Journal of Economic Perspectives—Volume 24, Number 4—Fall 2010—Pages 85–102

Macroeconomics after the Crisis: Time to Deal with the Pretense-of-Knowledge Syndrome

Ricardo J. Caballero

hhee rrecentecent fi nancialnancial crisiscrisis hashas damageddamaged tthehe rreputationeputation ofof macroeconomics,macroeconomics, llargelyargely forfor itsits inabilityinability toto predictpredict thethe impendingimpending fi nancialnancial andand economiceconomic T ccrisis.risis. ToTo bebe honest,honest, thisthis inabilityinability toto predictpredict doesdoes notnot concernconcern meme much.much. ItIt iiss aalmostlmost ttautologicalautological thatthat severesevere crisescrises areare essentiallyessentially unpredictable,unpredictable, forfor otherwiseotherwise ttheyhey wouldwould notnot ccauseause ssuchuch a hhighigh ddegreeegree ooff ddistress.istress. OOff ccourse,ourse, iitt iiss wwell-knownell-known tthathat ccertainertain eelementslements ccanan iincreasencrease tthehe ffragilityragility ooff a fi nancialnancial system,system, suchsuch asas highhigh levelslevels ofof lleverageeverage oorr mmismatchesismatches bbetweenetween sshort-termhort-term lliabilitiesiabilities aandnd llong-termong-term aassets,ssets, aandnd tthathat tthesehese iissuesssues maymay jjustifyustify policypolicy intervention.intervention. ButBut knowingknowing thesethese mechanismsmechanisms isis quitequite ddifferentifferent ffromrom aarguingrguing tthathat a ssevereevere ccrisisrisis ccanan bbee ppredicted.redicted. MModernodern CCassandrasassandras wwillill aalwayslways claimclaim ttoo hhaveave sseeneen tthehe ccrisisrisis ccoming.oming. WWhathat ttheyhey wwillill nnotot ssayay iiss hhowow mmanyany ttimesimes ttheyhey sawsaw thingsthings comingcoming thatthat nevernever materialized,materialized, oror howhow thethe specifispecifi c mechanismsmechanisms bbehindehind thethe crisiscrisis areare differentdifferent fromfrom tthosehose oonn wwhichhich ttheirheir ppredictionsredictions werewere based.based. IInn mmyy vview,iew, tthehe cconvictiononviction tthathat oonene ccanan fforetelloretell a ssevereevere ccrisisrisis iinn aadvancedvance iiss mmostlyostly a mmanifestationanifestation ofof ppareidolia—theareidolia—the ppsychologicalsychological pphenomenonhenomenon tthathat mmakesakes ppeopleeople sseeee ffacesaces aandnd aanimalsnimals iinn ccloudslouds aandnd tthehe llike.ike. WWhathat ddoesoes concernconcern meme aboutabout mymy discipline,discipline, however,however, isis thatthat itsits currentcurrent corecore ——byby wwhichhich I mainlymainly meanmean thethe so-calledso-called dynamicdynamic stochasticstochastic generalgeneral equilibriumequilibrium aapproach—haspproach—has becomebecome soso mesmerizedmesmerized withwith itsits ownown internalinternal logiclogic thatthat itit hashas begunbegun ttoo cconfuseonfuse thethe precisionprecision itit hashas achievedachieved aboutabout itsits ownown worldworld withwith thethe precisionprecision tthathat itit hashas aboutabout thethe realreal one.one. ThisThis isis dangerousdangerous forfor bothboth methodologicalmethodological andand ppolicyolicy reasons.reasons. OnOn thethe methodologymethodology front,front, macroeconomicmacroeconomic researchresearch hashas beenbeen inin ““fifi nne-tuning”e-tuning” modemode withinwithin thethe local-maximumlocal-maximum ofof thethe dynamicdynamic stochasticstochastic generalgeneral eequilibriumquilibrium wworld,orld, wwhenhen wwee sshouldhould bbee iinn ““broad-exploration”broad-exploration” mmode.ode. WWee aarere ttoooo ffarar

■ RRicardoicardo J.J. CaballeroCaballero isis thethe FordFord InternationalInternational ProfessorProfessor ofof ,Economics, MassachusettsMassachusetts Insti-Insti- ttuteute ofof Technology,Technology, andand alsoalso a ResearchResearch Associate,Associate, NationalNational BureauBureau ofof EconomicEconomic Research,Research, bbothoth iinn CCambridge,ambridge, MMassachusetts.assachusetts. HHisis ee-mail-mail aaddressddress iiss 〈[email protected]@mit.edu〉. doi=10.1257/jep.24.4.85 86 Journal of Economic Perspectives

ffromrom aabsolutebsolute ttruthruth ttoo bebe soso specializedspecialized andand toto makemake thethe kindkind ofof conficonfi dentdent quan-quan- ttitativeitative claimsclaims thatthat oftenoften emergeemerge fromfrom thethe core.core. OnOn thethe policypolicy front,front, thisthis confusedconfused pprecisionrecision createscreates thethe illusionillusion thatthat a minorminor aadjustmentdjustment iinn tthehe sstandardtandard policypolicy frame-frame- wworkork willwill preventprevent futurefuture crises,crises, andand byby doingdoing soso itit leavesleaves usus overlyoverly exposedexposed toto thethe nnewew aandnd uunexpected.nexpected. TToo bebe fairfair toto ourour fi eld,eld, anan enormousenormous amountamount ofof workwork atat thethe intersectionintersection ofof mmacroeconomicsacroeconomics aandnd ccorporateorporate fi nancenance hashas beenbeen chasingchasing manymany ofof thethe issuesissues thatthat pplayedlayed a centralcentral rolerole duringduring thethe currentcurrent crisis,crisis, includingincluding liquidityliquidity evaporation,evaporation, ccollateralollateral shortages,shortages, bubbles,bubbles, crises,crises, panics,panics, fi rere sales,sales, risk-shifting,risk-shifting, contagion,contagion, andand tthehe llike.ike.1 HHowever,owever, mmuchuch ofof thisthis literatureliterature belongsbelongs toto thethe periphery ooff macroeco-macroeco- nnomicsomics rratherather thanthan toto itsits core.core. IsIs tthehe ssolutionolution thenthen toto replacereplace thethe currentcurrent corecore forfor tthehe pperiphery?eriphery? I aamm ttempted—butempted—but I tthinkhink tthishis wwouldould aaddressddress oonlynly ssomeome ooff oourur pprob-rob- llems.ems. TThehe ddynamicynamic sstochastictochastic generalgeneral equilibriumequilibrium strategystrategy isis soso attractive,attractive, andand eveneven pplainlain aaddictive,ddictive, bbecauseecause iitt aallowsllows oonene ttoo ggenerateenerate iimpulsempulse rresponsesesponses tthathat ccanan bbee ffullyully ddescribedescribed inin termsterms ofof seeminglyseemingly scientifiscientifi c statements.statements. TheThe modelmodel isis anan irresistibleirresistible ssnake-charmer.nake-charmer. IInn ccontrast,ontrast, thethe peripheryperiphery isis notnot nearlynearly asas ambitious,ambitious, andand itit providesprovides mmostlyostly qqualitativeualitative insights.insights. SoSo wewe areare leftleft withwith thethe tensiontension betweenbetween a typetype ofof answeranswer ttoo wwhichhich wwee aaspirespire bbutut tthathat hhasas llimitedimited cconnectiononnection wwithith rrealityeality ((thethe ccore)ore) aandnd mmoreore ssensibleensible bbutut iincompletencomplete aanswersnswers ((thethe pperiphery).eriphery). TThishis ddistinctionistinction betweenbetween corecore andand peripheryperiphery iiss nnotot a mmatteratter ooff ffreshwaterreshwater vversusersus ssaltwateraltwater eeconomics.conomics. BBothoth tthehe rrealeal bbusinessusiness ccycleycle aapproachpproach aandnd iitsts NNewew KKeynesianeynesian ccounterpartounterpart belongbelong toto thethe core.core. Moreover,Moreover, therethere waswas a timetime whenwhen KeynesianKeynesian eco-eco- nnomicsomics wwasas mmoreore llikeike tthehe ccurrenturrent core,core, inin thethe sensesense ofof tryingtrying toto buildbuild quantitativequantitative aaggregativeggregative modelsmodels startingstarting fromfrom mmicro-foundedicro-founded cconsumptiononsumption functionsfunctions andand tthehe llike.ike. AAtt tthathat ttime,ime, iitt wwasas tthehe ““rational-expectations”rational-expectations” rrepresentativesepresentatives tthathat wwereere iinn tthehe iinsight-buildingnsight-building mmode,ode, iidentifyingdentifying kkeyey cconceptsoncepts fforor mmacroeconomicacroeconomic ppolicyolicy ssuchuch aass time-inconsistencytime-inconsistency andand endogenousendogenous expectations,expectations, withoutwithout anyany pretensepretense ofof beingbeing rrealisticealistic iinn aallll ddimensionsimensions ooff mmodelingodeling iinn oorderrder ttoo oobtainbtain qquantitativeuantitative aanswers.nswers. MMoreover,oreover, tthishis ttensionension iiss nnotot nnewew ttoo mmacroeconomicsacroeconomics oorr eevenven ttoo eeconomicsconomics mmoreore bbroadly.roadly. IInn hhisis NNobel-prizeobel-prize aacceptancecceptance llecture,ecture, HHayekayek ((1974)1974) wwrites:rites: ““OfOf ccourse,ourse, ccomparedompared wwithith tthehe ppreciserecise ppredictionsredictions wwee hhaveave llearntearnt ttoo eexpectxpect iinn tthehe pphysicalhysical ssciences,ciences, tthishis ssortort ooff mmereere ppatternattern ppredictionsredictions iiss a ssecondecond bbestest wwithith wwhichhich oonene ddoesoes nnotot llikeike ttoo hhaveave ttoo bbee ccontent.ontent. YYetet tthehe ddangeranger ooff wwhichhich I wwantant ttoo wwarnarn iiss ppreciselyrecisely tthehe bbeliefelief tthathat iinn oorderrder ttoo hhaveave a cclaimlaim bbee aacceptedccepted aass sscientificientifi c itit isis nnecessaryecessary ttoo aachievechieve mmore.ore. TThishis wwayay lliesies ccharlatanismharlatanism aandnd wworse.orse. TToo aactct oonn tthehe bbeliefelief tthathat wwee ppossessossess tthehe kknowledgenowledge aandnd tthehe ppowerower wwhichhich eenablenable uuss ttoo sshapehape tthehe pprocessrocess ooff ssocietyociety eentirelyntirely ttoo oourur lliking,iking, kknowledgenowledge wwhichhich iinn ffactact wwee ddoo nnotot ppossess,ossess, iiss llikelyikely ttoo mmakeake uuss ddoo mmuchuch hharm”arm” ((1974).1974). OOnene readingreading ofof Hayek’sHayek’s commentcomment isis asas a reminderreminder ofof thethe dangersdangers ofof ppresumingresuming a precisionprecision andand degreedegree ofof knowledgeknowledge wewe dodo notnot have.have. I suspectsuspect thatthat ifif HHayekayek wwasas cconfrontedonfronted wwithith tthehe llimitedimited cchoicehoice bbetweenetween tthehe ccoreore oorr tthehe pperipheryeriphery

1 In fact, at MIT we divide the fi rst-year Ph.D. macroeconomics sequence into four parts: methods, growth, fl uctuations, and crises. I will include specifi c references of this work in the main text, but see Part VI of Tirole (2006) for a nice survey and unifi ed explanation of several of these mechanisms. Ricardo J. Caballero 87

ooff macroeconomics,macroeconomics, hishis votevote todaytoday wouldwould bebe castcast forfor thethe periphery.periphery. ThisThis isis thethe sstartingtarting pointpoint ofof thethe themetheme I willwill developdevelop inin thethe fi rstrst partpart ofof thisthis paper.paper. ThereThere I willwill ddiscussiscuss thethe distinctiondistinction betweenbetween thethe corecore andand thethe peripheryperiphery ofof macroeconomicsmacroeconomics inin ggreaterreater detail,detail, asas wellwell asas thethe futilefutile nnatureature ofof thethe integrationistintegrationist movement—thatmovement—that is,is, tthehe processprocess ofof graduallygradually bringingbringing thethe insightsinsights ofof thethe peripheryperiphery intointo thethe dynamicdynamic sstochastictochastic ggeneraleneral eequilibriumquilibrium sstructure.tructure. HHowever,owever, wwhenhen wewe considerconsider Hayek’sHayek’s comment,comment, wewe fi nndd a silversilver lining:lining: a contem-contem- pporaryorary versionversion ofof hishis paragraph,paragraph, whichwhich wouldwould involveinvolve a discussiondiscussion ofof thethe corecore aandnd periphery,periphery, wouldwould confrontconfront oneone modelingmodeling approachapproach againstagainst otherother modelingmodeling aapproaches,pproaches, nnotot modelsmodels againstagainst narrative.narrative. ThisThis isis goodgood news.news. ThereThere isis nono doubtdoubt thatthat tthehe formalizationformalization ofof macroeconomicsmacroeconomics overover recentrecent decadesdecades hashas increasedincreased itsits poten-poten- ttial.ial. WeWe justjust needneed toto bebe carefulcareful toto notnot letlet thisthis formalizationformalization gaingain itsits ownown lifelife andand ddistractistract usus fromfrom thethe ultimateultimate goal,goal, whichwhich isis toto understandunderstand thethe mechanismsmechanisms thatthat drivedrive tthehe rrealeal eeconomy.conomy. TThishis pprogressrogress aalsolso ooffersffers hhopeope tthathat wwee mmayay fi ndnd wwaysays ttoo eexplorexplore fformallyormally aandnd eexplicitlyxplicitly tthehe llimitsimits ooff oourur aandnd eeconomicconomic aagents’gents’ kknowledge.nowledge. TThishis iiss tthehe ssecondecond tthemeheme I ddevelopevelop iinn tthishis ppaper.aper. TThehe iideadea iiss ttoo pplacelace aatt tthehe ccenterenter ooff tthehe aanalysisnalysis tthehe ffactact tthathat tthehe ccomplexityomplexity ooff mmacroeconomicacroeconomic iinteractionsnteractions llimitsimits tthehe kknowledgenowledge wwee ccanan eeverver aattain.ttain. IInn tthinkinghinking aaboutbout aanalyticalnalytical ttoolsools aandnd mmacroeconomicacroeconomic ppolicies,olicies, wwee sshouldhould sseekeek tthosehose tthathat aarere rrobustobust ttoo tthehe eenormousnormous uuncertaintyncertainty ttoo wwhichhich wwee aarere cconfionfi nned,ed, aandnd wwee sshouldhould cconsideronsider wwhathat tthishis ccomplexityomplexity ddoesoes ttoo tthehe aactionsctions aandnd rreac-eac- ttionsions ooff tthehe eeconomicconomic aagentsgents wwhosehose bbehaviorehavior wwee aarere ssupposedupposed ttoo bbee ccapturing.apturing. I ccannotannot bebe suresure thatthat shiftingshifting resourcesresources fromfrom thethe currentcurrent corecore toto thethe peripheryperiphery aandnd ffocusingocusing onon thethe effectseffects ofof (very)(very) limitedlimited knowledgeknowledge onon ourour modelingmodeling strategystrategy aandnd oonn tthehe aactionsctions ooff tthehe eeconomicconomic aagentsgents wwee aarere ssupposedupposed ttoo mmodelodel iiss tthehe bbestest nnextext sstep.tep. However,However, I amam almostalmost certaincertain thatthat ifif thethe goalgoal ofof macroeconomicsmacroeconomics isis toto provideprovide fformalormal fframeworksrameworks ttoo aaddressddress rrealeal eeconomicconomic pproblemsroblems rratherather tthanhan ppurelyurely lliterature-iterature- ddrivenriven ones,ones, wewe betterbetter startstart tryingtrying somethingsomething newnew ratherrather soon.soon. TheThe alternativealternative ofof ssegmenting,egmenting, wwithith aacademiccademic mmacroeconomicsacroeconomics pplayinglaying iitsts iinternalnternal ggamesames aandnd lleavingeaving tthehe rrealeal wworldorld pproblemsroblems mmostlyostly ttoo iinformalnformal ccommentatorsommentators aandnd ““policy”policy” ddiscussions,iscussions, iiss nnotot vveryery aattractivettractive eeither,ither, fforor tthehe llatteratter ooftenften ssufferuffer ffromrom aann eevenven ddeepereeper ppretense-retense- oof-knowledgef-knowledge ssyndromeyndrome tthanhan ddoo aacademiccademic mmacroeconomists.acroeconomists.

