/
How to Build an Economic Model in Your Spare Time by Hal R How to Build an Economic Model in Your Spare Time by Hal R

How to Build an Economic Model in Your Spare Time by Hal R - PDF document

tawny-fly
tawny-fly . @tawny-fly
Follow
511 views
Uploaded On 2014-10-06

How to Build an Economic Model in Your Spare Time by Hal R - PPT Presentation

Varian UC Berkeley December 1994 Current version July 25 2009 Abstract This is an essay for Passion and Craft Economists at Work edited by Michael Szenberg University of Michigan Press 1997 Keywords Address Hal R Varian Dean School of Informa ID: 2878

Varian Berkeley December

Share:

Link:

Embed:


Presentation Transcript

pChapter1HOWTOBUILDANECONOMICMODELINYOURSPARETIMEThisisalittlearticlethatIwrotetodescribehowIwork.ItcontainstheadvicethatIwishIhadreceivedwhenIwasjuststartingout,anditismeanttobeentertainingaswellasinstructive.Mostofmyworkineconomicsinvolvesconstructingtheoreticalmodels.Overtheyears,Ihavedevelopedsomewaysofdoingthisthatmaybeworthdescribingtothosewhoaspiretopracticethisart.Inrealitytheprocessismuchmorehaphazardthanmydescriptionwouldsuggest|themodelofresearchthatIdescribeisanidealizationofreality,muchliketheeconomicmodelsthatIcreate.Butthereisprobablyenoughconnectionwithrealitytomakethedescriptionuseful|whichIhopeisalsotrueformyeconomicmodels.1.1GettingideasThe rststepistogetanidea.Thisisnotallthathardtodo.Thetrickypartistogetagoodidea.Thewayyoudothisistocomeupwithlotsandlotsofideasandthrowoutalltheonesthataren'tgood.Butwheretogetideas,that'sthequestion.Mostgraduatestudentsareconvincedthatthewayyougetideasistoreadjournalarticles.Butinmyexperiencejournalsreallyaren'taverygoodsourceoforiginalideas.Youcangetlotsofthingsfromjournalarticles|technique,insight,eventruth.Butmostofthetimeyouwillonlygetsomeoneelse'sideas.True,theymayleaveafewlooseendslyingaroundthatyoucanpickupon,butthereasontheyarelooseisprobablythattheauthorthoughtaboutthemawhileandcouldn't gureoutwhattodowiththemordecidedtheyweretootedioustobotherwith|whichmeansthatitislikelythatyouwill ndyourselfinthesamesituation. 2HOWTOBUILDANECONOMICMODELINYOURSPARETIMEMysuggestionisratherdi erent:Ithinkthatyoushouldlookforyourideasoutsidetheacademicjournals|innewspapers,inmagazines,incon-versations,andinTVandradioprograms.Whenyoureadthenewspaper,lookforthearticlesabouteconomics:::andthenlookattheonesthataren'tabouteconomics,becauselotsofthetimetheyendupbeingabouteconomicstoo.Magazinesareusuallybetterthannewspapersbecausetheygointoissuesinmoredepth.Ontheotherhand,ashalloweranalysismaybemorestimulating:there'snothinglikeafallaciousargumenttostimulateresearch.1Conversations,especiallywithpeopleinbusiness,areoftenveryfruitful.Commerceisconductedinmanyways,andmostofthemhaveneverbeensubjectedtoaseriouseconomicanalysis.Ofcourseyouhavetobecarefulnottobelieveeverythingyouhear|peopleinbusinessusuallyknowasetofrulesthatworkwellforrunningtheirownbusiness,buttheyoftenhavenoideaofwheretheserulescomefromorwhytheywork,andthisisreallywhateconomiststendto ndinteresting.Inmanycasesyourideascancomefromyourownlifeandexperiences.OneofmyfavoritepiecesofmyownworkisthepaperIwroteon\AModelofSales".IhaddecidedtogetanewTVsoIfollowedtheadsinthenewspapertogetanideaofhowmuchitwouldcost.Inoticedthattheprices uctuatedquiteabitfromweektoweek.Itoccurredtomethatthechallengetoeconomicswasnotwhythepricesweresometimeslow(i.e.,duringsales)butwhytheywereeverhigh.Whowouldbesofoolishastobuywhenthepricewashighsinceeveryoneknewthattheitemwouldbeonsaleinafewweeks?Buttheremustbesuchpeople,otherwisethestoreswouldnever nditpro