15
Formal Approaches to Management Theory Ron Adner (Dartmouth) László Pólos (Durham) Michael Ryall (Melbourne) Olav Sorenson (Toronto) A perusal of the pages of the Academy of Management Review (AMR) over the past decade might lead one to conclude that formal methods have no place in the development of management theory. From 1998 to 2007, for example, we could only find one article in this journal that actually used a formal approach – either analytic methods, a simulation or formal logic – to build its propositions. One therefore might naturally wonder why formal approaches have not been more prevalent in management research. One possibility is that formal methods have little to contribute to management theory. That explanation seems unlikely to us (and not just because the articles in this issue serve as examples that strongly contradict it). Consider the disciplines on which management theory most commonly draws: economics and sociology. Nearly every theoretical article over the last decade that appeared in the American Economic Review used formal methods (almost exclusively analytic models). Even in sociology, the last ten years of the American Journal of Sociology (AJS) featured at least ten articles that built theory using formal approaches (divided fairly evenly across the methods featured in this special topic forum). Given that AJS publishes far fewer articles per year than AMR and that more than half of those articles are empirical, the ratio of ten to one would actually appear to understate the extent to which sociology relative to management favors the use of formal methods. Seeing as closely related social sciences have gained greatly from formal approaches, one might expect that management theory could benefit similarly.

Formal Approaches to Management Theory

Embed Size (px)

DESCRIPTION

Formal Approaches to Management Theory

Citation preview

Page 1: Formal Approaches to Management Theory

FormalApproachestoManagementTheory

RonAdner(Dartmouth)

LászlóPólos(Durham)

MichaelRyall(Melbourne)

OlavSorenson(Toronto)

AperusalofthepagesoftheAcademyofManagementReview(AMR)overthepast

decademightleadonetoconcludethatformalmethodshavenoplaceinthe

developmentofmanagementtheory.From1998to2007,forexample,wecould

onlyfindonearticleinthisjournalthatactuallyusedaformalapproach–either

analyticmethods,asimulationorformallogic–tobuilditspropositions.One

thereforemightnaturallywonderwhyformalapproacheshavenotbeenmore

prevalentinmanagementresearch.

Onepossibilityisthatformalmethodshavelittletocontributetomanagement

theory.Thatexplanationseemsunlikelytous(andnotjustbecausethearticlesin

thisissueserveasexamplesthatstronglycontradictit).Considerthedisciplineson

whichmanagementtheorymostcommonlydraws:economicsandsociology.Nearly

everytheoreticalarticleoverthelastdecadethatappearedintheAmerican

EconomicReviewusedformalmethods(almostexclusivelyanalyticmodels).Evenin

sociology,thelasttenyearsoftheAmericanJournalofSociology(AJS)featuredat

leasttenarticlesthatbuilttheoryusingformalapproaches(dividedfairlyevenly

acrossthemethodsfeaturedinthisspecialtopicforum).GiventhatAJSpublishesfar

fewerarticlesperyearthanAMRandthatmorethanhalfofthosearticlesare

empirical,theratiooftentoonewouldactuallyappeartounderstatetheextentto

whichsociologyrelativetomanagementfavorstheuseofformalmethods.Seeingas

closelyrelatedsocialscienceshavegainedgreatlyfromformalapproaches,one

mightexpectthatmanagementtheorycouldbenefitsimilarly.

Page 2: Formal Approaches to Management Theory

Asecondpossibilityisthatmanagementscholarshavenotinvestedinacquiringthe

skillsrequiredtouseformalmethods.Thatexplanationtooappearstofallshort,

especiallywhenoneconsiderstheresponsetothecallforthisspecialtopicforum.

Wereceivedatotalof75submissions,awealthofriches.Almosteveryoneofthese

submissionsmoreovercamefromsomeonesittinginabusinessschoolandmore

than90%oftheauthorsofthesemanuscriptsholddoctoraldegreesfrombusiness

schools(ratherthan,say,fromeconomicsdepartments).Wealsobenefitedfromthe

adviceofanextremelyhighqualitypoolofreviewers,manyofwhomhadnever

beforebeeninvolvedwithAMR.Thisoverwhelmingresponsesurprisedevenus,the

SpecialTopicForumguesteditors.Butitnonethelessrevealsthatalargesegmentof

managementscholarshavethetrainingtopursueformalapproachestobuilding

theory,andthusascarcityofskillscannotaccountfortheinfrequentusageofthese

toolsinmanagementresearch.