CCoreore aandnd PPeripheryeriphery

TThehe uultimateltimate ggoaloal ooff mmacroeconomicsacroeconomics iiss ttoo eexplainxplain aandnd mmodelodel tthehe ((simultaneous)simultaneous) aaggregateggregate ooutcomesutcomes tthathat aariserise ffromrom tthehe ddecisionsecisions mmadeade bbyy mmultipleultiple aandnd hheterogeneouseterogeneous eeconomicconomic aagentsgents iinteractingnteracting tthroughhrough ccomplexomplex rrelationshipselationships aandnd mmarkets.arkets. NNeithereither tthehe ccoreore nnoror tthehe pperipheryeriphery iiss aableble ttoo aaddressddress tthishis iincrediblyncredibly aambitiousmbitious ggoaloal vveryery ssatisfac-atisfac- ttorily.orily. TThehe pperipheryeriphery hhasas ffocusedocused oonn tthehe ddetailsetails ooff tthehe ssubproblemsubproblems aandnd mmechanismsechanisms bbutut hhasas ddownplayedownplayed ddistantistant aandnd ccomplexomplex ggeneraleneral eequilibriumquilibrium iinteractions.nteractions. TThehe ccoreore hhasas ffocusedocused oonn ((extremelyextremely sstylized)tylized) vversionsersions ooff tthehe generalgeneral equilibriumequilibrium interactionsinteractions aandnd hhasas ddownplayedownplayed tthehe ssubproblems.ubproblems. TThehe naturalnatural nextnext stepstep forfor thethe core,core, manymany wouldwould argue,argue, isis toto addadd thethe insightsinsights ofof tthehe peripheryperiphery ggraduallyradually intointo itsits dynamicdynamic stochasticstochastic generalgeneral equilibriumequilibrium structure.structure. 88 Journal of Economic Perspectives

I amam mmuchuch llessess optimisticoptimistic aboutabout thisthis strategy,strategy, asas I thinkthink itit isis plaguedplagued byby internalinternal iinconsistenciesnconsistencies aandnd ppretense-of-knowledgeretense-of-knowledge pproblems.roblems.

TThehe PPeripheryeriphery I believebelieve tthathat uupp ttoo nnowow tthehe iinsight-buildingnsight-building mmodeode ((bothboth ppastast aandnd ppresent)resent) ooff tthehe pperipheryeriphery ooff mmacroeconomicsacroeconomics hhasas pprovenroven ttoo bbee mmoreore uusefulseful tthanhan tthehe mmacro-machine-acro-machine- bbuildinguilding modemode ooff thethe ccoreore ttoo helphelp oourur uunderstandingnderstanding ooff ssignifiignifi cantcant macroeconomicmacroeconomic eevents.vents. ForFor eexample,xample, iinn tthehe ccontextontext ooff tthehe ccurrenturrent fi nancialnancial andand eeconomicconomic ccrisis,risis, tthehe pperipheryeriphery ggaveave uuss fframeworksrameworks ttoo uunderstandnderstand pphenomenahenomena ssuchuch aass sspeculativepeculative bbubbles,ubbles, lleverageeverage cycles,cycles, fi rere sales,sales, fl ightight toto quality,quality, margin-margin- andand collateral-constraintcollateral-constraint spirals,spirals, lliquidityiquidity runs,runs, andand soso on—phenomenaon—phenomena tthathat pplayedlayed a ccentralentral rroleole iinn bbringingringing tthehe wworldorld eeconomyconomy ttoo tthehe bbrinkrink ooff a ssevereevere ddepression.epression. TThishis lliteratureiterature aalsolso pprovidedrovided tthehe bbasisasis fforor tthehe policypolicy frameworkframework thatthat waswas usedused toto containcontain thethe crisis.crisis. AAllll inin all,all, I believebelieve itit wouldwould bbee ggoodood forfor macroeconomicsmacroeconomics ttoo ((re)orientre)orient a llargerarger ssharehare ooff iitsts hhumanuman ccapitalapital iinn tthishis ddirection,irection, nnotot jjustust fforor tthehe sstudytudy ooff ccrisesrises bbutut aalsolso fforor iitsts bbroaderroader cconcerns.oncerns. YYetet thethe peripheryperiphery ofof macroeconomicsmacroeconomics isis defidefi nedned notnot onlyonly byby itsits subjects,subjects, butbut aalso,lso, andand perhapsperhaps eevenven mmoreore sso,o, bbyy a mmethodologicalethodological ddecisionecision wwhichhich mmakesakes iitsts ggoalsoals nnarrowerarrower thanthan thosethose ofof thethe core.core. TheThe methodologymethodology ofof thethe peripheryperiphery isis designeddesigned ttoo isolateisolate insightsinsights (as(as micro-theorymicro-theory does),does), andand researchresearch onon thesethese topicstopics doesdoes notnot ttypicallyypically ddisplayisplay tthehe aaspirationspiration ttoo pproviderovide ccomprehensiveomprehensive aanswers—letnswers—let aalonelone qquan-uan- ttitativeitative answers—toanswers—to thethe overalloverall effectseffects onon thethe macroeconomy.macroeconomy. ItIt isis onlyonly naturalnatural forfor mmacroeconomistsacroeconomists toto wantwant more,more, butbut itit isis thethe rushedrushed processprocess toto fulfifulfi llll thisthis ambitionambition tthathat I bbelieveelieve hhasas lleded tthehe ccoreore rrightight iintonto HHayek’sayek’s pretense-of-knowledgepretense-of-knowledge syndrome.syndrome.

TThehe CCoreore TThehe corecore approachapproach toto macroeconomics,macroeconomics, asas itit isis taughttaught inin mostmost graduategraduate pprogramsrograms andand asas itit appearsappears inin leadingleading journals,journals, beginsbegins withwith a neoclassicalneoclassical growthgrowth mmodel.odel. TThishis modelmodel isis thenthen developeddeveloped intointo a stochasticstochastic form.form. TheThe earlyearly versionsversions wwereere calledcalled “real”“real” businessbusiness cycles,cycles, becausebecause thethe keykey shocksshocks occurredoccurred toto technology.technology. IInn tthehe bbasicasic rrealeal bbusinessusiness ccycleycle aapproach,pproach, hhouseholdsouseholds mmakeake ooptimizingptimizing ddecisionsecisions iinn eequatingquating theirtheir marginalmarginal raterate ofof substitutionsubstitution betweenbetween consumptionconsumption andand leisureleisure toto tthehe rrealeal wage,wage, whichwhich inin thethe basicbasic modelmodel isis determineddetermined byby thethe marginalmarginal productproduct ofof llabor.abor. HouseholdsHouseholds alsoalso makemake optimizingoptimizing decisionsdecisions inin thethe choicechoice betweenbetween consump-consump- ttionion andand saving,saving, wherewhere inin thisthis casecase thethe maximizingmaximizing cconditionondition involvesinvolves settingsetting thethe hhousehold’sousehold’s mmarginalarginal rrateate ooff substitutionsubstitution betweenbetween presentpresent andand futurefuture consumptionconsumption eequalqual ttoo tthehe rrateate ooff rreturn,eturn, whichwhich inin thethe basicbasic modelmodel isis determineddetermined byby thethe raterate ooff returnreturn thatthat fi rmsrms receivereceive onon investment.investment. FirmsFirms optimizeoptimize theirtheir useuse ofof laborlabor andand ccapitalapital accordingaccording toto a productionproduction function.function. TheThe standardstandard approachapproach inin macroeco-macroeco- nnomicsomics iiss tthenhen toto addadd toto thisthis corecore modelmodel a fewfew ingredients.ingredients. ForFor example,example, inin thisthis jjournal,ournal, GGalíalí aandnd GGertlerertler ((2007)2007) bbuilduild uupp a mmodelodel ooff tthishis ssortort aandnd tthenhen aadddd mmoney,oney, mmonopolisticonopolistic competitioncompetition (and(and priceprice mark-ups),mark-ups), andand nominalnominal priceprice rigidities.rigidities. Vari-Vari- aantsnts ooff tthishis mmodelodel havehave becomebecome thethe workhorseworkhorse modelmodel inin researchresearch departmentsdepartments ofof ccentralentral banks.banks. InIn thisthis symposium,symposium, thethe paperspapers byby OhanianOhanian andand HallHall alsoalso taketake thisthis ggeneraleneral approachapproach ofof startingstarting withwith a realreal businessbusiness cyclecycle modelmodel andand thenthen discussingdiscussing hhowow iitt mmightight bbee aadapteddapted ttoo ccaptureapture tthehe kkeyey eelementslements ooff tthehe fi nancialnancial crisis.crisis. Macroeconomics after the Crisis: The Pretense-of-Knowledge Syndrome 89

IIff wwee wwereere ttoo sstoptop tthere,here, aandnd ssimplyimply uusese tthesehese sstylizedtylized sstructurestructures aass jjustust oonene mmoreore ttoolool toto understandunderstand a piecepiece ofof thethe complexcomplex problem,problem, andand toto exploreexplore somesome potentiallypotentially pperverseerverse ggeneraleneral equilibriumequilibrium effecteffect wwhichhich couldcould aaffectffect thethe iinsightsnsights iisolatedsolated inin tthehe pperiphery,eriphery, tthenhen I wwouldould bebe fi nene withwith it.it. MyMy problemsproblems startstart whenwhen thesethese structuresstructures aarere ggiveniven llifeife oonn ttheirheir oown,wn, aandnd rresearchersesearchers cchoosehoose ttoo ““taketake tthehe mmodelodel sseriously”eriously” ((aa statementstatement thatthat signalssignals thethe timetime toto leaveleave a seminar,seminar, forfor itit isis alwaysalways followedfollowed byby a ssequenceequence ooff nnaiveaive aandnd ssurrealurreal claims).claims). TThehe qquantitativeuantitative iimplicationsmplications ooff tthishis ccoreore aapproach,pproach, wwhichhich aarere bbuiltuilt oonn ssuppos-uppos- eedlydly ““micro-founded”micro-founded” calibrationscalibrations ofof keykey pparameters,arameters, aarere ddefiefi nitelynitely onon thethe surrealsurreal sside.ide. TakeTake forfor exampleexample thethe preferredpreferred “microfoundation”“microfoundation” ofof thethe supplysupply ofof capitalcapital iinn tthehe workhorseworkhorse modelsmodels ofof thethe corecore approach.approach. A keykey parameterparameter toto calibratecalibrate inin tthesehese modelsmodels isis tthehe intertemporalintertemporal substitutionsubstitution elasticityelasticity ofof a representativerepresentative agent,agent, wwhichhich isis toto bebe estimatedestimated fromfrom micro-data.micro-data. A wholewhole literatureliterature developsdevelops aroundaround thisthis eestimation,stimation, whichwhich narrowsnarrows thethe parameterparameter toto certaincertain values,values, whichwhich areare thenthen toto bebe uusedsed aandnd hhonoredonored bbyy aanyonenyone wwantinganting ttoo ssayay ssomethingomething aaboutbout ““modern”modern” mmacroeco-acroeco- nnomics.omics. TThishis pparameterarameter maymay bbee a rreasonableeasonable estimateestimate forfor anan individualindividual agentagent facingfacing a specifispecifi c micromicro decision,decision, butbut whatwhat doesdoes itit havehave toto dodo withwith thethe aggregate?aggregate? WhatWhat hhappenedappened withwith thethe rolerole ofof ChineseChinese bureaucrats,bureaucrats, GulfGulf autocrats,autocrats, andand thethe like,like, inin thethe ssupplyupply ofof capital?capital? A typicaltypical answeranswer isis notnot toto worryworry aboutabout it,it, becausebecause thisthis isis allall “as“as if.”if.” BButut tthen,hen, wwhyhy ddoo wwee ccallall thisthis strategystrategy microfoundationsmicrofoundations rratherather tthanhan rreduced-form?educed-form? MMyy ppointoint iiss tthathat bbyy ssomeome sstrangetrange hherdingerding processprocess thethe corecore ofof macroeconomicsmacroeconomics sseemseems ttoo ttransformransform tthingshings tthathat mmayay hhaveave bbeeneen uusefulseful mmodelingodeling sshort-cutshort-cuts iintonto a ppartart ooff a nnewew andand artifiartifi cialcial “reality,”“reality,” andand nownow suddenlysuddenly everyoneeveryone usesuses thethe samesame language,language, wwhichhich inin thethe nextnext iterationiteration getsgets confusedconfused with,with, andand eventuallyeventually replaces,replaces, reality.reality. AAlonglong thethe way,way, thisthis processprocess ofof make-believemake-believe substitutionsubstitution raisesraises ourour presumptionpresumption ofof kknowledgenowledge aboutabout thethe workingsworkings ofof a complexcomplex economyeconomy andand increasesincreases thethe risksrisks ofof a ““pretensepretense ooff kknowledge”nowledge” aaboutbout wwhichhich HHayekayek wwarnedarned uus.s. AAfterfter muchmuch trialtrial andand error,error, thesethese corecore modelsmodels havehave managedmanaged toto generategenerate rreasonableeasonable nnumbersumbers fforor qquantitiesuantities dduringuring pplain-vanilla,lain-vanilla, ssecond-orderecond-order bbusinessusiness ccycleycle fl uctuations.uctuations. However,However, thethe structuralstructural interpretationinterpretation attributedattributed toto thesethese resultsresults isis ooftenften nnaïveaïve aatt bbest,est, aandnd mmoreore ooftenften iiss wworseorse tthanhan tthat.hat. FForor eexample,xample, wwhilehile tthesehese mmodelsodels hhaveave bbeeneen ssuccessfuluccessful iinn mmatchingatching ssomeome aaggregateggregate qquantities,uantities, ttheyhey hhaveave ddoneone mmuchuch moremore poorlypoorly onon prices.prices. ButBut inin whatwhat sensesense isis itit a goodgood generalgeneral equilibriumequilibrium fi t iiff tthehe qquantitiesuantities aarere rrightight bbutut nnotot tthehe pprices?rices? IIncidentally,ncidentally, thisthis processprocess ofof selectiveselective measuresmeasures ofof successsuccess alsoalso weakensweakens thethe iinitialnitial motivationmotivation forfor bbuildinguilding tthehe mmicrofoundationsicrofoundations ooff mmacroeconomics,acroeconomics, wwhichhich iiss ttoo mmakeake tthehe ttheoryheory ttestable.estable. A ttheoryheory iiss nnoo llongeronger ttestableestable wwhenhen rrejectionejection iiss uusedsed nnotot ttoo discarddiscard thethe theory,theory, butbut toto selectselect thethe datadata momentsmoments underunder whichwhich thethe corecore modelmodel iiss toto bebe judged.judged. ThisThis practicepractice meansmeans thatthat well-knownwell-known majormajor failuresfailures justjust becomebecome ““puzzles,”puzzles,” whichwhich aarere ssoonoon ppresumedresumed ttoo bbee oorthogonalrthogonal ttoo tthehe ooutpututput ffromrom tthehe qquan-uan- ttitativeitative mmodelodel tthathat iiss ttoo bbee ttakenaken ““seriously.”seriously.”2