tabletochargeahighprice.Armedwiththisinsight,Iwasabletogenerateamodelofsales.Inmymodelthereweretwokindsofconsumers:informedconsumerswhoreadtheadsanduninformedconsumerswhodidn'treadtheads.Thestoreshadsalesinordertopricediscriminatebetweentheinformedanduninformedconsumers.OnceIdevelopedthemodelIhadaresearchassistantgothroughacoupleofyears'worthoftheAnnArborNewssearchingforthepricesofcolorTVs.Muchtomydelightthegeneralpatternofpricingwassimilartothatpredictedbythemodel.And,yes,IdidmanagetogetaprettygooddealontheTVIeventuallybought.1.2Isyourideaworthpursuing?Solet'sassume(afavoritewordofeconomists)thatyouhaveanidea.Howdoyouknowifitisanygood?The rsttestistotrytophraseyourideainawaythatanon-economistcanunderstand.Ifyoucan'tdothis1Butwhichsourcestoread?IreadtheNewYorkTimes,theWallStreetJournalandtheEconomist;theseareprobablygoodplacestostart. BUILDINGYOURMODEL3it'sprobablynotaverygoodidea.Ifyoucanphraseitinawaythatanoneconomistcanunderstand,itstillmaybealousyidea,butatleastthere'shope.Beforeyoustarttryingtodecidewhetheryourideaiscorrect,youshouldstoptoaskwhetheritisinteresting.Ifitisn'tinteresting,noonewillcarewhetheritiscorrectornot.Sotryitoutonafewpeople|seeiftheythinkthatitisworthpursuing.Whatwouldfollowfromthisideaifitiscorrect?Wouldithavelotsofimplicationsorwoulditjustbeadeadend?Alwaysrememberthatworkingonthisparticularideahasanopportunitycost|youcouldbespendingyourtimeworkingonadi erentidea.Makesurethattheexpectedbene tscoverthatopportunitycost.Oneoftheprimarypurposesofeconomictheoryistogenerateinsight.Thegreatestcomplementis\Ah!Sothatexplainsit!"That'swhatyoushouldbelookingfor|forgetaboutthe\nicesolidwork"andtrytobecomeaWizardofAhs.1.3Don'tlookattheliteraturetoosoonThe rstthingthatmostgraduatestudentsdoistheyrushtotheliteraturetoseeifsomeoneelsehadthisideaalready.However,myadviceistowaitabitbeforeyoulookattheliterature.Eventuallyyoushoulddoathoroughliteraturereview,ofcourse,butIthinkthatyouwilldomuchbetterifyouworkonyourideaforafewweeksbeforedoingasystematicliteraturesearch.Thereareseveralreasonsfordelay.First,youneedthepracticeofdevelopingamodel.Evenifyouendupreproducingexactlysomethingthatisintheliteraturealreadyyouwillhavelearnedalotbydoingit|andyoucanfeelawfullygoodaboutyourselffordevelopingapublishableidea!(Evenifyoudidn'tgettopublishityourself:::)Second,youmightcomeupwithadi erentapproachthanisfoundintheliterature.Ifyoulookatwhatsomeoneelsedidyourthoughtswillbeshapedtoomuchbytheirviews|youaremuchmorelikelytobeoriginalifyouplungerightinandtrytodevelopyourowninsights.Third,yourideasneedtimetoincubate,soyouwanttostartmodelingasearlyaspossible.Whenyoureadwhatothershavedonetheirideascaninteractwithyoursand,hopefully,producesomethingnewandinteresting.1.4BuildingyourmodelSolet'sskiptheliteraturepartfornowandtrytogettothemodeling.Luckyforyou,alleconomicsmodelslookprettymuchthesame.Therearesomeeconomicagents.Theymakechoicesinordertoadvancetheirobjectives.Thechoiceshavetosatisfyvariousconstraintssothere'ssome-thingthatadjuststomakeallthesechoicesconsistent.Thisbasicstructure 4HOWTOBUILDANECONOMICMODELINYOURSPARETIMEsuggestsaplanofattack:Whoarethepeoplemakingthechoices?Whataretheconstraintstheyface?Howdotheyinteract?Whatadjustsifthechoicesaren'tmutuallyconsistent?Askingquestionslikethiscanhelpyoutoidentifythepiecesofamodel.Onceyou'vegotaprettygoodideaofwhatthepieceslooklike,youcanmoveontothenextstage.Moststudentsthinkthatthenextstageistoproveatheoremorrunaregression.No!Thenextstageistoworkanexample.Takethesimplestexample|oneperiod,2goods,2people,linearutility|whateverittakestogettosomethingsimpleenoughtoseewhatisgoingon.Onceyou'vegotanexample,workanotherone,thenanotherone.Seewhatiscommontoyourexamples.Istheresomethinginterestinghappen-inghere?Whenyourexampleshavegivenyouaninklingofwhatisgoingon,thenyoucantrytowritedownamodel.ThecriticaladvicehereisKISS:keepitsimple,stupid.Writedownthesimplestpossiblemodelyoucanthinkof,andseeifitstillexhibitssomeinterestingbehavior.Ifitdoes,thenmakeitevensimpler.SeveralyearsagoIgaveaseminaraboutsomeofmyresearch.Istartedoutwithaverysimpleexample.Oneofthefacultyintheaudienceinter-ruptedmetosaythathehadworkedonsomethinglikethisseveralyearsago,buthismodelwas\muchmorecomplex".Ireplied\MymodelwascomplexwhenIstarted,too,butIjustkeptworkingonittillitgotsimple!"Andthat'swhatyoushoulddo:keepatittillitgetssimple.Thewholepointofamodelistogiveasimpli edrepresentationofreality.Einsteinoncesaid\Everythingshouldbeassimpleaspossible:::butnosimpler."Amodelissupposedtorevealtheessenceofwhatisgoingon:yourmodelshouldbereducedtojustthosepiecesthatarerequiredtomakeitwork.Thistakesasurprisinglylongtime|thereareusuallylotsoffalsestarts,frustratingdiversions,andgeneralfumblingaround.Butkeepatit!Ifitwereeasytodo,itwouldhavealreadybeendone.1.5GeneralizingyourmodelSupposethatyou've nallymadeyourmodelassimpleaspossible.Atthispointyourmodelisprobablytoosimpletobeofmuchinterest:it'slikelyjustanexampleoraspecialcase.Butifyouhavemadeyourmodelassimpleaspossible,itwillnowbemucheasiertoseehowtogeneralizeitsinceyouknowwhatthekeypiecesarethatmakethemodelwork.Hereiswhereyoureducationcanbehelpful.Atlastyoucanuseallthosetechniquesyoulearnedingraduateschool.Mostofthetimeyouwereastudentyouprobablystudiedvariouscanonicalmodels:thingslikeconsumerchoice,andproducerchoice,generalequilibrium,gametheoryandsoon.Theprofessorprobablytoldyouthatthesewereverygeneralmodelsthatcouldencompasslotsofspecialcases. SEARCHINGTHELITERATURE5Well,itwasalltrue.Overthelast ftyyearseconomistshavecomeupwithsomeverygeneralprinciplesandmodels.Mostlikelyyourmodelisaspecialcaseofoneofthesegeneralmodels.Ifsoyoucanimmediatelyapplymanyoftheresultsconcerningthegeneralmodeltoyourspecialcase,andallthattechniqueyoulearnedcanhelpyouanalyzeyourmodel.1.6MakingmistakesThisprocess|simplifytogettheresult,complexifytoseehowgeneralitis|isagoodwaytounderstandyourmodel.MostofthetimethatIspendmodelingisinvolvedinthisback-and-forthprocess.Alongtheway,Imakealotofmistakes.AsPietHeinputsit:Theroadtowisdom?Wellit'splainandsimpletoexpress:Erranderranderragainbutlessandlessandless.Thisback-and-forthiterationinbuildingamodelislikesculpting:youarechippingawayalittlebithere,andalittlebitthere,hopingto ndwhat'sreallyinsidethatstubbornblockofmarble.Ichoosetheanalogywithsculptingpurposely:likesculpturemostoftheworkinbuildingamodeldoesn'tconsistofaddingthings,itconsistsofsubtractingthem.Thisisthemostfunpartofmodeling,anditcanbeveryexcitingwhentheformoftheideareallybeginstotakeshape.Inormallywalkaroundinabitofadazeatthisstage;andItrynottogettoofarawayfromayellowpad.Eventually,ifyou'relucky,theinnerworkingsofyourmodelwillrevealitself:you'llseethesimplecoreofwhat'sgoingonandyou'llalsounderstandhowgeneralthephenomenonreallyis.1.7SearchingtheliteratureAtthispointyoucanstartdoingyourliteraturesearch.Tellyourprofessorsaboutwhatyou'vediscovered|ninetimesoutoftenthey'lltellyoutolookinthe\1983AER"or\Econometrica77"orsometextbook(maybeeven 6HOWTOBUILDANECONOMICMODELINYOURSPARETIMEoneofmine).Andlotsofthetimethey'llberight.You'lllookthereand