Thoughitfallsbeyondthescopeofthisintroductoryessaytodetermineprecisely

thereasonsforthepaucityofformaltheorybuildinginmanagementresearch,we

speculatethatatleasttwofactorscomeintoplay.First,wesuspectthatreaders,

reviewersandeditorsmaynotfullyappreciatetherelativeadvantagesofformal

versusinformalapproaches.Theymaythereforefeelthattheinvestmentinvolved

inreadingandunderstandingtheseformalapproachesistoodear.Second,those

thatdouseformalmethodsmaynotconsiderAMRasapotentialoutletsimply

becauseithaspublishedsofewarticlesusingthesemethodsinthepast.Atleast

partlyinsupportofthissecondpoint,wewouldnotethatManagementScienceand

OrganizationScienceregularlypublishpapersthatuseanalyticmethods,

simulationsandformallogictobuildmanagementtheory.

Withthisspecialtopicforum,wehopetostimulateboththedemandforandthe

supplyofmanagementresearchusingformalapproaches.Bydescribingtheir

advantagesandbydemonstratingthemthroughexample,wehopethatreadersand

reviewerswillbecomemoreattunedtothevalueofresearchthatadoptsaformal

approachtotheorybuilding.And,bydemonstratingthatAMRhasaninterestin

Page 3: Formal Approaches to Management Theory

publishingthesepapers,wehopetoencouragemanagementresearcherstouse

themmoreoften.

Thebenefitsofbuildingtheoryformally

Formalapproachestotheorybuildingcomeinthreemainflavors,definedprimarily

bythemethodsusedforarticulatingassumptionsandthenmovingfromthose

assumptionstopropositions.Inmathematicalmodels,theresearcherbeginsby

outliningasetofmathematicalconditionsthatdescribethephenomenonofinterest

andthenusesmathematicalproofstodemonstratepropositionsthatfollowfrom

them.Insimulations(alsoknownascomputationalmodels),onesimilarlywritesa

setofconditionsthatrepresenttheassumptionsofthemodel,butinsteadof

validatingtheirimplicationsviaanalyticalproofs,onegeneratesoutcomes

computationallyacrossrangesofvaluesfortheparametersthatdeterminethose

outcomes.Formallogic,meanwhile,translatesnaturallanguageassumptionsinto

statementsinsymboliclogicandthenusesitsownmethodsforprovingthe

propositionsimpliedbythosestatements.Althoughtheseapproachesvary

somewhatintheirstrengthsandweaknessesforaddressingparticulartheoretical

questions,theynonethelessshareatleastthreestrengthsrelativetoverbal

theorizing(thenaturallanguageapproachmostcommonlypursuedinmanagement

research):(1)Precisionandtransparency;(2)Logicalconsistency;and(3)An

abilitytoidentifyunanticipatedimplications.

Precisionandtransparency:Debatesoverthemeaningofcertaintermsinthe

managementliteraturehavebecomenotorious.Whatconstitutesa“resource”inthe

resource‐basedview?Whatisa“dynamiccapability”?Whatdoesitmeanto“learn”,

ortobe“boundedlyrational”?Althoughthesedisputesoccasionallyconverge

towardoneaccepteddefinition,theycanjustaseasilycontinuewithoutany

apparentresolutiondespitedecadesofdebate.Management,moreover,isnotalone

inthissituation;allofthesocialscienceshavehadvigorous(andfrequently

unresolved)argumentsoverthemeaningsofkeyconcepts.

Page 4: Formal Approaches to Management Theory

Inlargepart,thesedebatesovertermsandtheapplicabilityoftheoriesreflectthe

imprecisionofnaturallanguages,suchasEnglish.Naturallanguageshaveevolved

overtimeandcontinuetochange.Foravarietyofreasons,many,ifnotmost,words

innaturallanguagesthereforehavemultiplemeanings.Thelistenerorreader

attemptstoinferwhichoneofthesemeaningsthecommunicatorintendedby

consideringthecontextinwhichtheyfindit.Butthisprocessleavesampleroomfor

ambiguityandsubjectiveinterpretation.Onelistenerorreadermightdecidethata

phrasemeanssomethingverydifferentfromwhatanotheronedoes.

Althoughonecanattempttocommunicateideasclearlythroughnaturallanguages,

theveryflexibilityandinstabilityoftheselanguagesmakesdoingsoanuphillbattle.