2 In a similar spirit, but applied to ad-hoc dynamics models, Lucas (2003) writes: “There’s an interesting footnote in Patinkin’s book. Milton Friedman had told him that the rate of change of price in any one market ought to depend on excess demand and supply in all markets in the system. Patinkin is happy 90 Journal of Economic Perspectives

BButut isn’tisn’t abstractionabstraction whatwhat goodgood economiceconomic modelsmodels areare about,about, forfor onlyonly thenthen cancan wwee isolateisolate thethe essenceessence ofof ourour concerns?concerns? Yes,Yes, butbut withwith certaincertain requirementsrequirements toto whichwhich I tthinkhink tthehe ccoreore hhasas ffailedailed toto adhere,adhere, yetyet thethe peripheryperiphery hashas gottengotten mostlymostly rright.ight. TThehe pperipheryeriphery uusesses aabstractionbstraction ttoo rremoveemove tthehe iinessential,nessential, butbut a ttypicalypical peripheryperiphery ppaperaper isis veryvery carefulcareful thatthat thethe mainmain objectobject ofof studystudy isis anchoredanchored byby sensiblesensible assump-assump- ttions.ions. ItIt isis fi nene toto bebe asas “goofy”“goofy” asas neededneeded toto makemake thingsthings simplersimpler alongalong inessentialinessential ddimensions,imensions, butbut itit isis iimportantmportant notnot toto soundsound “funny”“funny” onon thethe specifispecifi c issueissue thatthat isis toto bbee aaddressed.ddressed. IInstead,nstead, corecore mmacroeconomicsacroeconomics ooftenften hhasas aaimedimed nnotot fforor a rrealisticealistic aanchornchor aandnd a ssimplifiimplifi cationcation ofof thethe rest,rest, butbut forfor beingbeing onlyonly half-“goofy”half-“goofy” onon everything:everything: preferencespreferences aandnd pproductionroduction functionsfunctions thatthat dodo notnot representrepresent anyoneanyone butbut thatthat couldcould bebe foundfound iinn anan introductoryintroductory microeconomicsmicroeconomics textbook,textbook, thethe samesame forfor markets,markets, andand soso on.on. ByBy nnow,ow, ttherehere areare a wholewhole setset ofof conventionsconventions andand magicmagic parameterparameter valuesvalues resultingresulting iinn anan artifiartifi cialcial worldworld thatthat cancan bebe analyzedanalyzed withwith thethe rigorrigor ofof micro-theorymicro-theory butbut thatthat sspeakspeaks ooff nnoo pparticulararticular rreal-worldeal-world iissuessue wwithith anyany reliability.reliability.

IIntegration?ntegration? OOnene ppossibleossible rreactioneaction ttoo mmyy rremarksemarks iiss tthathat I aamm ttoooo iimpatient;mpatient; tthathat wwithith eenoughnough ttime,ime, wwee wwillill aarriverrive aatt aann EEll DDoradoorado ooff mmacroeconomicsacroeconomics wwherehere tthehe kkeyey iinsightsnsights ofof thethe peripheryperiphery areare incorporatedincorporated intointo a massivemassive dynamicdynamic stochasticstochastic ggeneraleneral equilibriumequilibrium model.model. AfterAfter all,all, therethere hashas beenbeen anan enormousenormous collectivecollective efforteffort iinn rrecentecent ddecadesecades iinn bbuildinguilding ssuchuch mmodels,odels, wwithith aann iincreasingncreasing nnumberumber ooff bbellsells aandnd whistleswhistles rrepresentingepresenting vvariousarious mmicroeconomicicroeconomic ffrictions.rictions. TThehe rresearchesearch ddepart-epart- mmentsents ofof centralcentral banksbanks aroundaround thethe worldworld havehave becomebecome eveneven moremore obsessedobsessed thanthan aacademicscademics wwithith tthishis aagenda.genda. HHowever,owever, I tthinkhink thisthis incrementalincremental strategystrategy maymay wellwell havehave overshotovershot itsits peakpeak andand mmayay lleadead uuss toto a minimumminimum ratherrather thanthan a maximummaximum inin termsterms ofof capturingcapturing realisticrealistic mmacroeconomicacroeconomic phenomena.phenomena. WeWe areare diggingdigging ourselves,ourselves, oneone stepstep atat a time,time, deeperdeeper aandnd deeperdeeper intointo a Fantasyland,Fantasyland, withwith economiceconomic agentsagents whowho cancan solvesolve richerricher andand rrichericher stochasticstochastic generalgeneral equilibriumequilibrium problemsproblems containingcontaining allall sortssorts ofof frictions.frictions. BBecauseecause thethe “progress”“progress” isis gradual,gradual, wewe dodo notnot seemseem toto noticenotice asas wewe acceptaccept whatwhat areare iincreasinglyncreasingly absurdabsurd behavioralbehavioral conventionsconventions andand stretchstretch thethe intelligenceintelligence andand infor-infor- mmationation ooff uunderlyingnderlying eeconomicconomic aagentsgents ttoo llevelsevels tthathat rrenderender tthemhem uunrecognizable.nrecognizable. TThehe beautybeauty ofof thethe simplestsimplest bbarebonesarebones realreal businessbusiness cyclecycle modelmodel is,is, iinn ffact,act, iinn iitsts ssimplicity.implicity. IItt iiss a ccoherentoherent descriptiondescription ofof equilibriumequilibrium inin a frictionlessfrictionless world,world, wherewhere itit iiss reasonablereasonable toto expectexpect thatthat humanshumans cancan dealdeal withwith itsits simplicity.simplicity. I wouldwould ratherrather stopstop ttherehere ((perhapsperhaps wwithith sspacepace fforor aaddingdding oonene nnominalominal rrigidity)igidity) aandnd ssimplyimply aacknowledgecknowledge tthathat itit isis a benchmark,benchmark, notnot a shellshell oror a steppingstonesteppingstone forfor everythingeverything wewe studystudy inin mmacroeconomics,acroeconomics, wwhichhich iiss uunfortunatelynfortunately tthehe wwayay tthehe ccoreore ttreatsreats iitt ttoday.oday.3 about this suggestion because he loves more generality, but if you think about Friedman’s review of Lange, of Lange’s book, what Friedman must have been trying to tell Patinkin is that he thinks the theory is empty, that anything can happen in this model. And I think he’s got a point.” 3 An extreme form of this view of the real business cycle model as the steppingstone for everything else is the so-called “gap approach,” which essentially views and constrains the research agenda of macro- economics to studying the failures of the maximization conditions of this very particular model (Chari, Ricardo J. Caballero 91

SSinceince thethe peripheryperiphery isis aboutabout isolatingisolating specifispecifi c mechanisms,mechanisms, itit surroundssurrounds thethe ssourcesources ooff tthesehese mmechanismsechanisms wwithith aassumptionsssumptions ddesignedesigned ttoo kkillill uunwantednwanted eeffectsffects tthathat wwouldould ppolluteollute tthehe mmessage.essage. IItt mmightight sseemeem aass iiff tthehe nnaturalatural pprocessrocess ttoo bbuilduild a qquantita-uantita- ttiveive aanswernswer fforor tthehe wwholehole wwouldould sstarttart wwithith bbringingringing bbackack ssomeome ooff tthehe rrealisticealistic uunwantednwanted eeffectsffects thatthat werewere removedremoved fforor aanalyticnalytic cconvenience,onvenience, aandnd ttoo mmodelodel tthehe iinteractionsnteractions aandnd complexitiescomplexities tthathat aariserise ffromrom tthehe ssimultaneousimultaneous ppresenceresence ooff aallll tthesehese pparts.arts. BButut iinstead,nstead, tthehe ccurrenturrent ccoreore aapproachpproach ooff mmacroeconomicsacroeconomics ppreservesreserves mmanyany ooff tthehe ooriginalriginal cconvenience-assumptionsonvenience-assumptions ffromrom tthehe rresearchesearch oonn tthehe pperipheryeriphery aandnd tthenhen oobsessesbsesses wwithith ““closing”closing” tthehe mmodelodel bbyy aaddingdding artifi cial ffactoractor ssupplyupply cconstraintsonstraints ((notenote tthathat tthehe eemphasismphasis iiss oonn tthehe wwordord aartifirtifi ccial,ial, nnotot oonn tthehe wwordord cconstraints).onstraints). AAllll tthathat wwee llearnearn ffromrom tthishis eexercisexercise iiss wwhathat tthesehese aartifirtifi ccialial cconstraintsonstraints ddoo ttoo tthehe sstylizedtylized mmechanisms,echanisms, nnotot wwhathat tthesehese mmechanismsechanisms ccanan ddoo ttoo tthehe wwholehole iinn a rrealisticealistic ssetting,etting, wwhichhich sshouldhould bbee oourur ggoal.oal.4 WWee nneedeed ttoo sstoptop tthishis ppractice,ractice, aatt lleasteast aass a nnorm,orm, eevenven iiff tthehe ccostost iiss tthathat wwee ccan’tan’t mmakeake wwelfareelfare sstatementstatements wwithith tthehe ssameame ddegreeegree ooff cconfionfi ddenceence tthathat a ffullyully sstructuraltructural mmodelodel wwouldould aallowllow uuss ttoo ddo—ito—it wwouldould bbee a ffalsealse ppretenseretense ooff kknowledgenowledge aanyway.nyway. MMoreover,oreover, tthehe pprocessrocess ofof bringingbringing thethe peripheryperiphery intointo thethe corecore raisesraises anan obviousobvious ttensionension aaboutbout thethe rolerole ofof rrationalational expectations.expectations. RationalRational expectationsexpectations isis a centralcentral iingredientngredient ofof thethe currentcurrent core;core; however,however, thisthis assumptionassumption becomesbecomes increasinglyincreasingly uuntenablentenable asas wewe continuecontinue toto addadd thethe realismrealism ofof thethe peripheryperiphery intointo thethe core.core.5 WhileWhile iitt ooftenften makesmakes sensesense toto assumeassume rationalrational expectationsexpectations forfor a limitedlimited applicationapplication toto iisolatesolate a particularparticular mechanismmechanism thatthat isis distinctdistinct fromfrom thethe rolerole ofof expectationsexpectations forma-forma- ttion,ion, thisthis assumptionassumption nono longerlonger makesmakes sensesense onceonce wewe assembleassemble thethe wholewhole model.model. AAgentsgents ccouldould bbee ffullyully rrationalational wwithith rrespectespect ttoo ttheirheir llocalocal eenvironmentsnvironments aandnd eeverydayveryday aactivities,ctivities, butbut theythey areare mostmost pprobablyrobably nearlynearly cluelessclueless withwith respectrespect toto thethe statisticsstatistics aaboutbout whichwhich currentcurrent macroeconomicmacroeconomic modelsmodels expectexpect themthem toto havehave fullfull informationinformation aandnd rrationalational iinformation.nformation. TThishis issueissue isis nnotot oneone thatthat cancan bebe addressedaddressed byby addingadding a parameterparameter capturingcapturing a llittleittle bbitit mmoreore rriskisk aaversionversion aboutabout macroeconomic,macroeconomic, ratherrather thanthan local,local, phenomena.phenomena. TThehe rreactioneaction ofof humanhuman bbeingseings ttoo tthehe ttrulyruly uunknownnknown iiss ffundamentallyundamentally differentdifferent ffromrom tthehe wwayay ttheyhey ddealeal wwithith tthehe rrisksisks aassociatedssociated wwithith a kknownnown ssituationituation aandnd eenviron-nviron- mmentent ((Knight,Knight, 11921;921; EEllsberg,llsberg, 1961).1961). InIn realistic,realistic, real-timereal-time settings,settings, bothboth economiceconomic aagentsgents andand researchersresearchers havehave a veryvery limitedlimited understandingunderstanding ofof thethe mechanismsmechanisms atat wwork.ork. ThisThis isis anan order-of-magnitudeorder-of-magnitude lessless knowledgeknowledge thanthan ourour corecore macroeconomicmacroeconomic mmodelsodels currentlycurrently assume,assume, andand hencehence itit isis highlyhighly likelylikely thatthat thethe optimaloptimal approxima-approxima- ttionion paradigmparadigm isis quitequite differentdifferent fromfrom ccurrenturrent workhorses,workhorses, bothboth forfor academicacademic andand

Kehoe, and McGrattan, 2007). Again, there is nothing wrong with this approach as a test of the workings of that particular model; the problem arises when it becomes a benchmark for something beyond that specifi c goal. 4 It is not rare to fi nd in the literature that some mechanism is called irrelevant because it is killed by the artifi cial constraints of the core. However, in many instances that can be corroborated by data, such results are really indictments of the artifi cial constraints, not of the mechanisms. 5 Of course, part of the realism added to the core could come from reducing the IQ of its agents. An important line of recent work started by Sims (2003) attempts to acknowledge human beings’ computa- tional capacity constraints by building on Shannon’s (1948) information theory work. However, much of this work has taken place within the context of the current core models of macroeconomics, and hence it still trivializes the environment economic agents have to confront. 92 Journal of Economic Perspectives

ppolicyolicy work.work. InIn tryingtrying toto addadd a degreedegree ofof complexitycomplexity toto thethe currentcurrent corecore models,models, byby bbringingringing inin aspectsaspects ofof thethe periphery,periphery, wewe areare simultaneouslysimultaneously makingmaking thethe rationalityrationality aassumptionsssumptions bbehindehind tthathat ccoreore aapproachpproach llessess pplausible.lausible. MMoreover,oreover, tthishis iintegrationistntegrationist strategystrategy doesdoes notnot comecome withwith anan assuranceassurance thatthat itit wwillill taketake usus ttoo tthehe rrightight pplace,lace, aass ttherehere iiss aann eenormousnormous aamountmount ooff ppathath ddependenceependence iinn tthehe pprocessrocess bbyy wwhichhich eelementslements aarere iincorporatedncorporated intointo thethe ccore;ore; aandnd tthehe bbaggageaggage wwee aarere aalreadylready ccarryingarrying hhasas tthehe ppotentialotential ttoo ddistortistort tthehe sselectionelection ooff tthehe mmechanismsechanisms ooff thethe peripheryperiphery thatthat areare incorporated.incorporated. GivenGiven thethe enormousenormous complexitycomplexity ofof thethe tasktask aatt hand,hand, wewe cancan spendspend anan unacceptablyunacceptably longlong timetime wanderingwandering inin surrealisticsurrealistic worldsworlds bbeforeefore ggainingaining aanyny ttractionraction iintonto rreality.eality. WWee ultimatelyultimately needneed toto revisitrevisit thethe ambitiousambitious goalgoal ofof thethe core,core, ofof havinghaving a frame-frame- wworkork fforor uunderstandingnderstanding tthehe wwhole,hole, ffromrom sshockshocks ttoo ttransmissionransmission cchannels,hannels, aallll ooff tthemhem iinteractingnteracting wwithith eeachach oother.ther. TThehe iissuessue iiss hhowow ttoo ddoo tthishis wwithoutithout oover-trivializingver-trivializing tthehe wworkingsorkings ooff tthehe eeconomyconomy ((inin tthehe ffundamentalundamental ssenseense ooff ooverestimatingverestimating tthehe ppowerower ooff ourour aapproximations)pproximations) ttoo a ddegreeegree tthathat mmakesakes tthehe fframeworkramework uuselessseless aass a ttoolool fforor uunderstandingnderstanding ssignifiignifi cantcant eventsevents andand dangerousdangerous forfor policypolicy guidance.guidance. I ddon’ton’t hhaveave tthehe aanswernswer ttoo tthishis ffundamentalundamental ddilemma,ilemma, bbutut iitt ddoesoes ppointoint iinn tthehe ddirectionirection ooff mmuchuch mmoreore diversifidiversifi ccationation ooff rresearchesearch andand methodologymethodology thanthan wewe currentlycurrently accept.accept. ItIt alsoalso ppointsoints iinn tthehe ddirectionirection ooff eembracing,mbracing, rratherather tthanhan ssweepingweeping uundernder tthehe rrug,ug, tthehe ccomplexityomplexity ooff tthehe mmacroeconomicacroeconomic eenvironment.nvironment. I tturnurn ttoo tthehe llatteratter tthemeheme nnext.ext.