nd\your"model|butitwillbemuchbetterdone,muchmorefullydeveloped,andmuchclearer.Hey,noonesaidresearchwouldbeeasy.Butthisisapointwhereyoureallyhaveachancetolearnsomething|readthearticle(s)carefullyandaskyourself\Whydidn'tIdothat?"Ifsomeonestartedwiththesameideaasyouandcarrieditfurther,youwanttoseewhatyoumissed.Ontheotherhand,ifyoureallyfollowedtheadviceIgaveyouabovetokeepitsimple,youmayhavecomeupwithsomethingthatismuchclearerthanthecurrenttreatments.Or,maybeyou'vefoundsomethingthatismoregeneral.Ifso,youmayhaveaworthwhileinsight.Gobacktoyouradvisorandtellhimorherwhatyouhavefound.Maybeyou'vegotanewangleonanoldideathatisworthfurtherexploration.Ifso,congratulations|youwouldneverhavefoundthisifyoudidtheliteraturesearchrightaway.Maybewhatyou've guredoutisnotalreadyintheliterature.Thenextpossibilityisthatyouarewrong.Maybeyouranalysisisn'tright,maybetheideaisjusto thewall.Thisiswhereyouradvisorcanplayabigrole.Ifyou'vereallymadeyouranalysisassimpleaspossible,itisa)lesslikelytocontainanerror,andb)anyerrorsthatremainwillbeeasierto nd.Thisbringsmetoanothercommonproblem.Whenyou'veworkedonatopicforseveralmonths|orevenseveralweeks|youtendtolosealotofperspective:::literally.You'rejusttooclosetotheworktoreallygetapictureofwhatisgoingon.Thislackofperspectivetakesoneoftwoforms: rst,youmaythinksomethingisobviouswhenitreallyisn't.Itmaybeobvioustoyou,butyou'vebeenthinkingaboutthisissueforseveralmonths|itprobablyisn'tsoobvioustosomeonewhodoesn'thavethebene tofthatexperience.Theotherpossibilityisthatyoumaythinksomethingiscomplicatedwhenitisreallyobvious|you'vewanderedintoaforestviaameanderingpath.Maybethere'sanicecleartrailjustafewfeetawaythatyou'vetotallymissed.Soatthispointyou'vegottostartgettingsomeindependentjudgmentofyourwork.Talktoyouradvisor,talktoyourfellowstudents,talktoyourwife,husband,girlfriend,boyfriend,neighbor,orpet:::whoeveryoucangettolisten.Andhere'swhatyou'll nd:they'vegotnoideaofwhatyouaretalkingabout(especiallyyourpet).Soyouhavetogobacktotryingto gureoutwhatyoureallyaretalkingabout:whatisthefundamentalideaofyourmodel?1.8GivingaseminarAfteryou'veboredyourfriends,relativesandpetstodeath,youshouldgiveaseminar.Thisisareallyimportantphase:themoreyoucantalk GIVINGASEMINAR7aboutyourwork,thebetterthe nalpaperwillbe.Thisisbecauseatalkforcesyoutogettothepoint.Ifyouwantyouraudiencetolistentoyou,you'vegottomakeyourideaclear,concise,andorganized|andtheexperiencethatyougainbydoingthisisextremelyusefulforwritingyourpaper.Ilistentoalotofstupidideas|butthat'swhatI'mpaidtodo.Lotsofpeoplelistentostupidideasfromme,too:mycolleaguesgetpaidtodoit,andthestudentsgetexaminedonit.Butmostpeopledon'thavetolistentoyou.Theydon'thavetoreadyourpaper.Theywon'tevenhavetoglanceattheabstractunlesstheyhaveareasonto.Thiscomesasabigshocktomostgraduatestudents.Theythinkthatjustbecausethey'veputalotofworkandalotofthoughtintotheirpaperthattherestoftheworldisobligedtopayattentiontothem.Alas,itisn'tso.HerbSimononcesaidthatthefundamentalscarcityinthemodernworldwasscarcityofattention|andbrother,isthatthetruth.Therearedemandsforeverybody'sattention,andifyouwantsomeonetopayattentiontoyou,youhavetogivethemareasontodoso.Aseminarisawaytogetthemtopayattention,sobesuretoexploitthisopportunitytogetpeopletolistentoyou.Theusefulthingaboutaseminaristhatyougetimmediatefeedbackfromtheaudience.Anaudiencewon'tputupwithalotofthethingsthatauthorstrytowriteinpapers:turgidprose,complexnotation,andtediousdetails.And,believeitornot,readerswon'tputupwiththesethingseither!Thetrickistousetheseminartogetallthosethingsoutofyourpaper|thatway,itmayactuallygetread.ControllingtheaudienceI'veseenitclaimedthatoneofthegreatestfearsthatmostpeoplehaveisspeakingbeforeagroup.Iimaginethatmostassistantprofessorshavethisproblem,butaftermanyyearsofgivinglecturesbeforeseveralhundredstudentsitgoesaway.Infact,lecturingcanbecomedownrightaddictive(asmyfamilyoftenremindsme.)AsthemathematicianR.H.Bingoncesaid:\WhenIwasyoung,Iwouldrathergivealectureonmathematicsthanlistentoone.NowthatIamolderandmorematureIwouldrathergivetwolecturesonmathematicsthanlistentoone."Givinglecturesisabitlikeeatingoysters.Your