Bycontrast,mathematicsandformallogichavebeendesignedwithprecisionin

mind.Becauseofthisprecision,translatinganaturallanguageideaintoaformal,

symbolicrepresentationcanbechallenging.Whereasonecanstateanassumption

orapropositioninnaturallanguagewithoutevenfullyunderstandingoneselfthe

meaningofthestatement,formallanguagesrequirethatthecommunicator

understandstheconceptdeeply.Infact,wewouldassertthatmuchofthevalueof

usingformalmethodsstemsfromthefactthatitforcestheresearchersusingthem

tothinkthoroughlythroughtheconceptsthattheyinvoke.Oncethetheoryhasbeen

translatedintoaformallanguage,everyonewhoseesitknowsexactlywhatit

means.Thoughtheinformalinterpretationofthetermsoftheformaltheorymight

varyfromreadertoreader,theformalcharacterizationensuresthatallofthese

interpretationshaveacommoncore.

Thatprecisionhasatleasttwoadvantagesinmanagementresearch.First,it

facilitatestheaccretionofknowledge.Ifdifferentresearchersusethesametermto

refertoseveraldifferentconceptswhentheydeveloptheoreticalpropositions,then

itbecomesexceedinglydifficulttounderstandwhetherandhowtheirarguments

interact.Sometimes,researchersduplicateeffortbyforwardingfundamentally

equivalentideasundernewnames.Atothertimes,progresscomestoahaltwhen

researchersappeartoarriveatcontradictoryconclusions(perhapsbecausethey

usethesamenaturallanguagetermsbutimbuethemwithdifferentmeanings).

Page 5: Formal Approaches to Management Theory

Second,empiricalresearcherscanmoreeasilyandaccuratelytesttheoretical

propositionswhenlittleambiguitysurroundsthemeaningof–andhencethe

appropriatemeasurementof–thecentralconcepts.Progressinempiricalresearch

comesprimarilyfromtheabilitytofalsifytheories.Butwithoutprecise

propositions,oneoftencannotsaywhetheranempiricaltestrefutesatheoryor

whetheritsimplyfailedtooperationalizeitadequately.

Logicalconsistency:Nooneintentionallyforwardsalogicallyinconsistentargument,

butitisnonethelesssurprisinglyeasytodo.Forexample,onemightnotrecognize

animplicitassumptionandthereforefailtoidentifysomecriticalcondition

necessaryforthetheorytohold.Or,onemightmissthefactthattwo(ormore)

assumptionsactinoppositedirections.Or,lesscrucially,onemightsimplyimpose

anunnecessary,possiblyredundant,assumption.Astheoriesbecomeincreasingly

involvedandincorporatemoreandmorefactors,theopportunitiesforerrorsof

omissionandcommissioninevitablyincrease.

Buteveniftheassumptionshavebeencorrectlyspecifiedandstated,peoplestill

regularlycometothewrongconclusionsabouttheimplicationsofthose

assumptions.Theproblemisthatwhenrelyingonverbaltheorizing,peoplerarely

thinklogicallythrougheachstageofthechainofanargument;rather,theyrelyon

theirintuitionstoidentifytheimplications.Ourintuitions,however,oftenprove

wrong.ConsidertheMontyHallproblem(Selvin,1975).Acontestantonagame

showcanchoosefromcurtainsA,BandC.Twoofthesecurtainshavegoatsbehind

themandonehasacar;ifhepickstheonewiththecar,hewinsit.Let’sassumethat

heinitiallychoosescurtainA.ThehostthenopenscurtainB,whichhasagoat

behindit,andasksthecontestantwhetherhewouldliketochooseadifferent

curtain.Shouldheswitch?Mostpeoplebelievethateachoftheremainingtwo

curtainshasa50%chanceofhavingthecarand,therefore,thatthecontestant

shouldbeindifferentbetweenstayingwithhisoriginalchoiceandshiftingtocurtain

C(GranbergandBrown,1999).Butmathematicsdemonstratesclearlythatthe

contestantshouldswitch.Infact,hisoriginalchoice,curtainA,hasonlyaone‐third

chanceofhavingthecarbehindit,whilecurtainChasatwo‐thirdsprobabilityof

Page 6: Formal Approaches to Management Theory

beingthewinningcurtain(becausethehostintentionallychoosesacurtainthat

doesnothavethecarbehindit,therebyrevealinginformationabouttheother

option).Onecouldthereforeeasilyimaginethatsomeonedevelopinganatural

languagetheoryoftheMontyHallproblemmightpositanintuitivelyappealingbut

incorrectproposition.Moregenerally,researchersthatrelyentirelyonverbal

theorizinghavelittletopreventthemfromarrivingaterrantconclusions.