FFacingacing andand EmbracingEmbracing EconomicEconomic ComplexityComplexity

I ssuspectuspect tthathat embracingembracing ratherrather thanthan fi ghtingghting complexitycomplexity andand whatwhat itit doesdoes toto ourour mmodelingodeling wouldwould helphelp uuss mmakeake pprogressrogress iinn uunderstandingnderstanding mmacroeconomicacroeconomic eevents.vents. OOnene ooff tthehe wweaknesseseaknesses ofof thethe corecore stemsstems fromfrom goinggoing tootoo directlydirectly fromfrom statementsstatements aaboutbout individualsindividuals toto statementsstatements aboutabout thethe aaggregate,ggregate, wherewhere thethe mmainain ddifferenceifference bbetweenetween thethe twotwo comescomes fromfrom stylizedstylized aggregateaggregate constraintsconstraints andand trivialtrivial interactions,interactions, rratherather tthanhan ffromrom tthehe rrichnessichness aandnd uunpredictabilitynpredictability ooff tthehe llinkagesinkages aamongmong tthehe pparts.arts. WWee needneed toto spendspend muchmuch moremore timetime modelingmodeling andand understandingunderstanding thethe topologytopology ofof llinkagesinkages aamongmong aagents,gents, mmarkets,arkets, iinstitutions,nstitutions, aandnd ccountries.ountries. BByy embracingembracing complexitycomplexity I dodo notnot meanmean thethe directdirect importationimportation ofof thethe modelsmodels ffromrom tthehe fformalormal ccomplexityomplexity literatureliterature inin thethe physicalphysical sciences,sciences, asas economicseconomics is,is, andand iiss likelylikely toto remain,remain, fundamentallyfundamentally reductionistreductionist (that(that is,is, itit seeksseeks toto understandunderstand thethe bbehaviorehavior ofof thethe wholewhole fromfrom thatthat ofof thethe parts).parts). TheThe nodesnodes ofof economiceconomic modelsmodels areare sspecial,pecial, fforor ttheyhey ccontainontain aagentsgents wwithith ffrontalrontal llobesobes wwhoho ccanan bbothoth sstrategizetrategize aandnd ppanic,anic, aandnd iitt iiss tthesehese ffeatureseatures thatthat introduceintroduce muchmuch ofof thethe unpredictabilityunpredictability inin thethe linkageslinkages I mmentionedentioned eearlier.arlier.6

6 Durlauf (2004) offers a thoughtful survey and discussion of the econophysics literature and its limita- tions for economic policy analysis, precisely because it “does not use models that adequately respect the purposefulness of individual behavior.” This of course is a statement of the current state of affairs in the literature, not of its potential, which is probably substantial. See Bak, Chen, Scheinkman, and Woodford (1993), Sheinkman and Woodford (1994), Arthur, Durlauf, and Lane (1997), Durlauf (1993, 1997), Macroeconomics after the Crisis: The Pretense-of-Knowledge Syndrome 93

HHavingaving saidsaid this,this, somesome ofof thethe motivationsmotivations forfor thethe econophysicseconophysics literatureliterature dodo sstriketrike a cchordhord withwith thethe tasktask aheadahead forfor macroeconomists.macroeconomists. ForFor example,example, AlbertAlbert andand BBarabásiarabási ((2002),2002), iinn aadvocatingdvocating fforor tthehe uusese ooff sstatisticaltatistical mmechanicsechanics ttoolsools fforor ccomplexomplex nnetworks,etworks, wwrite:rite:

Physics, a major benefi ciary of reductionism, has developed an arsenal of successful tools for predicting the behavior of a system as a whole from the properties of its constituents. We now understand how magnetism emerges from the collective behavior of millions of spins . . . The success of these modeling efforts is based on the simplicity of the interactions between the elements: there is no ambiguity as to what interacts with what, and the inter- action strength is uniquely determined by the physical distance. We are at a loss, however, to describe systems for which physical distance is irrelevant or for which there is ambiguity as to whether two components interact . . . there is an increasingly voiced need to move beyond reductionist approaches and try to understand the behavior of the system as a whole. Along this route, understanding the topology of the interactions between the components, i.e., networks, is unavoidable . . .

IInn anyany event,event, I willwill notnot reviewreview thisthis literatureliterature herehere andand insteadinstead willwill focusfocus onon aargumentsrguments aboutabout cumbersomecumbersome llinkagesinkages aandnd aagents’gents’ cconfusiononfusion aaboutbout tthesehese llinkagesinkages tthathat aarere ccloserloser ttoo mmainstreamainstream mmacroeconomics.acroeconomics.

DDominoesominoes aandnd AAvalanchesvalanches AAllenllen aandnd GGale’sale’s ((2000)2000) mmodelodel ooff ffinancialinancial nnetworksetworks aandnd tthehe iinherentnherent ffragilityragility ofof somesome ofof thesethese structuresstructures providedprovided anan earlyearly andand elegantelegant exampleexample ofof hhowow llinkagesinkages ccanan causecause substantialsubstantial iinstabilitynstability wwithith rrespectespect ttoo shocksshocks differentdifferent ffromrom tthosehose tthehe nnetworketwork wwasas ddesignedesigned ttoo hhandle.andle. RRecently,ecently, SShinhin ((2009)2009) sshowshows hhowow ffireire ssalesales ccanan ggreatlyreatly mmagnifyagnify tthehe ddominoomino mmechanismechanism hhighlightedighlighted bbyy AAllenllen aandnd GGaleale ((2000).2000). AAnothernother eexamplexample ooff ppromisingromising rresearchesearch iinn tthishis sstyletyle iiss RRotembergotemberg ((2009),2009), wwhichhich uusesses ggraphraph ttheoryheory ttoo sstudytudy hhowow iinterconnectednessnterconnectedness aaffectsffects ffirms’irms’ aabilitybility ttoo mmakeake uusese ooff anan exogenousexogenous amountamount ooff liquidity.liquidity. RotembergRotemberg sstudiestudies a ssituationituation iinn wwhichhich aallll ffirmsirms aarere ssolvent;olvent; tthathat iis,s, ““thethe ppaymentsayments tthathat aanyny pparticulararticular ffirmirm iiss eexpectedxpected ttoo mmakeake ddoo nnotot eexceedxceed tthehe ppaymentsayments iitt iiss eentitledntitled ttoo rreceive.”eceive.” HHee findsfinds thatthat iinterconnectednessnterconnectedness ccanan eexacerbatexacerbate ““thethe ddifficultiesifficulties tthathat ffirmsirms hhaveave iinn mmeetingeeting ttheirheir oobligationsbligations iinn pperiodseriods wwherehere lliquidityiquidity iiss mmoreore ddifficultifficult ttoo oobtain.”btain.” OOff course,course, thethe ccomplex-systemsomplex-systems lliteratureiterature itselfitself offersoffers ffascinatingascinating examplesexamples ooff tthehe ppowerower ofof interconnectedness.interconnectedness. Bak,Bak, Chen,Chen, Scheinkman,Scheinkman, andand WoodfordWoodford (1992)(1992) aandnd SSheinkmanheinkman andand WoodfordWoodford (1994)(1994) bringbring methodsmethods andand metaphorsmetaphors fromfrom statis-statis- tticalical mmechanicsechanics ttoo mmacroeconomics.acroeconomics. TTheyhey aarguergue tthathat llocal,ocal, nnonlinearonlinear iinteractionsnteractions

Brock (1993), Brock and Durlauf (2001), Acemoglu, Ozdaglar, and Tahbaz-Zalehi (2010) for early steps in this direction. 94 Journal of Economic Perspectives

ccanan aallowllow ssmallmall iidiosyncraticdiosyncratic sshockshocks ttoo ggenerateenerate llargearge aaggregateggregate fl uctuations,uctuations, ratherrather tthanhan washingwashing outout viavia thethe lawlaw ofof largelarge numbers.numbers. TheyThey discussdiscuss a kindkind ofof macroeco-macroeco- nnomicomic instabilityinstability calledcalled “self-organized“self-organized criticality,”criticality,” comparingcomparing thethe economyeconomy toto a ssandand hill:hill: atat fi rst,rst, a tinytiny graingrain ofof sandsand droppeddropped onon thethe hillhill causescauses nono aggregateaggregate effect,effect, bbutut aass tthehe sslopelope ofof thethe hillhill increases,increases, eventuallyeventually oneone graingrain ofof sandsand cancan bebe suffisuffi cientcient ttoo ccauseause aann aavalanche.valanche. InIn thethe limit,limit, aaggregateggregate fl uctuationsuctuations maymay emergeemerge fromfrom hard-hard- tto-detecto-detect aandnd ppurelyurely iidiosyncraticdiosyncratic shocks.shocks.

PPanicsanics IInn a complexcomplex environment,environment, agentsagents needneed toto makemake decisionsdecisions basedbased onon informa-informa- ttionion tthathat iiss aastonishinglystonishingly llimitedimited rrelativeelative ttoo aallll tthehe tthingshings tthathat aarere ggoingoing oonn aandnd tthathat hhaveave tthehe ppotentialotential ttoo ppercolateercolate tthroughhrough tthehe ssystem.ystem. TThishis ddegreeegree ooff iignorancegnorance iiss nnotot ssomethingomething agentsagents withwith frontalfrontal lobeslobes likelike toto face.face. AnxietyAnxiety isis alsoalso partpart ofof thethe frontalfrontal llobeobe ddomain!omain! RReactionseactions tthathat iincludenclude aanxietynxiety aandnd eevenven ppanicanic aarere kkeyey iingredientsngredients fforor mmacroeconomicacroeconomic ccrises.rises. PPutut differently,differently, a ccomplexomplex eenvironmentnvironment hhasas aann eenormousnormous potentialpotential toto generategenerate ttrulyruly confusingconfusing surprises.surprises. ThisThis factfact ofof lifelife needsneeds toto bebe mademade anan integralintegral partpart ofof mmacroeconomicacroeconomic modelingmodeling andand ppolicymaking.olicymaking. RealityReality isis iimmenselymmensely mmoreore ccomplexomplex tthanhan models,models, withwith millionsmillions ofof potentialpotential weakweak links.links. AfterAfter a crisiscrisis hashas occurred,occurred, iitt iiss rrelativelyelatively easyeasy toto highlighthighlight thethe linklink thatthat blewblew up,up, butbut beforebefore thethe crisis,crisis, itit isis a ddifferentifferent matter.matter. AllAll marketmarket participantsparticipants andand policymakerspolicymakers knowknow theirtheir ownown locallocal wworld,orld, bbutut uunderstandingnderstanding allall thethe ppossibleossible llinkagesinkages aacrosscross thesethese ddifferentifferent wworldsorlds iiss ttoooo complex.complex. TheThe extentextent toto whichwhich thethe lacklack ofof understandingunderstanding ofof thethe fullfull networknetwork mmattersatters ttoo eeconomicconomic agentsagents variesvaries overover thethe cycle.cycle. TheThe importanceimportance ofof thisthis lacklack ofof uunderstandingnderstanding isis atat itsits mostmost eextremextreme llevelevel dduringuring fi nancialnancial crises,crises, whenwhen seeminglyseemingly iirrelevantrrelevant andand distantdistant linkageslinkages areare perceivedperceived toto bebe relevant.relevant. Moreover,Moreover, thisthis changechange iinn paradigm,paradigm, fromfrom iirrelevantrrelevant toto criticalcritical linkages,linkages, cancan triggertrigger massivemassive uncertainty,uncertainty, wwhichhich ccanan uunleashnleash ddestructiveestructive fl ightsights toto quality.quality. BBenoitenoit Mandelbrot,Mandelbrot, thethe mathematicianmathematician perhapsperhaps bestbest knownknown forfor hishis workwork onon ffractalractal geometry,geometry, ooncence drewdrew a parallelparallel fromfrom economicseconomics toto storms,storms, whichwhich cancan onlyonly bbee predictedpredicted afterafter theythey form.form. MandelbrotMandelbrot (2008,(2008, inin a PBSPBS NewsHourNewsHour interviewinterview wwithith PaulPaul SolmanSolman onon OctoberOctober 21,21, 2008)2008) said:said: “[T]he“[T]he basisbasis ofof weatherweather forecastingforecasting iiss lookinglooking fromfrom a satellitesatellite andand seeingseeing a stormstorm coming,coming, butbut notnot predictingpredicting thatthat thethe sstormtorm wwillill form.form. TThehe bbehaviorehavior ofof economiceconomic phenomenaphenomena isis farfar moremore complicatedcomplicated tthanhan tthehe bbehaviorehavior ooff lliquidsiquids oror gases.”gases.” FFinancialinancial ccrisesrises rrepresentepresent aann eextremextreme mmanifestationanifestation ooff ccomplexityomplexity iinn mmac-ac- rroeconomics,oeconomics, butbut thisthis elementelement probablyprobably permeatespermeates thethe entireentire businessbusiness cycle,cycle, inin ppartart throughthrough fl uctuationsuctuations inin thethe perceivedperceived probabilityprobability thatthat complexitycomplexity andand itsits cconsequencesonsequences wwillill bbee uunleashednleashed iinn tthehe nnearear ffuture.uture. TThesehese fl uuctuationsctuations ccouldould bbee eendogenousndogenous aandnd aarisingrising ffromrom llocalocal pphenomenahenomena aass hhighlightedighlighted bbyy tthehe fformalormal ccom-om- pplexitylexity literatureliterature iinn pphysicalhysical ssciences,ciences, bbutut iitt aalsolso ccouldould bbee iinn rresponseesponse ttoo a mmoreore cconventionalonventional macroeconomicmacroeconomic sshock,hock, ssuchuch aass anan oiloil oror aggregateaggregate demanddemand shock,shock, eespeciallyspecially onceonce tthesehese iinteractnteract wwithith mmoreore cconventionalonventional fi nancial-accelerator-typenancial-accelerator-type mmechanismsechanisms (like(like tthosehose iinn KKiyotakiiyotaki aandnd MMoore,oore, 11997;997; BBernankeernanke aandnd GGertler,ertler, 11989).989). Ricardo J. Caballero 95