rstonerequiressomecourage,butafteryoudevelopatasteforthem,itcanbehardtostop.Therearethreepartstoaseminar:theintroduction,thecontent,andtheconclusion.Myadviceaboutintroductionsissimple:don'thaveone.Ihaveseenmanyseminarsruinedbylong,pretentious,contentlessintroduc-tions.Justsayafewsentencesaboutthebigpictureandthengetdowntobusiness:showthemwhatyou'vegotandwhyit'simportant.Theprimary 8HOWTOBUILDANECONOMICMODELINYOURSPARETIMEreasontogetdowntobusinessrightawayisthatyouraudiencewillonlyrememberabouttwentyminutesofyourtalk|andthatisusuallythe rsttwentyminutes.Somakesurethatyougetsomeusefulinformationintothat rsttwentyminutes.Asforconclusions,themostcommonproblemislettingtheseminartrailo intosilence.Thiscanruinagoodtalk.IalwaysliketospendthelastcoupleofminutessummarizingwhatIaccomplishedandwhytheaudienceshouldcare.Afterall,thisiswhattheywillwalkawaywith,soyoumightaswelltellthemwhattheyshouldrememberratherthanmakethem gurethisoutforthemselves.Nowadayseveryoneseemstouseoverheadsfortheirlectures.Thedown-sideofthisisthattheseminarisn'tveryspontaneous|buttheupsideisthattheseminarisusuallybetterorganized.Myadviceistolimityourselftooneortwoslidesforaintroductionandoneforaconclusion.Thatwayyouwillbeforcedtogettoyourcontributionsoonerratherthanlater.Andmakeyouroverheadsbig;uselargetypeanddon'ttrytosaytoomuchoneachone.Therearetwothingstoavoidinyourpresentation:don'tletyouraudi-encegotosleep,anddon'tletthemgettoolively.Youwanttheaudiencetohearwhatyouhavetosay.Theywon'thearyourmessageiftheyaresleeping,andtheywon'thearyourmessageiftheyaretalkingmorethanyouare.Sodon'tloosecontrolofyourseminar!Thekeytomaintainingcontrolistoestablishcredibilityearlyon.Thewaytodoistogointogreatdetailinthepresentationofyour rstresult|atheorem,aregression,adiagram,whatever.Spellouteachaspectofyourresultinexcruciatingdetailsonoonecanpossiblymisunderstand.Whenyoudothisyouwillcertainlygetquestionslike\Willthisgeneralizetonagents?"or\Haveyoucorrectedforheteroskedasticity?"Ifyouknowtheanswertothequestion,goaheadandanswerit.Ifyoudon'tknowtheanswer|orthequestionsistotallyo thewall|say\That'sagoodquestion;letmecomebacktothatattheendoftheseminar."(Ofcourseyouneverwill.)Don'tgetsidetracked:thepointofgoingthroughtheinitialresultingreatdetailistoestablishcredibility.Onceyou'vepresentedyourresultandyouseethattheaudiencehasunderstoodthepoint|theirheadsarenoddingbutnotnoddingo |youcangoontothegeneralizationsandelaborations.Ifyou'vedoneagoodjobatestablishingyourcredibilityinitiallynowtheaudiencewillbelieveanythingyousay!Ofcourseyoushouldn'tabusethistrust,butitisusefultoexploititintherestofyourpresentation.Thisisthefundamentalreasonforstartingsimple:ifyoustartoutwithadelicateargument,itwillbehardfortheaudiencetounderstandandyouwillneverestablishtrust.Whenyouaredonewithyourtalkyoushouldtakeafewminutestojotdownsomenotes:whatwasdicultforpeopletounderstand?Whatquestionsdidtheyask?Whatsuggestionsdidtheymake?Whatreferencesdidtheygiveyou?Youmaythinkthatyouwillrememberthesepoints,but PLANNINGYOURPAPER9quiteoftenyouwon't.Theaudienceisaveryusefulresourceforclarifyingyourthoughts|makesureyouuseitwell.1.9PlanningyourpaperAlmosteveryonewritesoncomputersthesedays.Iknowthatcomputersaregreattimesavers:IgetalmostasmuchworkdonenowasIgotdonebeforeIstartedusingcomputers.2IthoughtthatIwouldspendabitoftimetalkingabouthowIusecomputers,notbecauseitisallthatimportant,butbecausenooneelseeverdiscussessuchmundanematters.SinceIamwellknownasacomputernerd,peoplealwaysaskmewhatIuse,andI gureIcansavetimebypointingthemtothisarticle.Undoubtedlythiswillalllookincrediblyarchaicinafewyears,butthat'sthecostofbeingonthebleedingedgeoftechnology.IcurrentlyuseaUnixmachine,butmostofwhatIsayappliesequallywelltootherenvironments.IhaveadirectoryonmycomputercalledPapersandwhenIstarttoworkonanewtopicIcreateasubdirectoryunderpapers.