOs
maystillfearthatexantepolicieswillleadtoprivateantitrustactions.Thisisa
reasonableconcern,particularlyinthewakeoftheGoldenBridgev.Nokiarulingthat
groupboycottsinSSObargainingcanbefoundperseillegal.However,statements
bytheDOJandFTCsuggestthatconductinexantepricenegotiationswilltypically
bejudgedunderaruleofreasonstandard,reflectingboththecoordinationbenefits
associatedwithjointtechnicaldecisionmakinginsideSSOsandthepotentialforex
antenegotiationstothwartpatenthold‐up.
Amorecontroversialrecommendationisthatpolicy‐makersnudgeSSOstowards
providingmoreexplicitguidanceondisclosurerulesandlicensingcommitmentsas
partoftheirIPRpolicies.14Onewayforpolicy‐makerstoencouragemoreclarity
mightbetoindicateapreferenceforexplicitIPRpoliciesandproceduresin
governmentpurchasingguidelines,suchasOMBCircularA‐119.Thecurrent
guidelinesindicatethevoluntaryconsensusstandardsmust“includeprovisions
requiring[that]intellectualproperty[bemade]availableonanon‐discriminatory,
13See Dell Computer Corp., 121 F.T.C. 616 (1996); Union Oil Co. of Cal., FTC Docket No. 9305, and Rambus, Inc., FTC Docket No. 9302. 14 The Rambus court made several disparaging remarks about the lack of clarity in VITA’s IPR policies. However, my impression is that subsequent efforts at many SSOs have focused on clarifying procedural requirements as opposed to the substance of these rules.
19
royalty‐freeorreasonableroyaltybasistoallinterestedparties”andencourage
agenciestouseofvoluntarystandardswhereverpractical.Buttheseguidelinessay
verylittleaboutthechoiceamongcompetingprivatestandards.Whengovernment
agenciesactasconsumers,itwouldbequitereasonableforthemtofavorstandards
thatinvolvemorecertaintyandlessriskofexpostpatenthold‐up.Andsincethese
purchasingguidelinesleaveagencieswithconsiderablediscretion,thereislittlerisk
thatsuchachangewouldleadtoexcessiveuniformityofSSOpoliciesand
proceduresshouldvaguenessbetheoptimalIPRpolicyforcertaintechnologiesor
markets.
Finally,thoughitisoutsidethescopeofthisessay,hold‐upproblemscouldbe
mitigatedthroughpatentsystemreformsthatreducethemyriadsearchand
transactioncoststhatmakemarketsforintellectualpropertyfunctionratherpoorly
intheICTsector.Inparticular,thepatenthold‐upproblemwoulddisappearifit
werepracticalforfirmstoidentifyrelevantpatentsandtake‐outallofthenecessary
licensesbeforedesigning,buildingandmarketingnewproducts.Thisisfarfromthe
currentreality.WhileprivatesolutionssuchasSSOsandpatentpoolsmayprovidea
work‐aroundtotheproblemsofpatenthold‐upandroyalty‐stacking,thisessayhas
highlightedsomeoftheinherentcosts,limitationsandtrade‐offsoftheprimarily
privateapproach.
References
ANSI(2011).GuidelinesforImplementationoftheANSIPatentPolicy.NewYork,NY:AmericanNationalStandardsInstitute.Retrievedfromhttp://publicaa.ansi.org/
Biddle,B.,A.WhiteandS.Woods(2010).“HowManyStandardsinaLaptop?(AndOtherEmpiricalQuestions)”AvailableatSSRN:http://ssrn.com/abstract=1619440.David,P.A.,&Greenstein,S.(1990).Theeconomicsofcompatibilitystandards:Anintroductiontorecentresearch.EconomicsofInnovationandNewTechnology,1,3‐42.Farrell,J.,&Shapiro,C.(2008).Howstrongareweakpatents?AmericanEconomicReview,98(4),1347‐1369.
20
Farrell,J.,Pil‐Choi,J.,Edlin,A.,Greenstein,S.,Hall,B.H.,&Saloner,G.(2004).BriefamicuscuriaeofeconomicsprofessorsandscholarsintheMatterofRambus,Inc.,U.S.FederalTradeCommission,DocketNo.9302.Retrievedfromhttp://www.ftc.gov/os/adjpro/d9302/040415scholarsamicusbrief.pdf.Farrell,J.,Hayes,J.,Shapiro,C.,&Sullivan,T.(2007).Standardsetting,patentsandhold‐up.AntitrustLawJournal,74,603–670.FTC(2011).TheevolvingIPmarketplace:aligningpatentnoticeandremedieswithcompetition.ReportoftheU.S.FederalTradeCommission.Retrievedfromhttp://www.ftc.gov/bc/workshops/ipmarketplace/.Grindley,P.andD.Teece.(1997).ManagingIntellectualcapital:licensingandcross‐licensinginsemiconductorsandelectronics.,CaliforniaManagementReview,39(2),8‐41.Herman,M.(2010).Negotiatingstandards‐relatedpatentlicenses:howthedealisdone.Landslide,3(1),31.Layne‐Farrar,A.,Padilla,J.,&Schmalensee,R.(2007).Pricingpatentsforlicensinginstandardsettingorganizations:MakingsenseofFRANDcommitments.Workingpaper,retrievedfromhttp://papers.ssrn.com/sol3/papers.cfm?abstract_id=996700.Lemley,M.A.(2002).Intellectualpropertyrightsandstandardsettingorganizations.CaliforniaLawReview,90,1889‐1981.
Lemley,M.A.(2007).Tenthingstodoaboutpatentholdupofstandards(andonenotto).BostonCollegeLawReview,48,149‐168.
Lemley,M.A.,&Shapiro,C.(2007).Patenthold‐upandroyaltystacking.TexasLawReview,85,1991‐2049.
Majoras,D.(2005).Recognizingthepro‐competitivepotentialofroyaltydiscussionsinstandardsetting.Retrievedfromhttp://www.ftc.gov/speeches/majoras/050923stanford.pdfMiller,J.(2007).Standard‐setting,patentsandaccesslock‐in:RANDlicensingandthetheoryofthefirm.IndianalawReview,40,351‐358.Rysman,M.,&Simcoe,T.(2008).Patentsandtheperformanceofvoluntarystandardsettingorganizations.ManagementScience,54(11),1920‐1934.
Shapiro,C.&Varian,H.(1998).Informationrules:Astrategicguidetothenetworkeconomy.Boston,MA:HarvardBusinessSchoolPress.
Simcoe,T.(2006).Explainingtheincreaseinintellectualpropertydisclosure.InThestandardsedge:Goldenmean,S.Bolin(ed.).TheBolinGroup.
Simcoe,T.,Graham,S.,&Feldman,M.(2009).Competingonstandards?Entrepreneurship,intellectualpropertyandplatformtechnologies.JournalofEconomicsandManagementStrategy,18(3),775–816.
21
Swanson,D.,&Baumol,W.(2005).Reasonableandnondiscriminatory(RAND)royalties,standardsselection,andcontrolofmarketpower.AntitrustLawJournal,73(1),51‐56.
Williamson,O.(1985).Theeconomicinstitutionsofcapitalism.TheFreePress.