TThehe peripheryperiphery hashas mademade somesome progressprogress onon partsparts ofof thesethese mechanisms.mechanisms. IInn CCaballeroaballero aandnd SSimsekimsek ((2009a,2009a, bb),), wwee ccaptureapture tthehe iideadea ooff a ssuddenudden rriseise iinn ccomplexityomplexity ffollowedollowed bbyy wwidespreadidespread ppanicanic iinn tthehe fi nancialnancial sector.sector. IInn oourur mmodel,odel, bbanksanks nnormallyormally ccollectollect bbasicasic iinformationnformation aaboutbout ttheirheir ddirectirect ttradingrading ppartners,artners, wwhichhich sserveserves toto aassuressure tthemhem ooff tthehe ssoundnessoundness ooff tthesehese rrelationships.elationships. HHowever,owever, wwhenhen aacutecute fi nancialnancial ddistressistress eemergesmerges iinn ppartsarts ooff tthehe fi nancialnancial network,network, itit isis notnot eenoughnough ttoo bebe informedinformed aboutabout thesethese directdirect tradingtrading partners,partners, butbut itit alsoalso becomesbecomes importantimportant fforor thethe banksbanks ttoo llearnearn aaboutbout tthehe hhealthealth ooff tthehe ppartnersartners ooff ttheirheir ttradingrading ppartnersartners ttoo aassessssess thethe chanceschances ofof anan indirectindirect hhit.it. AAss cconditionsonditions ccontinueontinue ttoo ddeteriorate,eteriorate, bbanksanks mmustust llearnearn aaboutbout tthehe hhealthealth ooff tthehe ttradingrading ppartnersartners ooff tthehe ttradingrading ppartnersartners ooff ttheirheir ttradingrading ppartners,artners, aandnd ssoo oon.n. AAtt ssomeome ppoint,oint, tthehe ccostost ooff iinformationnformation ggatheringathering bbecomesecomes ttoooo llargearge aandnd tthehe bbanks,anks, nnowow ffacingacing eenormousnormous uuncertainty,ncertainty, cchoosehoose ttoo wwithdrawithdraw fromfrom loanloan commitmentscommitments aandnd iilliquidlliquid ppositions.ositions. HHaldanealdane ((2009)2009) mmaster-aster- ffullyully capturescaptures tthehe eessencessence ooff tthehe ccounterpartyounterparty uuncertaintyncertainty pproblemroblem tthathat ccanan aariserise iinn a complexcomplex mmodernodern fi nancialnancial network:network: “Knowing“Knowing yyourour uultimateltimate ccounterparty’sounterparty’s rriskisk thenthen becomesbecomes aakinkin ttoo ssolvingolving a hhigh-dimensionigh-dimension SSudokuudoku ppuzzle.”uzzle.” A fl ight-to-ight-to- qqualityuality ensues,ensues, aandnd tthehe fi nancialnancial crisiscrisis sspreads.preads. TTakingaking a vveryery differentdifferent aagencygency approach,approach, aandnd tthinkinghinking aaboutbout tthehe specificsspecifics ooff thethe high-frequencyhigh-frequency reporepo market,market, Dang,Dang, Gorton,Gorton, andand HolmströmHolmström (2009)(2009) showshow hhowow a negativenegative aggregateaggregate shockshock cancan causecause debtdebt toto becomebecome informationinformation sensi-sensi- ttive,ive, impedingimpeding tthehe eefficientfficient ttradingrading ooff aassets.ssets. TTheyhey ppointoint ooutut tthathat a ssecurityecurity tthathat payspays offoff thethe samesame inin allall statesstates ofof thethe worldworld wouldwould bebe trulytruly informationinformation iinsensitive.nsensitive. GGiveniven llimitedimited lliability,iability, ddebtebt iiss thethe securitysecurity thatthat bestbest approximatesapproximates tthishis iinformation-insensitivenformation-insensitive ssecurityecurity iinn tthehe rrealeal wworldorld aandnd pprovidesrovides tthehe lleasteast iincentivencentive fforor tthehe ccreationreation ooff pprivaterivate iinformation.nformation. HHowever,owever, wwhenhen a bbadad sshockhock cconcentratesoncentrates aagents’gents’ bbeliefseliefs oonn sstatestates ooff tthehe wworldorld wwherehere ddebtebt ddoesoes nnotot ppayay ooffff iinn full,full, agentsagents wouldwould generategenerate informationinformation beforebefore trading,trading, whichwhich raisesraises possi-possi- bbilitiesilities ofof adverseadverse selectionselection andand impedesimpedes trade.trade. InIn thisthis model,model, unexpectedlyunexpectedly eenough,nough, oopacitypacity wwouldould rreduceeduce tthehe eextentxtent ooff aadversedverse sselectionelection aandnd tthushus wwouldould eencouragencourage ttrade.rade. IInn CaballeroCaballero andand KrishnamurthyKrishnamurthy (2008a),(2008a), wwee illustrateillustrate withwith a modelmodel andand eexamplesxamples ofof thethe amplifiamplifi ccationation rroleole ooff KKnightiannightian uncertainty,uncertainty, wwhichhich rrefersefers ttoo rriskisk tthathat cannotcannot bbee mmeasuredeasured aandnd tthushus ccannotannot bbee hhedged.edged. WWee ppointedointed ooutut tthathat mmostost fl ight-to-qualityight-to-quality episodesepisodes aarere triggeredtriggered byby unusualunusual oror unexpectedunexpected events.events. InIn 1970,1970, tthehe defaultdefault byby PPennenn CCentralentral RRailroad’sailroad’s pprime-ratedrime-rated ccommercialommercial ppaperaper ccaughtaught tthehe marketmarket byby ssurprise.urprise. IInn OOctoberctober 11987,987, tthehe sspeedpeed ooff tthehe sstocktock mmarket’sarket’s ddeclineecline lleded investorsinvestors toto questionquestion theirtheir models.models. IInn tthehe ffallall ofof 1998,1998, thethe co-movementco-movement ooff RRussian,ussian, Brazilian,Brazilian, andand U.S.U.S. bondbond spreadsspreads surprisedsurprised eveneven sophisticatedsophisticated mmarketarket pparticipants.articipants. IInn thethe recentrecent fi nancialnancial crisis,crisis, anotheranother defaultdefault onon commercialcommercial paper—paper— tthishis time,time, byby LehmanLehman Brothers—createdBrothers—created ttremendousremendous uuncertainty.ncertainty. TThehe LLehmanehman bbankruptcyankruptcy alsoalso causedcaused profoundprofound disruptiondisruption iinn thethe marketsmarkets forfor creditcredit defaultdefault sswapswaps andand iinterbanknterbank lloans.oans. TThehe ccommonommon aaspectsspects ooff iinvestornvestor bbehaviorehavior aacrosscross tthesehese eepisodes—re-evaluationpisodes—re-evaluation ooff mmodels,odels, cconservatism,onservatism, aandnd ddisengagementisengagement ffromrom rriskyisky aactivities—indicatectivities—indicate tthathat tthesehese eepisodespisodes iinvolvednvolved KKnightiannightian uuncertaintyncertainty aandnd nnotot mmerelyerely anan increaseincrease inin riskrisk exposure.exposure. TheThe extremeextreme emphasisemphasis onon tailtail outcomesoutcomes andand 96 Journal of Economic Perspectives

wworst-caseorst-case sscenarioscenarios iinn aagents’gents’ ddecisionecision rrulesules ssuggestsuggests aaversionversion ttoo tthishis kkindind ooff uuncertainty.ncertainty.7 NNoveltyovelty aandnd uuncertaintyncertainty pplaylay aann iimportantmportant rroleole iinn tthesehese eepisodes;pisodes; fforor eexample,xample, thethe ccollapseollapse ooff AAmaranthmaranth hhedgeedge ffundund iinn 22006006 ccausedaused llittleittle ddisruptionisruption ttoo fi nnancialancial mmarkets,arkets, wwhereashereas tthehe llossesosses aatt LLong-Termong-Term CCapitalapital MManagementanagement inin 19981998 ccontributedontributed toto worldwideworldwide crisiscrisis despitedespite thethe rescuerescue thatthat waswas ultimatelyultimately organized.organized. SSomeome observersobservers ssimilarlyimilarly aarguedrgued tthathat tthehe ooilil ppricerice sspikespikes ooff tthehe 11970s970s wwereere aassoci-ssoci- aatedted withwith muchmuch wworseorse mmacroeconomicacroeconomic ooutcomesutcomes tthanhan tthosehose iinn tthehe 22000s000s bbecauseecause aagentsgents ccameame ttoo eexpectxpect vvolatilityolatility aandnd hhenceence tthehe rrecentecent sshockshocks wwereere nnotot aass ““earth-earth- sshaking”haking” ((Nordhaus,Nordhaus, 22007).007). HHaldanealdane (2009)(2009) comparescompares tthehe rrecentecent fi nancialnancial crisiscrisis toto thethe SevereSevere AcuteAcute RRespiratoryespiratory SSystemystem ((SARS)SARS) ooutbreakutbreak earlierearlier inin thethe decade.decade. MorbidityMorbidity andand mortalitymortality rratesates ffromrom SSARSARS wwere,ere, ““byby eepidemiologicalpidemiological standards,standards, modest.”modest.” YetYet SARSSARS triggeredtriggered a worldwideworldwide panic,panic, reducingreducing growthgrowth ratesrates acrossacross AsiaAsia byby 1–41–4 percentagepercentage points.points. PParentsarents keptkept theirtheir childrenchildren homehome fromfrom schoolschool inin Toronto,Toronto, andand ChineseChinese restaurantsrestaurants iinn tthehe UUnitednited SStatestates wwereere tthehe ttargetsargets ooff bboycotts.oycotts. FacedFaced withwith KnightianKnightian uncertainty,uncertainty, ppeopleeople cconflonfl atedated thethe possibilitypossibility ofof catastrophecatastrophe withwith catastrophecatastrophe itself.itself.

SSomeome PPolicyolicy IImplicationsmplications ooff a CConfusingonfusing EEnvironmentnvironment TThehe ccentralityentrality ooff ssurprisesurprises iinn fi nancialnancial andand economiceconomic crisescrises seemsseems discouragingdiscouraging ssinceince itit wouldwould seemseem tthathat iitt iiss ddiffiiffi cultcult toto fi ghtght somethingsomething thatthat isis essentiallyessentially impossibleimpossible ttoo ppredict,redict, tthathat kkeepseeps cchanging,hanging, aandnd tthathat iiss nnotot uunderstoodnderstood untiluntil afterafter itit happens.happens. HHowever,owever, somesome iinsightsnsights andand systematicsystematic patternspatterns areare stillstill possible.possible. Certainly,Certainly, iitt rremainsemains uusefulseful ttoo tthinkhink aaboutbout ppoliciesolicies oorr iinstitutionsnstitutions that,that, byby affectingaffecting factorsfactors llikeike capitalcapital requirements,requirements, leverage,leverage, andand maturitymaturity mismatches,mismatches, cancan reducereduce thethe riskrisk ooff crises.crises. ButBut fi nancialnancial andand economiceconomic crisescrises areare likelylikely toto happenhappen anyway,anyway, andand soso iitt wwouldould bebe usefuluseful toto considerconsider inin advanceadvance howhow policypolicy mightmight respond,respond, ratherrather thanthan nneedingeeding ttoo iimprovise.mprovise. FForor example,example, oneone commoncommon patternpattern acrossacross allall episodesepisodes ofof thisthis kindkind isis thatthat thethe cconfusiononfusion triggerstriggers panics,panics, andand panicspanics triggertrigger spikesspikes inin thethe demanddemand forfor explicitexplicit andand iimplicitmplicit iinsurance.nsurance. ThisThis observationobservation immediatelyimmediately hintshints aatt tthehe ccoreore ooff tthehe rrequiredequired ppolicyolicy rresponse:esponse: WhenWhen a llargearge ssystemicystemic ccrisisrisis ooff uuncertaintyncertainty sstrikes,trikes, tthehe ggovernmentovernment

7 In Caballero and Krishnamurthy (2008b), we place the origins of the current crisis in this framework. We argue that perhaps the single largest change in the fi nancial landscape over the last fi ve years was in complex credit products: collateralized debt obligations, collateralized loan obligations, and the like. Market participants had no historical record to measure how these fi nancial structures would behave during a time of stress. These two factors, complexity and lack of history, are the preconditions for rampant uncertainty. When the AAA subprime tranches began to experience losses, investors became uncertain about their investments. Had the uncertainty remained confi ned to subprime mortgage investments, the fi nancial system could have absorbed the losses without too much dislocation. However, investors started to question the valuation of the other credit products—not just mortgages—that had been structured in much the same way as subprime investments. The result was a freezing up across the entire credit market. The policy response to this initial freezing was timid, which kept the stress on the fi nancial system alive until a full blown “sudden fi nancial arrest” episode developed (after Lehman’s demise). See Caballero (2009) for an analogy between sudden cardiac arrest and sudden fi nancial arrest. Macroeconomics after the Crisis: The Pretense-of-Knowledge Syndrome 97

mmustust qquicklyuickly pproviderovide aaccessccess ttoo rreasonablyeasonably ppricedriced bbalance-sheetalance-sheet insuranceinsurance toto fragilefragile aandnd ssystemicallyystemically iimportantmportant iinstitutions.nstitutions. IInn CaballeroCaballero andand KrishnamurthyKrishnamurthy (2008a),(2008a), wewe showedshowed thatthat inin anan episodeepisode ofof KKnightiannightian uncertainty,uncertainty, a governmentgovernment oror centralcentral bankbank concernedconcerned withwith thethe aggregateaggregate wwillill wantwant toto provideprovide insuranceinsurance againstagainst extremeextreme events,events, eveneven ifif itit hashas nono informa-informa- ttionalional advantageadvantage overover thethe privateprivate sector.sector. TheThe reasonreason isis thatthat duringduring a panicpanic ofof thisthis kkind,ind, eacheach individualindividual bankbank andand investorinvestor fearsfears beingbeing inin a situationsituation worseworse thanthan thethe aaverage,verage, anan eeventvent tthathat ccannotannot bebe ttruerue fforor tthehe ccollectiveollective ((individualindividual agentsagents knowknow tthishis aggregateaggregate constraint,constraint, butbut theythey assumeassume theythey willwill bebe onon thethe shortshort endend ofof things).things). BByy providingproviding a broadbroad guarantee,guarantee, thethe governmentgovernment getsgets thethe privateprivate sectorsector toto reactreact mmoreore tthanhan oone-for-onene-for-one withwith thisthis guarantee,guarantee, becausebecause itit alsoalso closescloses thethe gapgap betweenbetween tthehe ttruerue aaverageverage andand thethe averageaverage ofof panic-drivenpanic-driven expectations.expectations. ManyMany ofof thethe actualactual pprogramsrograms iimplementedmplemented dduringuring tthehe ccrisisrisis hhadad eelementslements ooff gguarantees,uarantees, aalthoughlthough ttheyhey couldcould probablyprobably havehave gonegone further.further.8 FForor eexample,xample, RRossoss ((2009)2009) arguesargues convinc-convinc- iinglyngly tthathat tthehe governmentgovernment shouldshould supplysupply impairmentimpairment guaranteesguarantees alongalong thethe lineslines ooff thosethose usedused byby thethe government-sponsoredgovernment-sponsored enterprisesenterprises FannieFannie MaeMae andand FreddieFreddie MMacac ttoo iimprovemprove tthehe lliquidityiquidity ofof thethe poolpool ofof legacylegacy assetsassets inin banksbanks balancebalance sheets.sheets. SSimilarly,imilarly, iinn CCaballeroaballero andand KurlatKurlat (2009),(2009), wewe proposedproposed a policypolicy frameworkframework whichwhich wwouldould notnot onlyonly guaranteeguarantee accessaccess toto insuranceinsurance inin thethe eventevent ofof a panic,panic, butbut itit wouldwould ddoo soso inin a fl exibleexible mannermanner thatthat integratesintegrates thethe rolerole ofof thethe governmentgovernment asas anan insurerinsurer ooff llastast rresortesort wwithith pprivaterivate ssectorector iinformationnformation oonn tthehe ooptimalptimal aallocationllocation ooff ccontingentontingent iinsurance.nsurance.9 ForFor examplesexamples ofof relatedrelated publicpublic guaranteeguarantee programsprograms andand proposals,proposals, seesee CCaballeroaballero ((2009a,2009a, bb);); MMehrlingehrling aandnd MMilneilne ((2008);2008); aandnd MMilneilne ((2009).2009). A rrelatedelated argumentargument isis developeddeveloped byby GeanakoplosGeanakoplos inin a seriesseries ofof papers.papers. GGeanakoploseanakoplos (2003,(2003, 2009)2009) andand GeanakoplosGeanakoplos andand FostelFostel (2008)(2008) highlighthighlight thethe rolerole ooff mmarginsargins iinn a ttheoryheory ooff “leverage“leverage cycles.”cycles.” InIn thesethese models,models, thethe supplysupply andand demanddemand fforor lloansoans ddeterminesetermines notnot onlyonly thethe interestinterest rate,rate, butbut alsoalso equilibriumequilibrium leverage.leverage. GGeanakoploseanakoplos (2010)(2010) suggestssuggests a three-prongedthree-pronged approachapproach forfor governmentgovernment policypolicy inin tthehe aftermathaftermath ofof a leverageleverage cycle.cycle. First,First, thethe governmentgovernment shouldshould addressaddress thethe precipi-precipi- ttatingating causecause ofof thethe crisis:crisis: thethe “scary“scary badbad news”news” andand “massive“massive uncertainty,”uncertainty,” whichwhich iinn tthehe pparticulararticular ccasease ooff tthehe rrecentecent crisis,crisis, aaffectedffected thethe housinghousing market.market. Second,Second, tthehe governmentgovernment shouldshould createcreate a lendinglending facilityfacility toto restorerestore “reasonable”“reasonable” levelslevels ofof lleverage.everage. TThird,hird, tthehe ggovernmentovernment sshouldhould rrestoreestore ““optimistic”optimistic” ccapital,apital, fforor eexample,xample,