(Forexample,thispaperisinadirectoryPapers/how-I-work.)WhenIcreatethedirectoryIcreateanotes.txt le:thiscontainsmyinitialideas,aroughoutline,whatever.Forexample,thenotes.txt leforthispaperinitiallyhadentriesinitlike:*readthenewspaper*simplify*writeandtalk**ifyoudon'tgrabtheminthe rstpage,theywon'treaditIcreateanotes lelikethiswhenI rststarttoworkonatopic|IjotdowntheinitialideasIhave,whichareusuallyprettysketchy.InthefollowingdaysandweeksIoccasionallytakealookatthisoutline.WhenIlookatitImovethingsaround,addmaterialandsoon.Irarelytakeanythingoutcompletely|Ijustmovematerialtotheendofthe le.Afterall,Imightwantthosenotesagain.AfterorganizingtheseideasforseveralweeksormonthsIamreadytowritethe rstdraftofthepaper.Iusuallytrytodothisinadayortwo,tokeepitallfresh.Inormallyputthenotesinonewindowandthepaper2Ifatrainstopsatatrainstation,whatdoyouthinkhappensataworkstation? 10HOWTOBUILDANECONOMICMODELINYOURSPARETIMEintheotherandwritethepaperwhileIreferbackandupdatethenotestokeeptheminsyncwiththepaper.OncethepaperiswrittenIputitasideforacoupleofweeks.Papersneedtoagelike necheese|it'struethatmoldmightdevelop,butthe avorisoftenenhanced.Moreimportantly,itgivesyoursubconsciousmindachancetoworkontheidea|maybeitwillcomeupwithsomethingyourconsciousmindhasmissed.WhenIcomebacktothepaperItrytoreaditwithafreshmind,likesomeonewhohasneverseenitbefore.3OnrareoccasionsIlikewhatIread,butusuallyIhavelotsofcriticisms.WheneverIhavetopauseandthink\whatdoesthatmean?"Irewrite|Iaddmoreexplanation,changethenotation,orwhateverisnecessarytomakethepaperclearer.WhenI'mdonewiththisprocessIhavea rstdraft.Inextcheckthisdraftintoarevisioncontrolsystem.Thisisapieceofsoftwarethatkeepstrackoftherevisionsofapaper.Itdocumentsallofthechangesyoumakeandallowsyoutorestoreanypreviousversionofapaper.IusetheUnixutilityrcsbutIknowtherearemanyothersystemsavailable.Revisioncontrolsystemsareespeciallyvaluableifyouareworkingwithacoauthorsincetheykeeptrackofwhichpersonmadewhichchangeswhen.Ithenrepeattheprocess:letthepapersitforafewmoreweeksormonths,thencomebacktoit,readitwithafreshmindandreviseitaccordingly.Itisparticularlyusefultodoarevisionrightafteryougiveaseminar.Re-memberthosenotesItoldyoutowriteafteryourseminarended?Sitdownwiththepaperandgooverthequestionstheaudiencehadandthesugges-tionstheymade.Canyouanswertheirquestionsinyourpaper?Canyouincorporatetheirsuggestions?Besuretomodifythenotes/outline/slidesforyourtalkwhenyouincorporatetheaudience'ssuggestions.BibliographicsoftwareOneveryusefulcomputertoolisabibliographicsystem.Thisisapieceofsoftwaredesignedtomanagedalistofreferences.Thereisamasterdatabaseofreferencesthatisstoredonyourcomputer.YouassignakeytoeacharticlelikeArrow70orArrowRisk.Whenyouwanttorefertoapaperyouusethekey,bysayingsomethinglikencitefArrow70g.Thebib-liographicprogramthenlooksuptheappropriatecitationinyourdatabaseandputsitinthelistofreferencesattheendofyourarticle.IusethesystemcalledBibTEX,sinceitworkswellwithTEX.However,therearemanyothersystemsavailablethatworkforotherwordprocessing3Thisismucheasieronceyoureachmiddleage. WHENTOSTOP11packages.It'sagoodideatogetinthehabitofusingasystemlikethis.Overtheyearsyouwillbuildupacomprehensivebibliographyfortheareasyouworkin.Butwheredoyougetyourreferencesinthe