Allformalapproachessharetheadvantageofhavingtoolsavailableforensuring

logicalconsistency.Analyticmodelsandformallogicuseproofs.Simulations

meanwhilecalculatetheimplicationsofavarietyofscenariosnumerically.Inanyof

theseapproaches,researchersmustexplicateallofthenecessaryassumptions

because,withoutthem,onecouldnotderivetheresults.Redundantandsuperfluous

assumptionsmoreoverbecomerelativelyobviouswhenwritingproofs(though

thoseusingsimulationscannotspotthemsoeasily).Proofsandcomputational

methods,moreover,ensurethattheresultsdofollowfromtheassumptions.Inthe

MontyHallexample,onecouldhaveusedanyoftheseapproachestoarriveatthe

correctanswer.Theresearchermightmakeanerrorintheapplicationofthese

methods,but,becausetherulesareclearlydefined,readersandreviewersalikecan

easilyaudittheassumptionsandthechainsoflogicleadingtothepropositions.

Unanticipatedimplications:Theavailabilityofthesetoolshasanotheraddedbenefit:

onecanusethemtoderiveunanticipatedimplicationsoftheassumptions.Most

exercisesinverbaltheorizinginvolverecognizingarelationshipbetweentwo

factors,sayresourcesandprofitability,andthentryingtodetermineasetof

assumptionsthatwouldleadtosucharelationship.Inessence,onemoves

backwardsfrompropositionstoassumptions.Acleardisadvantageofthisapproach

isthatonecannotfindtheunexpected(thoughitmaylaterrearitsheadinempirical

researchtestingthesepropositions).Bycontrast,formalapproachescanhelpthe

researchertoidentifyeventhemostsurprisingandcounterintuitiveimplications

thatmightfollowfromasetofassumptions.

Page 7: Formal Approaches to Management Theory

Consumersofresearchusingformalapproachesoftencannotfullyappreciatethis

abilitytoidentifyunanticipatedimplicationsfortworeasons.Insomecases,when

theseimplicationsclearlyruncountertoreality,theymayforcethetheoristto

revisittheinitialpremisesofthemodelandthereforetheymayappearonlyasan

intermediate(andunpublished)stageoftheresearch.Inothercases,bythetime

theyappearinprint,notonlyhavetheseimplicationsbeenderived,butalsothe

intuitionunderlyingthemhasbeenexplicatedandthereforetheresults–though

originallysurprising–mayevenseemsomewhatobvious(thoughonlyin

retrospect).

Butthevalueofthisabilitytoexplicatemorecompletelytheimplicationsofasetof

assumptionscanbeseeninthebreadthofpropositionsderivedfromrelatively

simplemodels.Readersnewtoformalapproachesareoftensurprisedbythe

complexityofthebehaviorsthatemergefromrelativelyfewassumptions.

Simulationsofthebehaviorofflocksofbirds,herdsofanimalsandschoolsoffish,

forexample,havedemonstratedthataverysmallnumberofrules(asfewasthree)

cangeneraterealisticsimulationsofgroupsofanimalsinmotion(Reynolds,1987).

Inmanagementresearch,Postrel(thisissue)alsoillustratesthewiderangeof

implicationsthatonecanderivefromasmallsetofassumptions.

Beyondsimplygeneratingadditionalinterestingpropositions,thisabilitytoidentify

unintendedimplicationsalsoacceleratesprogressinunderstandingaphenomenon.

Additionalpropositionsofferempiricalresearchersmoretargets,more

opportunitiestofalsifythetheory.Ifthepropositionsfollowfromtheassumptions

andempiricalanalysisfalsifiesoneormoreofthesepropositions,thenitsuggests

thanoneormoreoftheassumptionsmustnotholdaswell.Researchersmustthen

revisittheirtheories.

Analyticalmodels,simulationsandformallogic

Despitetheseadvantages,eachapproachtoformaltheorybuildingisnotwithoutits

limitations.Theuseofmathematicalmethods,forexample,sometimesrequires

unappealingassumptions.Simulationshavethedisadvantageofonlybeingableto

Page 8: Formal Approaches to Management Theory

provetheexistenceofarelationshipbetweenasetofinputsandanoutcome;they

cannotdeterminethenecessityofasetofconditions.Formallogic,meanwhile,does

notasreadilyaccommodatefunctionalformsofrelationshipsandthereforecannot

alwaysadjudicatebetweentheexpectedeffectsofcountervailingforces.