8 Insurance programs during the crisis included: a temporary program created by the U.S. Treasury Department to insure money-market funds; nonrecourse funding for the purchase of asset-backed securities through the Term Asset-Backed Securities Loan Facility (TALF); and a temporary increase in deposit insurance from $100,000 to $250,000. A notable example from outside the United States was the U.K. Asset Protection Scheme, which backed over half a trillion pounds in post-haircut assets for two British banks. See Caballero (2009), Madigan (2009), and IMF (2009). 9 Under our proposal, the government would issue tradable insurance credits (TICs) which would be purchased by fi nancial institutions, some of which would have minimum holding requirements. During a systemic crisis, each TIC would entitle its holder to attach a government guarantee to some of its assets. All regulated fi nancial institutions would be allowed to hold and use TICs, and possibly hedge funds, private equity funds, and corporations as well. In principle, TICs could be used as a fl exible and readily available substitute for many of the facilities that were created by the Federal Reserve during the crisis. 98 Journal of Economic Perspectives

tthroughhrough bailoutsbailouts (although(although GeanakoplosGeanakoplos isis moremore sanguinesanguine thanthan I amam aboutabout thethe eeffiffi ccacyacy ooff ““bailoutsbailouts wwithith ppunishment,”unishment,” aass ddiscussediscussed iinn CCaballero,aballero, 22010).010). FForor thethe purposepurpose ofof thisthis essay,essay, moremore importantimportant thanthan thethe specifispecifi c proposalsproposals isis tthehe observationobservation thatthat thethe veryvery acceptanceacceptance ofof thethe keykey rolerole playedplayed byby complexitycomplexity inin ssignifiignifi cantcant macroeconomicmacroeconomic eventsevents shouldshould bebe enoughenough toto pointpoint usus inin thethe directiondirection ofof tthehe kkindind ooff ppoliciesolicies tthathat ccanan hhelpelp ttoo llimitimit mmacroeconomicacroeconomic tturbulence.urbulence.

RRobustnessobustness A nnumberumber ooff rresearchersesearchers hhaveave ssoughtought ttoo ddesignesign ppolicyolicy fframeworksrameworks tthathat aarere rrobustobust ttoo ssmallmall mmodelingodeling mmistakesistakes bbyy tthehe ppolicymaker.olicymaker. FForor eexample,xample, HHansen,ansen, SSargent,argent, aandnd ttheirheir cco-authorso-authors hhaveave mmadeade ssubstantialubstantial pprogressrogress iinn iincorporatingncorporating rrobustobust ccontrolontrol ttechniquesechniques ttoo eeconomicconomic ppolicyolicy aanalysisnalysis ((forfor eexample,xample, HHansen,ansen, SSargent,argent, aandnd TTallarini,allarini, 11999;999; HHansenansen aandnd SSargent,argent, 22007;007; CCogley,ogley, CColacito,olacito, HHansen,ansen, aandnd SSargent,argent, 22008;008; KKarantounias,arantounias, HHansen,ansen, aandnd SSargent,argent, 22009).009). WWoodfordoodford ((2010)2010) hhasas eexploredxplored tthehe ssameame bbroadroad iissuessue iinn tthehe ccontextontext ooff tthehe sstandardtandard NNew-Keynesianew-Keynesian mmodelodel uusedsed iinn ccentralentral bbanks’anks’ rresearchesearch ddepartments.epartments. TThishis sstrategytrategy iiss cclearlylearly a ssteptep iinn tthehe rrightight ddirection,irection, aalthoughlthough I ssuspectuspect tthehe ddeviationseviations ttheyhey cconsideronsider ffromrom tthehe ccoreore mmodelsodels aarere sstilltill ttoooo llocalocal ttoo ccaptureapture tthehe eenormousnormous uuncertaintiesncertainties aandnd cconfusiononfusion tthathat ppolicymakersolicymakers ffaceace iinn rrealisticealistic nnontrivialontrivial sscenarios.cenarios. BButut tthishis lliteratureiterature hhasas mmanyany ooff tthehe rrightight wwordsords iinn iit.t. TThehe naturalnatural nextnext stepstep forfor thisthis robustnessrobustness literatureliterature isis toto incorporateincorporate massivemassive uuncertainty.ncertainty. ThisThis stepstep maymay alsoalso harborharbor somesome ofof thethe answersanswers onon howhow toto dealdeal withwith qquantitativeuantitative statementsstatements inin a highlyhighly uncertainuncertain environment.environment. AsAs a startingstarting point,point, wewe pprobablyrobably needneed toto relaxrelax thethe artifiartifi cialcial micro-foundationmicro-foundation constraintsconstraints imposedimposed justjust forfor tthehe ssakeake ooff beingbeing ableable toto generategenerate “structural”“structural” generalgeneral equilibriumequilibrium simulations.simulations. WWhenhen cclosinglosing a quantitative model,model, therethere shouldshould bebe anan explicitexplicit correspondencecorrespondence bbetweenetween thethe knowledgeknowledge wewe assumeassume inin suchsuch closureclosure andand thethe statestate ofof ourour knowledgeknowledge aaboutbout suchsuch closure.closure. ThisThis meansmeans replacingreplacing artifiartifi cialcial structuralstructural equationsequations forfor looserlooser oones,nes, oorr eevenven fforor rreduced-formeduced-form data-relationshipsdata-relationships ifif thisthis allall thatthat wewe reallyreally know.know. IItt iiss nnotot nnearlyearly enoughenough (although(although itit isis progress)progress) toto dodo BayesianBayesian estimationestimation ofof thethe ddynamicynamic sstochastictochastic ggeneraleneral equilibriumequilibrium mmodel,odel, fforor tthehe aabsencebsence ooff kknowledgenowledge iiss ffarar mmoreore ffundamentalundamental tthanhan ssuchuch aann approachapproach admitsadmits (for(for example,example, Fernandez-Fernandez- VVillaverde,illaverde, 22009,009, pprovidesrovides a ggoodood ssurveyurvey ooff tthishis literature).literature). WWee needneed toto reworkrework thethe mechanismmechanism thethe corecore currentlycurrently usesuses toto gogo fromfrom insightsinsights dderivederived iinn tthehe pperipheryeriphery ttoo qquantitativeuantitative ggeneraleneral eequilibriumquilibrium oones.nes. TTherehere hhasas bbeeneen cconsiderableonsiderable progressprogress inin formalizingformalizing aargumentsrguments fforor hhowow ddynamicynamic aadjustmentsdjustments hhappen,appen, butbut wwhilehile tthishis iiss cclearlylearly pprogress,rogress, iitt iiss nnotot a ssubstituteubstitute fforor tthehe kkindind ooff pprogressrogress I aadvocatedvocate hhere,ere, wwhichhich iiss ttoo aacknowledgecknowledge eexplicitlyxplicitly oourur ddegreeegree ooff iigno-gno- rrance.ance.1100 TTherehere areare manymany instancesinstances inin whichwhich ourour knowledgeknowledge ofof thethe truetrue structuralstructural

10 As Lucas (2003), following on a Friedman (1946) lead, points out: “[T]he theory is never really solved. What are the predictions of Patinkin’s model? The model is too complicated to work them out. All the dynamics are the mechanical auctioneer dynamics that Samuelson introduced, where anything can happen. . . . You can see from his verbal discussion that he’s reading a lot of economics into these dynamics. What are people thinking? What are they expecting? . . . He’s really thinking about intertem- poral substitution. He doesn’t know how to think about it, but he is trying to.” In the specifi c example used here by Lucas, the Lucas–Rapping (1969) model, for example, did represent signifi cant progress Ricardo J. Caballero 99

rrelationshipelationship iiss eextremelyxtremely llimited.imited. IInn ssuchuch ccases,ases, tthehe mmainain pproblemroblem iiss nnotot iinn hhowow ttoo formalizeformalize anan intuition,intuition, butbut inin thethe assumptionassumption thatthat thethe structuralstructural relationshiprelationship isis kknownnown withwith precision.precision. SuperimposingSuperimposing a specifispecifi c optimizationoptimization paradigmparadigm isis notnot thethe ssolutionolution ttoo tthishis ppervasiveervasive pproblem,roblem, aass mmuchuch ooff tthehe ddiffiiffi ccultyulty lliesies ppreciselyrecisely iinn nnotot kknowingnowing which,which, andand whose,whose, optimizationoptimization problemproblem isis toto bebe solved.solved. ForFor thisthis reasonreason tthehe solutionsolution iiss nnotot ssimplyimply ttoo eexplorexplore a wwideide rrangeange ooff pparametersarameters fforor a sspecifipecifi c mmechanism.echanism. TThehe pproblemroblem iiss tthathat wwee ddoo nnotot kknownow tthehe mechanism, nnotot jjustust tthathat wwee ddon’ton’t knowknow iitsts sstrength.trength. BButut howhow dodo wewe gogo aboutabout doingdoing policypolicy analysisanalysis inin modelsmodels withwith somesome looselyloosely sspecifipecifi eedd bblockslocks notnot pinnedpinned downdown byby specifispecifi c fi rrst-orderst-order cconditions?onditions? WWelcomeelcome ttoo tthehe realreal world!world! ThisThis tasktask isis whatwhat actualactual policymakerspolicymakers face.face. AcademicAcademic modelsmodels oftenoften pproviderovide ppreciserecise ppolicyolicy pprescriptionsrescriptions bbecauseecause tthehe sstructure,tructure, sstates,tates, aandnd mmecha-echa- nnismsisms areare sharplysharply defidefi ned.ned. InIn contrast,contrast, policymakerspolicymakers dodo notnot havehave thesethese luxuries.luxuries. TThoughtfulhoughtful ppolicymakersolicymakers uusese aacademiccademic iinsightsnsights ttoo tthinkhink aaboutbout tthehe ttypeype ooff ppoliciesolicies ttheyhey maymay wwantant ttoo cconsider,onsider, bbutut tthenhen ttryry ttoo uunderstandnderstand tthehe iimplicationsmplications ooff ssuchuch ppoliciesolicies whenwhen somesome ((oror mmost)ost) ooff tthehe aassumptionsssumptions ooff tthehe uunderlyingnderlying ttheoreticalheoretical mmodelodel dodo notnot hhold.old. HHowever,owever, tthishis kkindind ooff rrobustnessobustness aanalysisnalysis iiss nnearlyearly aabsentbsent iinn oourur modeling.modeling. IInn tthishis ssense,ense, aandnd aass I mmentionedentioned eearlier,arlier, tthehe wworkork ooff HHansenansen aandnd SSargentargent (2007)(2007) aandnd oothersthers oonn rrobustobust ccontrolontrol iinn ppolicymakingolicymaking ppointsoints iinn tthehe rrightight ddirection,irection, althoughalthough I tthinkhink wwee nneedeed ttoo ggoo mmuch,uch, mmuchuch ffurtherurther iinn rreducingeducing tthehe aamountmount andand type ooff knowledgeknowledge policymakerspolicymakers and eeconomicconomic agentsagents areare assumedassumed ttoo possess.possess. OOnene primaryprimary ddrivingriving forceforce behindbehind modernmodern macroeconomicsmacroeconomics (both(both corecore andand pperiphery)eriphery) waswas aann aattemptttempt ttoo ccircumventircumvent tthehe LLucasucas ccritique—theritique—the aargumentrgument tthathat mmarketarket pparticipantsarticipants taketake thethe ppolicyolicy regimeregime iintonto aaccountccount aandnd ssoo eestimatesstimates ooff eeconomicconomic pparametersarameters fforor oonene ppolicyolicy rregimeegime mmayay wwellell nnotot bbee vvalidalid iiff tthehe ppolicyolicy rregimeegime cchanges.hanges. IIff wwee nownow replacereplace somesome fi rst-orderrst-order conditionsconditions byby empiricalempirical relationshipsrelationships andand theirtheir ddistributions,istributions, doesn’tdoesn’t thisthis critiquecritique returnreturn toto haunthaunt us?us? TheThe answeranswer mustmust bebe “yes,”“yes,” atat lleasteast ttoo ssomeome eextent.xtent. BButut iiff wwee ddoo nnotot hhaveave ttruerue kknowledgenowledge aaboutbout tthehe rrelationshipelationship aandnd iitsts ssource,ource, tthenhen aassumingssuming tthehe wwrongrong sspecifipecifi c fi rst-orderrst-order conditioncondition cancan alsoalso bebe a ssourceource ofof misguidedmisguided policypolicy prescription.prescription. BothBoth thethe ad-hocad-hoc modelmodel andand thethe particularparticular sstructuraltructural modelmodel mmakeake uunwarrantednwarranted sspecifipecifi c assumptionsassumptions aboutabout agents’agents’ adaptationadaptation ttoo tthehe nnewew ppolicyolicy environment.environment. TheThe LucasLucas critiquecritique isis clearlyclearly valid,valid, butbut forfor manymany ((most?)most?) ppolicyolicy questionsquestions wewe haven’thaven’t yetyet foundfound thethe solution—wesolution—we onlyonly havehave thethe ppretenseretense ooff a ssolution.olution. UUltimately,ltimately, fforor policypolicy prescriptions,prescriptions, itit isis importantimportant toto assignassign differentdifferent weightsweights ttoo tthosehose tthathat ffollowollow ffromrom bblockslocks ooverver wwhichhich wwee hhaveave ttruerue kknowledge,nowledge, aandnd tthosehose tthathat ffollowollow ffromrom vveryery llimitedimited kknowledge.nowledge. SSomeome ooff tthishis hhasas aalreadylready bbeeneen ddoneone iinn tthehe aassetsset ppricingricing literature:literature: forfor example,example, Ang,Ang, Dong,Dong, andand PiazzesiPiazzesi (2007)(2007) useuse arbitragearbitrage ttheoryheory ttoo constrainconstrain anan otherwiseotherwise nonstructuralnonstructural econometriceconometric studystudy ofof thethe yieldyield curvecurve aandnd Taylor’sTaylor’s rule.rule. PerhapsPerhaps a similarsimilar routeroute cancan bebe followedfollowed inin macroeconomicsmacroeconomics toto ggaugeauge thethe orderorder ooff mmagnitudeagnitude ooff ssomeome kkeyey eeffectsffects aandnd mmechanisms,echanisms, wwhichhich ccanan tthenhen over the Patinkin modeling of labor dynamics. It solved the “how” part of the formal underpinning of Patinkin’s dynamics. But as I point out in the text, this kind of solution is insuffi cient. 100 Journal of Economic Perspectives

bbee ccombinedombined withwith peripheryperiphery insightsinsights toto generategenerate back-of-the-envelope-typeback-of-the-envelope-type calcula-calcula- ttions.ions. ForFor now,now, wewe shouldn’tshouldn’t pretendpretend thatthat wewe knowknow moremore thanthan this,this, althoughalthough thisthis iiss nono reasonreason toto givegive upup hope.hope. WeWe havehave mademade enormousenormous progressprogress overover thethe lastlast fewfew ddecadesecades inin thethe formalizationformalization ofof macroeconomics.macroeconomics. WeWe justjust gotgot a littlelittle carriedcarried awayaway wwithith tthehe bbeautifuleautiful structuresstructures thatthat eemergedmerged ffromrom tthishis pprocess.rocess.