rstplace?Well,onewayistoaskpeople:youradviser,yourcolleagues,yourfriends,andsoon.Thisisstilloneofthemostreliableways.ButnowadaysthereareanumberofcomputerizeddatabasesavailableonlineoronCDsthatallowforeasysearch.YoucanopentheCDfortheJournalofEconomicLiterature,typeinafewkeywordslike\pricediscrimination"andgetthelast10years'worthofabstractsofpublishedarticlesthatcontainthewords\pricediscrimination."Asyoulookatthesearticlesyouwillseeafew\classic"articlescited.WhenyouidentifytheseclassicarticlesgototheSocialScienceCitationIndexandsearchforalltherecentpapersthathavecitedtheseclassics.Thisprocessshouldgiveyouanup-to-datebibliographyprettyquickly.Oftenyoucandownloadthecitationsyougetdirectlyintoyourbibliographydatabaseprogram.1.10ThestructureofthepaperThere'sanoldjokeaboutacademicpapers.Theyareallsupposedtohavethreeparts.The rstpart,everyonecanunderstand.Thesecond,onlyahandfulofreaderscanunderstand.Thelastpartnoonecanunderstand|that'showthereadersknowit'saseriouspieceofwork!Thebigmistakethatauthorsmakethesedaysistoleaveoutthe rstpartofthepaper|thatpartthateveryonecanunderstand.Buttheintro-ductionisthemostimportantpartofthepaper.You'vegottograbthereaderonthe rstpage.Nomatterhowbrillianttherestofthepaperis,itwon'tbenoticedifnoonereadsit.Andnoonewillreaditifyoudon'tgettheirinterestinthe rstfewparagraphs.Ifyoureallyknowwhatyourpaperisabout,youshouldn't ndithardtoexplainthistoyourreadersinacoupleofparagraphs.Mybasicadviceistomakeyourpaperlooklikeyourtalk.Gettothepoint.Useexamples.Keepitsimple.Tellpeoplewhywhatyoudidisimportantafteryou'vedoneit.Putthetediousstu intheappendix.Endwithasummaryofwhatyouhaveaccomplished.Ifyouhavereallywrittenagoodpaper,peoplewon'thavetolistentoyourseminarto ndoutwhatyouhavedone:theycanjustreaditinyourpaper.1.11WhentostopYoucantellwhenyourworkisgettingreadyforpublicationbythere-actionsintheseminars:peoplestopaskingquestions.(Oratleast,thepeoplewhohavereadyourpaperstopaskingquestions.)Ifyou'vefollowed 12HOWTOBUILDANECONOMICMODELINYOURSPARETIMEmyadvice,you'vealreadyaskedtheirquestions|andansweredthem|inyourpaper.Onceyou'vemadeyourpoint,stop.Lotsofpapersdragontoolong.Isaidearlierthatpeopleonlyrememberabout20minutesofyourseminar(ifyou'relucky),andtheyonlyrememberabout10pagesofyourpaper.Youshouldbeabletosaymostofwhatyouwanttosayinthatlength.Onceyourpaperiswritten,youcansubmitittoajournal.Idon'thavetoomuchtosayaboutthis;DanHamermeshhaswrittenanicearticlethatdescribestheprocedurebetterthanIcan.4AllIcansayistoechohisadvicethatyougooverthearticlewitha netoothcombbeforesendingitin.Nothingturnso aneditororarefereemorethanto ndtypos,missingreferencesandsloppyeditinginthearticlestheydealwith.1.12WritingtextbooksMostofwhatI'vehadtosaysofarhastodowithwritingarticles.ButIsupposeIreallyshouldsayabitabouttheotherkindofwritingI'vedone:textbooks.5My rsttext,MicroeconomicAnalysisreallywasn'tplanned;itjusthap-pened.WhenI rststartedmyprofessionalcareeratMITin1973Iwasaskedtoteachthe rstyeargraduatemicrocourse.Thetext,suchasitwas,consistedofabout20pagesofnoteswrittenbyBobHall,maybe40pagesofnotesfromDanMcFaddenandSidWinter,andafewjournalar-ticles.Thenoteswereawfullysketchy,andthejournalarticlesweremuchtooadvancedfor rstyearstudents.SoIhadtowritemyownnotesforthestudents.The