Eachoftheselimitationsneverthelessappliesonlytoaparticularformalapproach.

Thesetradeoffsthereforemightconstituteareasonforchoosingoneformalmethod

overanother,butnoneofthemsuggeststhatonewouldwanttoeschewformal

methodsaltogether.Althoughitisbeyondthescopeofthisintroductiontoreview

eachapproachindetail,letusexpandalittleontherelativestrengthsofeach.1

Analyticmodels:Someseeanalyticmodelsasthe“goldstandard”offormal

approaches.Thetranslationofassumptionsanddefinitionsintomathematical

statementsandthentheuseofproofstovalidatepropositionsmeanthatthese

methodsimposeastrictdisciplineonclarityandlogicalconsistency.Theuseof

mathematics,moreover,allowsonetomodeltheinteractionsofmany,many

“movingparts”–piecesofthemodelthatinfluenceitsbehavior–withthe

confidencethatone’sconclusionsdoindeedfollowfromone’spremises.

Boththegreateststrengthandthegreatestlimitationofanalyticmethodsstemfrom

theneedtoproduceamodelwithanalyticallytractablesolutions.Onthepositive

side,mathematicalmethodsencouragethetheoristtodistillasituationtoits

essence.Consider,forexample,thePrisoner’sDilemma.Thisscenarioelegantly

capturestheproblemofshirkinginteamcooperation.Itdoesnotdescribeallofthe

complexityofareal‐lifeencounter,butitdoescapturetheproblem’sessential

features.Byabstractingawayfromthedetailsofaparticularcase,itprovidesdeep

insightintoanentireclassofsituations.Indeed,theartofanalyticanalysisappears

intheabilitytoidentify,toabstractandtomatchtoamathematicalapproachthe

1Forthosewhowishtodelvemoredeeplyintooneormoreoftheseapproaches,wewouldrecommendHarrisonetal(2007)forsimulationandHannan,PólosandCarroll(2007)forformallogic.Mathematicsincludessuchavastarrayofapproachesandtoolsthatwecannotrecommendanysinglesurvey.

Page 9: Formal Approaches to Management Theory

essentialelementsofthesituationbeingmodeled.Butsometimesthedepictionofa

situationdoesnotfitwellwithanyanalyticalframework.Attemptstoforceitinto

onemaythereforeresultinatheorywithlittlefacevalidity.

Simulation:Insituationsthateludemathematicalmethods,simulationmayprovide

anattractivealternative.Beforemovingforward,weshouldnotethatalthough

simulationresearchsometimesreferstoparticularsetsofinputsas“experiments”

andtheresultsoftheseexperimentsas“data”,simulationisalmostalwaysan

exerciseintheorybuilding;itisnotempiricalresearch.

Simulationsrunthegamutfromthesimpletothecomplex.Theprimaryadvantage

ofthesecomputationalmodelsrelativetoanalyticmethodsisthattheyallowmuch

greaterflexibilityincertainsortsofassumptionsmadebythetheorist.Forexample,

simulationscanaccommodateanyfunctionalformthatonemightimaginelinking

two(ormore)variables.Butsimulationsalsofacetheirownconstraints.The

investigationofsomeclassesofproblems–suchasgameswithaninfinitenumber

ofperiods–remaincomputationallyintractableevenwithaccesstocolossal

computerpower.Whethermathematicalmethodsorsimulationoffersthe

preferableapproach,therefore,dependsontheproblem.

Relativetoanalyticmodels,simulationshaveatleasttwodisadvantages.First,

becausethemethoditselfimposesrelativelylessdisciplineonthetheorist,these

modelscanbecomesocomplexthattheirinnerworkingsendupbeingopaque.One

canstillobservehowoutcomesvarywithchangesintheinputs,buttheinteractions

becomesoconvolutedastopreventtheresearcherfrombeingabletodetermine

whytheserelationshipsexist.Whentheindividualcomponentsofthesimulationare

wellunderstood,however,evensuchhighlydetailedandcomplexsimulationscan

beuseful.Forexample,physicistsoftenbuildincrediblycomplicatedsimulationsto

determinetheexpectedbehavioroflarge‐scalesystems.Theyunderstandthe

behaviorofeachcomponent–forexample,amoleculeoranelementaryparticle–

extremelywell,andtheyalsohavedetailedknowledgeofhowsmallnumbersof

Page 10: Formal Approaches to Management Theory

thesecomponentsinteract,buttheymustnonethelessrelyonsimulationstopredict

thebehaviorofhundreds,thousands,millionsorevenbillionsofthesecomponents.