TThehe PPretenseretense ooff KKnowledgenowledge

TThehe rootroot causecause ofof thethe poorpoor statestate ofof affairsaffairs inin thethe fi eldeld ofof macroeconomicsmacroeconomics lieslies iinn a ffundamentalundamental tensiontension inin academicacademic macroeconomicsmacroeconomics betweenbetween thethe enormousenormous ccomplexityomplexity ooff iitsts ssubjectubject aandnd tthehe mmicro-theory-likeicro-theory-like pprecisionrecision ttoo wwhichhich wwee aaspire.spire. TThishis ttensionension iiss nnotot nnew.ew. TThehe ooldld iinstitutionalnstitutional sschoolchool cconcludedoncluded tthathat tthehe ttaskask wwasas iimpossiblempossible aandnd hhenceence nnotot wworthorth fformalizingormalizing iinn mmathematicalathematical ttermserms ((forfor eexample,xample, SSamuels,amuels, 1987,1987, andand referencesreferences therein).therein). NarrativeNarrative waswas thethe chosenchosen tool,tool, asas nono math-math- eematicalmatical modelmodel couldcould capturecapture thethe richnessrichness ofof thethe worldworld thatthat isis toto bebe explained.explained. HHowever,owever, tthishis aapproachpproach diddid notnot solvesolve thethe conundrum;conundrum; itit merelymerely postponedpostponed it.it. TheThe mmodernodern corecore ofof macroeconomicsmacroeconomics swungswung thethe pendulumpendulum toto thethe otherother extreme,extreme, andand hhasas specializedspecialized inin quantitativequantitative mathematicalmathematical formalizationsformalizations ofof a preciseprecise butbut largelylargely iirrelevantrrelevant world.world. ThisThis approachapproach hashas notnot solvedsolved thethe conundrumconundrum either.either. I wishwish thethe ssolutionolution waswas toto bebe foundfound somewheresomewhere inin betweenbetween thesethese polarpolar opposites,opposites, butbut itit isis notnot cclearlear whatwhat “in“in between”between” meansmeans forfor a rangerange thatthat hashas a frameworkframework basedbased onon verbalverbal ddiscussionsiscussions ofof thethe realreal worldworld onon oneone endend andand oneone basedbased onon quantitativequantitative analysisanalysis ofof aann ““alternative”alternative” wworld,orld, onon thethe oother.ther. TThehe peripheryperiphery ofof macroeconomicsmacroeconomics hashas muchmuch toto offeroffer inin termsterms ofof specifispecifi c iinsightsnsights andand mechanisms,mechanisms, butbut toto fulfifulfi llll thethe ambitionambition ofof thethe corecore wewe needneed toto changechange tthehe pparadigmaradigm ttoo ggoo ffromrom tthesehese iinsightsnsights oonn tthehe ppartsarts ttoo tthehe bbehaviorehavior ooff tthehe wwhole.hole. IItt iiss nnotot aaboutbout eembeddingmbedding tthesehese iintonto ssomeome vversionersion ooff tthehe ccanonicalanonical rrealeal bbusinessusiness ccycleycle mmodel.odel. IItt iis,s, aamongmong ootherther tthings,hings, aaboutbout ccapturingapturing ccomplexomplex iinteractionsnteractions aandnd tthehe cconfusiononfusion tthathat ttheyhey ccanan ggenerate.enerate. FFromrom a ppolicyolicy perspective,perspective, thethe specifispecifi cscs ofof a crisiscrisis areare onlyonly knownknown onceonce thethe ccrisisrisis starts.starts. ForFor thisthis reason,reason, mymy sensesense isis that,that, contrarycontrary toto thethe hopehope ofof policymakerspolicymakers aandnd rregulators,egulators, therethere isis limitedlimited scopescope forfor policypolicy thatthat cancan inin advanceadvance eliminateeliminate thethe rriskisk oror costscosts ofof fi nancialnancial crisis,crisis, beyondbeyond somesome common-sensecommon-sense measuresmeasures (like(like capitalcapital rrequirementsequirements forfor fi nancialnancial institutions)institutions) andand veryvery generalgeneral public–privatepublic–private insuranceinsurance aarrangementsrrangements (like(like depositdeposit insurance).insurance). ByBy thethe timetime a ttruerue fi nancialnancial crisiscrisis isis underway,underway, tthehe immediatelyimmediately relevantrelevant policypolicy issuesissues areare nono longerlonger aboutabout whetherwhether interventionintervention mmightight breedbreed moralmoral hazard,hazard, butbut aboutabout a sociallysocially wastefulwasteful reluctancereluctance toto investinvest andand toto hhire,ire, aandnd tthehe eextentxtent ttoo wwhichhich ppredatoryredatory ttradingrading oorr fi rere salessales cancan bebe minimized.minimized. GGoingoing bbackack ttoo oourur mmacroeconomicacroeconomic mmodels,odels, wwee nneedeed ttoo sspendpend mmuchuch mmoreore eeffortffort iinn uunderstandingnderstanding thethe topologytopology ofof interactionsinteractions inin realreal economies.economies. TheThe fi nan-nan- ccialial sectorsector andand itsits recentrecent strugglesstruggles havehave mademade tthishis nneedeed vvividlyividly cclear,lear, bbutut tthishis iissuessue iiss ccertainlyertainly nnotot eexclusivexclusive ttoo tthishis ssector.ector. TThehe cchallengeshallenges areare big,big, butbut macroeconomistsmacroeconomists cancan nono longerlonger continuecontinue playingplaying iinternalnternal ggames.ames. TThehe aalternativelternative ooff lleavingeaving aallll tthehe iimportantmportant sstufftuff ttoo tthehe ““policy”-typespolicy”-types Macroeconomics after the Crisis: The Pretense-of-Knowledge Syndrome 101

aandnd iinformalnformal commentatorscommentators cannotcannot bebe thethe rightright approach.approach. I dodo notnot havehave thethe aanswer.nswer. BButut I suspectsuspect thatthat whateverwhatever thethe solutionsolution ultimatelyultimately is,is, wewe willwill accelerateaccelerate ourour cconvergenceonvergence toto it,it, andand rreduceeduce thethe ddamageamage wwee ddoo aalonglong tthehe ttransition,ransition, iiff wwee ffocusocus oonn rreducingeducing tthehe eextentxtent ooff oourur ppretense-of-knowledgeretense-of-knowledge ssyndrome.yndrome.

■ I thank Daron Acemoglu, David Autor, Abhijit Banerjee, Olivier Blanchard, , Francesco Giavazzi, Jonathan Goldberg, Chad Jones, Bengt Holmström, Arvind Krishnamurthy, John List, Guido Lorenzoni, James Poterba, Alp Simsek, , and Timothy Taylor for their comments. Of course they are not responsible for my tirade.

References

Acemoglu, Daron, Asuman Ozdaglar, and in a Flight to Quality Episode.” Journal of Finance, Alireza Tahbaz-Zalehi. 2010. “Cascades in Networks 63(5): 2195–2229. and Aggregate Volatility.” Available at: http:// Caballero, Ricardo J., and Arvind Krish- econ-www.mit.edu/faculty/acemoglu/paper. namurthy. 2008b. “Knightian Uncertainty and Albert, Réka, and Albert-László Barabási. 2002. Its Implications for the TARP.” Financial Times “Statistical Mechanics of Complex Networks.” Economists’ Forum, November 24. Review of Modern Physics, 74(1): 47–97. Caballero, Ricardo J., and Pablo Kurlat. 2009. Allen, Franklin, and Douglas Gale. 2000. “Finan- “The ‘Surprising’ Origin and Nature of Financial cial Contagion.” Journal of Political Economy, 108(1): Crises: A Macroeconomic Policy Proposal.” Prepared 1–33. for the Jackson Hole Symposium on Financial Ang, Andrew, Sen Dong, and Monika Piazzesi. Stability and Macroeconomic Policy, August. http:// 2007. “No-Arbitrage Taylor Rules.” Available at www.kansascityfed.org/publicat/sympos/2009 SSRN: http://ssrn.com/abstract=621126. /papers/caballeroKurlat.07.29.09.pdf. Arthur, William B., Steven N. Durlauf, and Caballero, Ricardo J., and Alp Simsek. 2009a. David A. Lane, eds. 1997. The Economy as an Evolving “Complexity and Financial Panics.” NBER Working Complex System II. Reedwood City: Addison-Wesley. Paper 14997. Bak, Per, Kan Chen, Jose Scheinkman, and Caballero, Ricardo J., and Alp Simsek. 2009b. Michael Woodford. 1993. “Aggregate Fluctuations “Fire Sales in a Model of Complexity.” http:// from Independent Sectoral Shocks: Self-Organized econ-www.mit.edu/fi les/4736. Criticality in a Model of Production and Inventory Chari, V. V., Patrick J. Kehoe, and Ellen R. Dynamics.” Richerche Economiche, 47(1): 3–30. McGrattan. 2007. “Business Cycle Accounting.” Bernanke, Ben, and Mark Gertler. 1989. Econometrica, 75(3): 781–836. “Agency Costs, Net Worth and Business Fluctua- Cogley, Timothy, Riccardo Colacito, Lars Peter tions.” The American Economic Review, 79(1): 14–31. Hansen, and Thomas J. Sargent. 2008. “Robustness Brock, William A. 1993. “Pathways to Random- and U.S. Monetary Policy Experimentation.” Avail- ness in the Economy: Emergent Nonlinearity able at SSRN: http://ssrn.com/abstract=1267033. and Chaos in Economics and Finance.” Estudios (Forthcoming in Journal of Money, Credit, and Económicos, 8(1): 3–55. Banking). Brock, William, and Steven N. Durlauf. 2001. Dang, Tri Vi, Gary Gorton, and Bengt “Discrete Choice with Social Interactions.” Review Holmström. 2009. “Ignorance and the Opti- of Economic Studies, 68(2): 235–60. mality of Debt for the Provision of Liquidity.” Caballero, Ricardo J. 2009. “Sudden Financial http://www4.gsb.columbia.edu/null Arrest.” Prepared for the Mundell-Fleming Lecture /download?&exclusive=filemgr.download&file delivered at the Tenth Jacques Polak Annual _id=7213758. Research Conference, IMF, November 8. Durlauf, Steven N. 1993. “Nonergodic Caballero, Ricardo J. 2010. “Crisis and Reform: Economic Growth.” Review of Economic Studies, Managing Systemic Risk.” Prepared for the XI 60(2): 349–66. Angelo Costa Lecture delivered in Rome on Durlauf, Steven N. 1997. “Statistical Mechanics March 23. Approaches to Socioeconomic Behavior.” In The Caballero, Ricardo J., and Arvind Krish- Economy as a Complex Evolving System II, ed. W. Brian namurthy. 2008a. “Collective Risk Management Arthur, Steven N. Durlauf, and David A. Lane, 102 Journal of Economic Perspectives

81-104. Redwood City, CA: Addison-Wesley. Knight, Frank H. 1921. Risk, Uncertainty and Durlauf, Steven N. 2004. “Complexity and Profi t. Boston: Houghton Miffl in. Empirical Economics.” Economic Journal, 115(504): Lucas, Robert E. 2003. “My Keynesian F225–F243. Education.” Keynote Address to the 2003 HOPE Ellsberg, Daniel. 1961. “Risk, Ambiguity, and Conference. the Savage Axioms.” Quarterly Journal of Economics, Lucas, Robert E., and Leonard A. Rapping. 75(4): 643–69. 1969. “Real Wages, Employment, and Infl ation.” Fernández-Villaverde, Jesús. 2009. “The Econo- Journal of Political Economy, 77(5): 721–54. metrics of DSGE Models.” NBER Working Paper Madigan, Brian F. 2009. “Bagehot’s Dictum 14677. in Practice: Formulating and Implementing Friedman, Milton. 1946. “Lange on Price Policies to Combat the Financial Crisis.” Speech Flexibility and Employment: A Methodological for Federal Reserve Bank of Kansas City’s Annual Critcism,” American Economic Review, 36(4): Economic Symposium, Jackson Hole, Wyoming, 613–631. (Reprinted in Essays in Positive August 21. Economics, pp. 277–300.) Mandelbrot, Benoit. 2008. PBS NewsHour Galí, Jordi, and Mark Gertler. 2007. “Macroeco- interview with Paul Solman, October 21. nomic Modeling for Monetary Policy Evaluation.” Mehrling, Perry, and Alistair Milne. 2008. Journal of Economic Perspectives, 21(4): 25–45. “Government’s Role as Credit Insurer of Last Geanakoplos, John. 2003. “Liquidity, Default, Resort and How It Can Be Fulfi lled.” Unpublished and Crashes: Endogenous Contracts in General paper. http://www.econ.barnard.columbia.edu Equilibrium.” In Advances in Economics and /faculty/mehrling/creditinsureroflastresort Econometrics: Theory and Applications, Eighth World fi nal09Oct2008.pdf. Conference, Vol. 2, 170–205. Milne, Alistair. 2009. The Fall of the House of Monographs. Credit. Cambridge University Press, Cambridge and Geanakoplos, John. 2009. “The Leverage Cycle.” New York. Cowles Foundation Discussion Paper 1715. Avail- Nordhaus, William. 2007. “Who’s Afraid of a Big able at SSRN: http://ssrn.com/abstract=1441943. Bad Oil Shock?” Prepared for Brookings Panel on Geanakoplos, John. 2010. “Solving the Present Economic Activity, September. http://nordhaus. Crisis and Managing the Leverage Cycle.” Cowles econ.yale.edu/Big_Bad_Oil_Shock_Meeting.pdf. Foundation Discussion Paper No. 1751. http:// Reinhart, Carmen M., and Kenneth S. Rogoff. cowles.econ.yale.edu/P/cd/d17b/d1751.pdf. 2009. This Time Is Different: Eight Centuries of Finan- Geanakoplos, John, and Ana Fostel. 2008. cial Folly. Princeton University Press. “Leverage Cycles and the Anxious Economy.” Ross, Stephen A. 2009. “A Modest Proposal.” American Economic Review, 98(4): 1211–44. Unpublished paper, MIT. Haldane, Andrew G. 2009. “Rethinking the Rotemberg, Julio J. 2009. “Liquidity Needs Financial Network.” Speech delivered at the Finan- in Economies with Interconnected Financial cial Student Association in Amsterdam on April 28. Obligations.” Unpublished paper, May 14. (Also an Hansen, Lars Peter, Thomas J. Sargent, and August, 2008, NBER Working Paper, No. 14222.) Thomas D. Tallarini, Jr. 1999. “Robust Permanent Samuels, Warren J. 1987. “Institutional Income and Pricing.” The Review of Economic Studies, Economics.” In The New Palgrave: A Dictionary of 66(4): 873–907. Economics, vol. 2, ed. by Murray Mitgate, Peter Hansen, Lars Peter, and Thomas J. Sargent. Newman, and John Eatwell. Macmillan. 2007. Robustness. Princeton University Press. Scheinkman, Jose, and Michael Woodford. 1994. Hayek, Friedrich A. 1974. “The Pretence of “Self-Organized Criticality and Economic Fluctua- Knowledge.” Prize Lecture, The Sveriges Riksbank tions.” American Economic Review, 84(2): 417–21. Prize in Economic Sciences in Memory of Alfred Shannon, Claude E. 1948. “A Mathematical Nobel. Theory of Communication.” Bell System Technical International Monetary Fund. 2009. Global Journal, Vol. 27, July, pp. 379–423, and Vol. 27, Financial Stability Report: Navigating the Financial October, 623–56. Challenges Ahead, October. Preliminary version Shin, Hyun Song. 2009. “Financial Intermedia- downloaded October 15, 2009. tion and the Post-Crisis Financial System.” Paper Jackson, Matthew O. 2008. Social and Economic presented at the 8th BIS Annual Conference, June. Networks. Princeton University Press. Sims, Christopher A. 2003. “Implications of Karantounias, Anastasios G., Lars Peter Rational Inattention.” Journal of Monetary Economics, Hansen, and Thomas J. Sargent. 2009. “Managing 50(3): 665–90, April. Expectations and Fiscal Policy.” Federal Reserve Tirole, Jean. 2006. The Theory of Corporate Bank of Atlanta Working Paper 2009-29. October. Finance. Princeton University Press. Kiyotaki, Nobuhiro, and . 1997. Woodford, Michael. 2010. “Robustly Optimal “Credit Cycles.” Journal of Political Economy, 105(2): Monetary Policy with Near-Rational Expectations.” 211–48. American Economic Review, 100(1): 274–303. This article has been cited by:

1. Guidon Fenig, Luba Petersen. 2017. Distributing scarce jobs and output: experimental evidence on the dynamic effects of rationing. Experimental Economics . [CrossRef] 2. Arne HeiseDie Ökonomik als wissenschaftliches Macht- und Schlachtfeld 55-81. [CrossRef] 3. Hanno Pahl, Jan SparsamDSGE-Makroökonomik und die Krise 135-158. [CrossRef] 4. Nicola Giocoli. 2016. Truth or precision? Some reflections on the economists’ failure to predict the financial crisis. The Review of Austrian Economics 29:4, 371-386. [CrossRef] 5. Lawrence H. WhiteHayek and Contemporary Macroeconomics 41-61. [CrossRef] 6. Ismail Erturk. 2016. Shadow banking: a story of the Doppelgänger (the Double) in science of finance. Journal of Cultural Economy 1-16. [CrossRef] 7. Joshua R. Hendrickson. 2016. Interest rates and investment coordination failures. The Review of Austrian Economics . [CrossRef] 8. Matteo G. Richiardi. 2016. The Future of Agent-Based Modeling. Eastern Economic Journal . [CrossRef] 9. Arne Heise, Sebastian Thieme. 2016. The Short Rise and Long Fall of Heterodox Economics in Germany After the 1970s: Explorations in a Scientific Field of Power and Struggle. Journal of Economic Issues 50:4, 1105-1130. [CrossRef] 10. Dirk J. Bezemer. 2016. Towards an ‘accounting view’ on money, banking and the macroeconomy: history, empirics, theory. Cambridge Journal of Economics 40:5, 1275-1295. [CrossRef] 11. Nicolás Cachanosky, Peter Lewin. 2016. An empirical application of the EVA® framework to business cycles. Review of Financial Economics 30, 60-67. [CrossRef] 12. Davoud Taghawi-Nejad, Vipin P. Veetil. 2016. The Complexity of Coordination. Eastern Economic Journal . [CrossRef] 13. RÜDIGER BACHMANN, LIN MA. 2016. Lumpy Investment, Lumpy Inventories. Journal of Money, Credit and Banking 48:5, 821-855. [CrossRef] 14. Adam Fforde. 2016. Confirmation bias: methodological causes and a palliative response. Quality & Quantity . [CrossRef] 15. Vipin P. Veetil, Lawrence H. White. 2016. Towards a New Austrian Macroeconomics. The Review of Austrian Economics . [CrossRef] 16. Teodoro Dario Togati. 2016. How can we explain the persistence of the Great Recession? A balanced stability approach. Cambridge Journal of Economics 40:4, 1077-1101. [CrossRef] 17. Stanislao Gualdi, Marco Tarzia, Francesco Zamponi, Jean-Philippe Bouchaud. 2016. Monetary policy and dark corners in a stylized agent-based model. Journal of Economic Interaction and Coordination . [CrossRef] 18. Peter LewinPolicy Design and Execution in a Complex World: Can We Learn from the Financial Crisis? 265-283. [CrossRef] 19. Apostolos Serletis. 2016. INTRODUCTION TO MACROECONOMIC DYNAMICS SPECIAL ISSUE ON COMPLEXITY IN ECONOMIC SYSTEMS. Macroeconomic Dynamics 20:02, 461-465. [CrossRef] 20. Nicolás Cachanosky, Alexander W. Salter. 2016. The view from Vienna: An analysis of the renewed interest in the Mises-Hayek theory of the business cycle. The Review of Austrian Economics . [CrossRef] 21. Jianjun Miao. 2016. Introduction to the symposium on bubbles, multiple equilibria, and economic activities. Economic Theory 61:2, 207-214. [CrossRef] 22. Hanno Pahl. 2016. A Changing Face of Mainstream Economics? Agentenbasierte Modellierungen in der Makroökonomik. Soziale Systeme 20:1. . [CrossRef] 23. Lukasz Hardt. 2016. The Recent Critique of Theoretical Economics: A Methodologically Informed Investigation. Journal of Economic Issues 50:1, 269-287. [CrossRef] 24. Ivan Boldyrev, Ekaterina SvetlovaAfter the Turn: How the Performativity of Economics Matters 1-27. [CrossRef] 25. Eduardo Pol. 2015. A theorem on the methodology of positive economics. Cogent Economics & Finance 3:1. . [CrossRef] 26. S. SaadahChaos and Complex in Measuring Sustainability of Economic Condition in Indonesia 1-4. [CrossRef] 27. JW Nevile, GC Harcourt, Peter Kriesler. 2015. Macroeconomic Policy for the Real World: A Post- Keynesian Perspective. Economic Papers: A journal of applied economics and policy 34:3, 108-117. [CrossRef] 28. Gurbachan Singh. 2015. Thinking afresh about central bank’s interest rate policy. Journal of Financial Economic Policy 7:3, 221-232. [CrossRef] 29. Michael G. Jacobides. 2015. What drove the financial crisis? Structuring our historical understanding of a predictable evolutionary disaster. Business History 57:5, 716-735. [CrossRef] 30. Benjamin Braun. 2015. Governing the future: the European Central Bank’s expectation management during the Great Moderation. Economy and Society 44:3, 367-391. [CrossRef] 31. Nicolas Cachanosky. 2015. Expectation in Austrian business cycle theory: Market share matters. The Review of Austrian Economics 28:2, 151-165. [CrossRef] 32. Jakob Grazzini, Matteo Richiardi. 2015. Estimation of ergodic agent-based models by simulated minimum distance. Journal of Economic Dynamics and Control 51, 148-165. [CrossRef] 33. Hanno Pahl. 2015. Steuerungsabstinenz als Ordnungsvision: Soziologische Beobachtungen zur Rational Expectations Revolution in der Makroökonomik. Zeitschrift für kritische Sozialtheorie und Philosophie 2:2. . [CrossRef] 34. Stanislao Gualdi, Marco Tarzia, Francesco Zamponi, Jean-Philippe Bouchaud. 2015. Tipping points in macroeconomic agent-based models. Journal of Economic Dynamics and Control 50, 29-61. [CrossRef] 35. Hiroshi Yoshikawa. 2015. Stochastic macro-equilibrium: a microfoundation for the Keynesian economics. Journal of Economic Interaction and Coordination 10:1, 31. [CrossRef] 36. Itay Goldstein, Dirk Hackbarth. 2014. Corporate finance theory: Introduction to special issue. Journal of Corporate Finance 29, 535-541. [CrossRef] 37. Nicolas Cachanosky. 2014. The Mises-Hayek business cycle theory, fiat currencies and open economies. The Review of Austrian Economics 27:3, 281-299. [CrossRef] 38. John C. Driscoll, Steinar Holden. 2014. Behavioral economics and macroeconomic models. Journal of Macroeconomics 41, 133-147. [CrossRef] 39. Pierre-Richard Agénor, Luiz A. Pereira da Silva. 2014. Macroprudential regulation and the monetary transmission mechanism. Journal of Financial Stability 13, 44-63. [CrossRef] 40. Jianjun Miao. 2014. Introduction to economic theory of bubbles. Journal of Mathematical Economics 53, 130-136. [CrossRef] 41. Nicolas Cachanosky. 2014. The effects of U.S. monetary policy on Colombia and Panama (2002– 2007). The Quarterly Review of Economics and Finance 54:3, 428-436. [CrossRef] 42. Arne Heise. 2014. The Future of Economics in a Lakatos–Bourdieu Framework. International Journal of Political Economy 43:3, 70-93. [CrossRef] 43. 김김김, Kyuho Kang. 2014. Analysis on Recent Changes in the Covered Interest Rate Parity Condition. KDI Journal of Economic Policy 36:2, 103-136. [CrossRef] 44. Nicola Giocoli. 2014. Network efficiency and the banking system. International Review of Economics 61:3, 203. [CrossRef] 45. Tamim Bayoumi, Giovanni Dell'Ariccia, Karl Habermeier, Tommaso Mancini Griffoli, Fabian Valencia. 2014. Monetary Policy in the New Normal. Staff Discussion Notes 14:3, 1. [CrossRef] 46. Niels Anthonisen. 2013. On the long-run relationship between inflation and output in a spatial overlapping generations model. Journal of Economic Dynamics and Control 37:12, 2500-2524. [CrossRef] 47. Moritz Schularick. 2013. Public and Private Debt: The Historical Record (1870-2010). German Economic Review n/a-n/a. [CrossRef] 48. Elias Hurmekoski, Lauri Hetemäki. 2013. Studying the future of the forest sector: Review and implications for long-term outlook studies. Forest Policy and Economics 34, 17-29. [CrossRef] 49. Camelia Minoiu, Javier A. Reyes. 2013. A network analysis of global banking: 1978–2010. Journal of Financial Stability 9:2, 168-184. [CrossRef] 50. Jean-Pierre Allegret, Cécile Couharde, Valérie Mignon. 2013. Introduction: Recent international macroeconomic and financial issues. International Economics 133, 1-7. [CrossRef] 51. Costas Karfakis. 2013. Credit and business cycles in Greece: Is there any relationship?. Economic Modelling 32, 23-29. [CrossRef] 52. Danilo Delpini, Stefano Battiston, Massimo Riccaboni, Giampaolo Gabbi, Fabio Pammolli, Guido Caldarelli. 2013. Evolution of Controllability in Interbank Networks. Scientific Reports 3. . [CrossRef] 53. Guido Caldarelli, Alessandro Chessa, Fabio Pammolli, Andrea Gabrielli, Michelangelo Puliga. 2013. Reconstructing a credit network. Nature Physics 9:3, 125-126. [CrossRef] 54. Pedro de Araujo, Roisin O’Sullivan, Nicole B. Simpson. 2013. What Should be Taught in Intermediate Macroeconomics?. The Journal of Economic Education 44:1, 74-90. [CrossRef] 55. Luis-Felipe Arizmendi. 2013. An Extended Model of Currency Options Applicable as Policy Tool for Central Banks with Inflation Targeting and Dollarized Economies. Theoretical Economics Letters 03:03, 164-167. [CrossRef] 56. William D Craighead. 2013. Expectations, Employment and Prices, by Roger E.A. Farmer How the Economy Works: Confidence, Crashes and Self-Fulfilling Prophecies, by Roger E.A. Farmer. Eastern Economic Journal 39:1, 140-142. [CrossRef] 57. Andrew Mearman. 2012. ‘Heterodox economics’ and the problems of classification. Journal of Economic Methodology 19:4, 407-424. [CrossRef] 58. S. Cincotti, D. Sornette, P. Treleaven, S. Battiston, G. Caldarelli, C. Hommes, A. Kirman. 2012. An economic and financial exploratory. The European Physical Journal Special Topics 214:1, 361-400. [CrossRef] 59. ANDREA TEGLIO, MARCO RABERTO, SILVANO CINCOTTI. 2012. THE IMPACT OF BANKS' CAPITAL ADEQUACY REGULATION ON THE ECONOMIC SYSTEM: AN AGENT-BASED APPROACH. Advances in Complex Systems 15:supp02, 1250040. [CrossRef] 60. Stijn Claessens, M. Ayhan Kose, Marco E. Terrones. 2012. How do business and financial cycles interact?. Journal of International Economics 87:1, 178-190. [CrossRef] 61. Julien Chevallier. 2012. Global imbalances, cross-market linkages, and the financial crisis: A multivariate Markov-switching analysis. Economic Modelling 29:3, 943-973. [CrossRef] 62. In the Wake of the Crisis: Leading Economists Reassess Economic Policy . [CrossRef] 63. Sharon Kozicki. 2011. Macro has progressed. Journal of Macroeconomics . [CrossRef] 64. M. HASHEM PESARAN, RON P. SMITH. 2011. BEYOND THE DSGE STRAITJACKET1. The Manchester School 79, 5-16. [CrossRef] 65. Joseph E. Stiglitz. 2011. RETHINKING MACROECONOMICS: WHAT FAILED, AND TO HOW REPAIR IT. Journal of the European Economic Association 9:4, 591-645. [CrossRef] 66. Anjan V. Thakor. 2011. Incentives to innovate and financial crises. Journal of Financial Economics . [CrossRef] 67. Gregory Waymire, Sudipta Basu. 2011. Economic crisis and accounting evolution. Accounting and Business Research 41:3, 207-232. [CrossRef] 68. Roger Koppl, William J. Luther. 2011. Hayek, Keynes, and modern macroeconomics. The Review of Austrian Economics . [CrossRef] 69. Edoardo Gaffeo, Roberto Tamborini. 2011. If the Financial System Is Complex, How Can We Regulate It?. International Journal of Political Economy 40:2, 79-97. [CrossRef] 70. Neil Hart. 2011. Mainstream Macroeconomics: A ‘Keynesian’ Revival?. The Economic and Labour Relations Review 22:1, 17-40. [CrossRef] 71. M. Ayhan Kose. 2011. Journal of International Economics . [CrossRef] 72. Javier A. Reyes, Camelia Minoiu. 2011. A Network Analysis of Global Banking:1978-2009. IMF Working Papers 11:74, 1. [CrossRef] 73. İsmail Şiriner, Keremet ShaiymbetovaImpacts of Global Financial Crisis and Changes in Monetary Policy of Central Banks: 474-504. [CrossRef]