rstyearIwroteabout50pages;thenextyearanother50,andtheyearafterthatanother50.Thestudentswhousedthemweregreat.Theyreadthemcarefullyandtoldmewhatwaswrong:wheretheobscuritieswere,wheretheerrorswere,whatwastooadvanced,andwhatwastoosimple.Iowemuchofthesuccessofthatbooktothefactitwasclasstestedbeforeahighlycriticalaudience.DuringthisperiodIhappenedtomeetRichardHamming,anelectricalengineerwhohadwrittenseveraltexts.Hegavemeakeypieceofadvice:\Gettogethertheproblemsthatyouwantyourstudentstobeabletosolveafterthey'vereadyourbook|andthenwritethebookthatwillteachthemhowtosolvethem."4DanielS.Hamermesh,\TheYoungEconomist'sGuidetoProfessionalEtiquette",JournalofEconomicPerspectives,6:1,169{180.5ThereadermayrecallDisraeli'swarning:\Anauthorwhospeaksabouthisownbooksisalmostasbadasamotherwhotalksaboutherownchildren." WRITINGTEXTBOOKS13Thiswasgreatadvice.Ifollowedittosomedegreewiththegradu-atetext,butlater,whenIwrotetheundergraduatetext,Ifolloweditreligiously|butmoreaboutthatbelow.6Onedayapublishercameintomyoceandasked(astheyoftendo)\Areyouwritingabook?"Isaidthatwouldbeasillythingforanassistantprofessortodo|butasamatteroffact,IdidhavesomeclassnotesthatIhadbeenworkingonforafewyears.NextthingIknew,Ihadseveralpublishersinterestedinmynotes.IspentasemesteratBerkeleyin1977andusedthatopportunitytohammerthemintoshape.Muchtomysurprisethenoteseventuallybecomeabookandendedupbeingverywidelyused.Ididasecondeditionin1983andIshouldhavedonearevisionin1987orso|butinsteadIdecidedtowriteanundergraduatetext.IwantedtowriteanundergraduatebookbecauseIwasfedupwiththebooksIhadbeenusing.Ihadtriedseveraldi erentones,butcouldn't ndanyIreallyliked.IrememberonesemesterIsatdownandtriedtowriteamidtermexam|butthebookIhadbeenusingwassovapidthatIcouldn'tthinkofanyproblemsthatthestudentscouldsolveusingthetoolsthathadbeenpresentedinthebook!AtthatpointI guredIcouldproducesomethingbetter.Aboutthesametimeoneofmyundergraduateshadpickedupawork-bookbyMarciaStigumcalled,Ibelieve,ProblemsinMicroeconomics.Thestudentfoundthisveryhelpfulinunderstandingtheconceptsofeconom-ics,andIrememberedwhatHamminghadtoldmeabouthowtowriteatextbook.SoIaskedmycolleagueTedBergstromifhewouldliketoworkwithmetocreateaseriousworkbook.7Tedcreatedproblemsasthetextwasbeingwritten,andIhadtomakesurethatthetextcontainedeverythingnecessarytosolvetheproblemshecreated.Icreatedproblemstoo,butthosewereautomaticallycoordinatedwiththetextbook|theex-ternalstimulusimposedbyTed'sproblemswasmuchmoreimportantinshapingthecontentofthebook.Ifthestudentsweren'tabletosolvetheproblems,Ihadtoaddexplanationstothetextuntiltheycould|andifwecouldn'tcreateaproblemtoillustratesomepoint,thepointprobablywasn'timportantenoughtoputinthetext.It'sapitythatmostworkbooksarecreatedasafterthoughts.Creatingtheworkbookreallyshouldbeanintegralpartofthewritingprocess,asHammingsuggested.Youwantthestudentstobeabletousethematerial6ThegeneralprinciplethatIfollowed(andstillfollow)withthegraduatetextisthatitshouldgivethestudenttheinformationtheyneedtoknowtoreadamicroeconomicspaperintheAmericanEconomicReview.EverynowandthenIgothroughafewissuesoftheAERandnotetopicsthatshouldgointhenexteditionofthebook.7Asitturnedout,itwasn'tquiteasseriousasIhadexpected|infact,Ithinkthatitisquitefunny,butthatisduetoTed'suniquesenseofhumorratherthanmyintentions. 14HOWTOBUILDANECONOMICMODELINYOURSPARETIMEyouteachthem,sothe rstorderofbusinessisto gureoutwhatitisthatyouwantthemtobeabletodo.Thelatestbuzzwordineducationis\learningbydoing"butasfarasI'mconcernedthat'salwaysbeentheonlywaytogo.Theundergraduatetextturnedouttobeprettysuccessfulaswell.Andtheworkbookhasendedupsellingtwoorthreetimesasmuchasanyofitscompetitors|whichgoestoshowthattherestillisamarketforaqualityproductinthetextbookmarket.1.13SummaryIsaidthateverytalkshouldhaveasummary|soIsupposeIhavetofollowmyownadvice.Herearethepointstotakeaway:Lookforideasintheworld,notinthejournals.Firstmakeyourmodelassimpleaspossible,thengeneralizeit.Lookattheliteraturelater,notsooner.Modelyourpaperafteryourseminar.Stopwhenyou'vemadeyourpoint.Andnowmypointshavebeenmade,soI'mdutyboundtostop.Goforthandmodel!