Second,simulationdoesnotconstituteaproof.Simulationssampleafinite–though

potentiallyverylarge–numberofcasesfromaninfinitesetofpossibilities(atleast

whenthemodelincludescontinuousparameters).Onethereforecansaythatsome

specificsetofassumptionsleadstoaparticularoutcome,butonedoesnotknow

whetheranyofthoseassumptionsarenecessary,orwhethersomeunanalyzedset

ofassumptionsmightalsoprovesufficienttoyieldtheresults.Samplingthe

parameterspaceaswidelyandwithasfinearesolutionaspossiblecanhelptoallay

theseconcerns,butitdoesnotalleviatethem.

Inmanycases,theoristscancombinesimulationmethodswithanalytictechniques,

capturingalittlebitofthebestofbothapproaches.Onemight,forexample,builda

simplifiedmodelthatdoesnotincorporatealloftheinterestingfeaturesofa

phenomenonbutthatdoesfitwithintherestrictionsofsomemathematical

framework.Aftercharacterizingthatmodel,onemightthenexploitcomputational

methodstorelaxsomeoftheassumptionsortoextendthefeaturesofthemodel.

Suchanapproachnicelyexploitstheabilityofanalyticmethodstoidentifyboth

necessaryandsufficientconditions,whilestillallowingtheconsiderationofthose

casesthatstymieanalyticalanalysis(butthatremainamenabletosimulation).

Formallogic:Inmanyways,formallogicissimilartomathematicalmethods.Indeed,

mathematicsusesfirst‐orderlogic.Sincemanyoftheapplicationsofformallogicto

managementandorganizationsalsorelyonafirst‐orderlogic,suchasPéli(this

issue),thesepapersdifferlittleintheirrelativestrengthsandlimitationsfrom

mathematicalmethods.

Thebiggestdifferencebetweenformallogicandbothanalyticandsimulation

methodsariseswhenresearchersselectordefinealternativelogics(i.e.otherthan

first‐orderlogic).Inmanagementresearch,forexample,Hannan,PólosandCarroll

(2007)havedevelopedanon‐monotoniclogic,aformallanguage,fortheory

building(foranexampleoftheapplicationofthislogic,seeKuilman,Vermuelenand

Page 11: Formal Approaches to Management Theory

Li,thisissue).Thislogicincludesanumberofmodifiersandrelationshipsthatallow

theresearchertorepresentmoreaccuratelytheincompletenessofourempirical

understandingofaphenomenon.Forexample,itallowsquantifierslike“normally”

and“presumably”thatfirst‐orderlogiccannotaccommodate(anddefinesa

consistentsetofrulesfordealingwiththesequantifiers).Itthereforecanprovidea

morerealisticbasisfordevelopingandcombiningtheoryfragmentsthatdescribe

phenomenathatstillrequireagreatdealofempiricalexploration.

Althoughtheabilitytodevelopthelanguageusedtobuildtheorymightseemto

reducedramaticallythedisciplineimposedbyformalizationontheorists,logicians

havedevelopedastrictsetofrulesthatallformallanguagesshouldmeettoensure

theirprecisionandcompleteness.Allformallogicsthereforehavealltherigorof

mathematicalmethods,requiringpreciseassumptionsanddefinitionsandallowing

onetoprovepropositionsandtoidentifynecessaryandsufficientconditions.

Thoughformalmethodsprovidepowerfultoolsfordevelopingtheory,wewouldnot

claimthatverbaltheorizing–thatis,theuseofnaturallanguagestobuildtheory–

hasnoplaceinmanagementresearch.Whereasformalmethodsoperate

deductively,naturallanguagesmorereadilyaccommodateinductionandintuition.

Hence,whenaninterestingphenomenonisfirstidentified,conjectures,expressedin

naturallanguages,canprovideausefulstartingpoint.Theseconjecturescanhelpto

guideempiricalresearchersintermsofwhatdatatheyshouldcollectandwhat

questionstheyshouldattempttoanswer.Asthephenomenonbecomesbetter

understood,researcherscanthenrevisitthetheorywithformaltoolstodevelopa

moreprecise,transparentandlogicallyconsistentdescriptionofit.

Thepapersinthisissue

Asnotedabove,wehadanembarrassmentofrichesfromwhichtochoose.We

selectedpapersforthisspecialtopicforumnotjustfortheirhighqualityandthe

importanceoftheirmanagerialimplications,butalsofortheirabilitytohighlightthe

varietyandvalueofformalapproaches.

Page 12: Formal Approaches to Management Theory

Becauseofthebreadthofpapersinthespecialissue,onecouldchoosetoorder

theminmultipleways.Wehavechosenaroughgroupingbysubject,butthepapers

havemanyotherpointsofcontact.

Thefirstthreepapersdealbroadlywiththeissueofentrepreneurialentry,buteach

approachesitusingadifferentformalmethod.ParkerandAlvarez(thisissue)

modelthedecisionproblemofhowapairofentrepreneursshouldallocatethe

rightstocontrolanewventure.Althoughperfectlyrationalactorswouldgive

controltothehigherabilitymemberoftheteam,theyinterestinglydemonstrate

that,underconditionsofsubjectiverationality,theteamwillallocatecontrolrights

tothemoreoptimisticofthetwo.2Thismisallocationofcontrolrightscanlead

startupstoperformpoorly.Giventhewell‐knownpropensityofentrepreneurstobe

over‐optimistic,thisresultstrikesusaparticularlyinteresting.

GancoandAggrawal(thisissue)meanwhiledevelopasimulationtodevelopabetter

understandingoftheperformancedifferentialsbetweenstartupsanddiversifying

entrants,andhowthosedifferentialsdependontheenvironment.Theybuildtheir

simulationonavariantoftheNKmodelthathasbeenwidelyusedinmanagement

research(e.g.,Rivkin,2000).Theirmodel,extendingthecounter‐intuitiveideathat

adaptationactuallyhurtsorganizationsinturbulentenvironments,canbothaccount

foralloftheexistingempiricalregularitiesthathavebeenfoundandsuggests

severalnewpropositionsthatresearchersshouldexplore.

Finally,Kuilman,VermuelenandLi(thisissue)considerhowthenatureofthe

environmentatthetimeoffoundinginfluencesthelifechancesofentrants.

Empiricalresearchhasusuallyfoundthatfirmsthatenterduringperiodsofintense

competitivecrowdingsufferhighermortalityrates.Butsomestudieshavealso

foundtheoppositepattern–lowermortalityratesamongthosethatenterduring

timesofhigherdensity.Thearticlesreportingthesefindingsappeartoarriveat

2Subjectiverationalityreferstotheideathatactorsmustbehaveinawaythatappearsoptimalgiventheirbeliefsandobservationsoftheenvironment,butthatmaynotberationalfortheomniscientactor(Ryall,2003).

Page 13: Formal Approaches to Management Theory

contradictoryconclusions,butKuilmanetal.demonstratethatonecanincorporate

themintoasingletheorybyintroducingthestageofthepopulation’shistoryasan

additionfactor.Whetherdensityatthetimeoffoundingconstitutesanadvantageor

adisadvantagedependsonwhetherthepopulationremainsinthelegitimating

phaseorwhetherithasachievedlegitimacyandinteractionshavebecomeprimarily

competitive.Inessence,theyintroduceascopecondition.

Weshouldnotethattheapplicationofthenon‐monotoniclogicinthispaperis

particularlyinstructive.WhereasthelogicthatKuilmanetal.usewouldhave

allowedthemtoresolvetheseconflictingfindingssimplybyassumingthat

legitimacytakesprecedenceovercompetition(orviceversa),theyavoidthiseasy

solution.Theyavoiditbecausetheyhavenogoodsubstantivereasontointroducea

precedenceorderingbetweencompetitionandlegitimacy.Theythereforewisely

avoidsuchanadhocassumption.

Thenexttwopapersinvestigate,againbroadly,theproblemofcoordinationand

incentives.Postrel(thisissue)introducesthenovelideaofgoalambiguityintothe

well‐understoodeffectsofsubstitutionversuscomplementarityinjointproduction.

Heshowsthatgoalambiguitycanleadtoafailureintheallocationofefforteven

whenbothactorshaveincentivesonlyintermsoftheirjointoutput.Althoughthis

failurelookslikeacoordinationproblem,standardsolutionstocoordination

failures–suchassequencingactors’decisions–maynotsolvethisproblemand

mayevenexacerbateit.Themanagerialimplicationsareclear:coordinationand

goal‐clarificationhavesynergisticeffectsonteamperformance.

CoffandMakadok(thisissue)similarlyexploreanissueofclearmanagerial

importance:Howcanaprincipal(anowner)ensurethatagents(employees)donot

freeride?Indevelopingtheirtheory,theypointoutthatthemanageractuallyhasat

leastthreeleversunderhercontrolinachievingthisoutcome:authority,incentives

andownership.Allofthesevaryfrombeingmorefirm‐liketobeingmoremarket‐

like.Managersthereforehaveeightdifferentformsofhybridorganizationattheir

Page 14: Formal Approaches to Management Theory

disposal,andCoffandMakadokdemonstratethatatleastoneofthemshould

generatetheoptimalallocationofeffortacrosstheagentsforanytypeoftask.

Finally,weendwithtwopiecesthatprovidenewinsightintomacro‐organizational

behavior.CowanandJonard(thisissue)developamodelofallianceformationin

whichsuccessfulpartnershiprequiresthattwofirmshavemoderatelysimilar

characteristics.Theydemonstratethatthissingleassumptioncangeneratea

surprisinglywidevarietyofnetworkstructures,includingatendencyforrepeated

interactionandsmall‐worldstructures.Theirresultsthereforeraisethepossibility

thattheneedforsomewhatsimilarpartners,ratherthansocialfactors,mayexplain

anumberofinterorganizationalstructures.Sinceonecouldimaginethe

developmentofaparallelmodelattheleveloftheindividual,theirresultsmayalso

provideinsightintointerpersonalnetworks.

Péli(thisissue)meanwhileattemptstoreconciletheideathatinertiaand

imprintingimplythatpopulation‐levelchangeshouldoccurthroughtheselection

(failure)oforganizationswiththeintuitionthatpopulationssometimesshift

throughtheadaptationoftheirindividualmembers.Toresolvethisapparent

inconsistency,hesuggeststhatenvironmentalchangemayeitherbequantitativeor

qualitativeinnature.Thefirst,aquantitativechange,mightforcefirmstoadjust

theirintensityofvariousactivities,butthelatterrequiresthemtodoentirely

differentthingstosurvive.Theformerthereforeallowsadaptation,whileselection

reignssupremeinthelatterregime.Hisapproachtoresolvingthisconundrum

thereforerecallsasimilarsolutionthathepositedtowhenbroadversusnarrow

scopewouldbenefitorganizations(Péli,1997).

WeapplaudtheauthorsincludedinthisSpecialTopicForumfortheircreativeand

compellingcontributionstomanagementtheoryandfortheirableapplicationsof

formalmethods.Wealsoappreciatetheeffortsofallofthosethattookaninterestin

ourcall.Wehopethatthepapersthatappearherebecomebutthebeginningina

longlineofinterestingandinsightfulinvestigationsthatuseformalapproachesto

developmanagementtheory.

Page 15: Formal Approaches to Management Theory

References

Granberg,Donald,andThadA.Brown(1999)“TheMontyHalldilemma.”Personality

andSocialPsychologyBulletin,21:711‐729

Hannan,MichaelT.,LászlóPólosandGlennR.Carroll(2007)LogicsofOrganization

Theory:Audiences,CodesandEcologies.PrincetonUniversityPress

Harrison,J.Richard,ZhiangLin,GlennR.CarrollandKathleenM.Carly(2007)

“Simulationmodelinginorganizationalandmanagementresearch.”Academyof

ManagementReview,32:1229‐1245

Péli,Gabor(1997).“Thenichehiker’sguidetopopulationecology:Alogical

reconstructionoforganizationalecology’snichetheory.”SociologicalMethodology,

27:1‐46

Reynolds,CraigW.(1987)“Flocks,herdsandschools:Adistributedbehavioral

model.”ComputerGraphics,21:25‐34

Rivkin,JanW.(2000)“Imitationofcomplexstrategies.”ManagementScience,46:

824‐844

Ryall,MichaelD.(2003)“Subjectiverationality,self‐confirmingequilibriumand

corporatestrategy.”ManagementScience,49:936‐949

Selvin,Steve(1975)“OntheMontyHallproblem.”AmericanStatistician,29:134

acer
Typewriter
http://dc122.4shared.com/download/rRbBfDAq/WALT_Time_Management_Theory.doc?tsid=20110407