62-Why-Newton-was-wrong

MomentuminfinancialmarketsWhyNewtonwaswrongJan6th2011|fromPRINTEDITIONTheorysaysthatthepastperformanceofsharepricesisnoguidetothefuture.PracticesaysotherwiseWHATgoesupmustcomedown.Itisnaturaltoassumethatthelawofgravityshouldalsoapplyinfinancialmarkets.Afterall,isn'ttheoldestpieceofinvestmentadvicetobuylowandsellhigh?Butin2010Europeaninvestorswouldhaveprosperedbyfollowingadifferentrule.Anyonewhoboughtthebest-performingstocksofthepreviousyearwouldhaveenjoyedreturnsmorethan12percentagepointshigherthansomeonewhobought2009'sworstperformers.Thiswasnotunusual.Sincethe1980sacademicstudieshaverepeatedlyshownthat,onaverage,sharesthathaveperformedwellintherecentpastcontinuetodosoforsometime.Longer-termstudieshaveconfirmedthatthismomentumeffecthasbeenobservableformuchofthepastcentury.Noristhephenomenonconfinedtothestockmarket.Commoditypricesandcurrenciesareremarkablypersistent,risingorfallingforlongperiods.Themomentumeffectdrivesajuggernautthroughoneofthetenetsoffinancetheory,theefficient-markethypothesis.Initsstrongestformthisstatesthatpastpricemovementsshouldgivenousefulinformationaboutthefuture.Investorsshouldhavenologicalreasontohavepreferredthewinnersof2009tothelosers;bothshouldbefairlypricedalready.Marketsdothrowupoccasionalanomalies-forinstance,theoutperformanceofsharesinJanuaryortheirpoorperformanceinthesummermonths-thatmaybetoosmallorunreliabletoexploit.Butthemomentumeffectishuge.ElroyDimson,PaulMarshandMikeStauntonoftheLondonBusinessSchool(LBS)lookedatthelargest100stocksintheBritishmarketsince1900.Theycalculatedthereturnfrombuyingthe20bestperformersoverthepast12monthsandthenholdingthem,rebalancingtheportfolioeverymonth.Thisproducedanannualaverageof10.3percentagepointsmorethanastrategyofbuyingtheprevious12months'worstperformers.Aninvestmentofpound1in1900wouldhavegrownintopound2.3mbytheendof2009;thesamesuminvestedintheloserswouldhaveturnedintojustpound49(seechart1).MessrsDimson,MarshandStauntonappliedasimilarapproachto19marketsacrosstheworldandfoundasignificantmomentumeffectin18ofthem,datingbackto1926inAmericaand1975inlargerEuropeanmarkets.AstudybyAQRCapitalManagement,ahedgefund,foundthattheAmericanstockswiththebestmomentumoutperformedthosewiththeworstbymorethantenpercentagepointsayearbetween1927and2010(seechart2).AQRhassetupaseriesoffundsthatattempttoexploitthemomentumanomaly.Toocostly,toorisky?Eventhehighpriestsofefficient-markettheoryhaveacknowledgedthemomentumeffect.Well-paidfundmanagershavespentdecadestryingtofindwaystobeatthemarket.Butyouhavetowonderwhytheybotherdevotingsomuchmoneyandefforttoresearchingthefortunesofindividualcompanieswhenthemomentumapproachappearstobeeasytoexploitandhasbeenaroundforalongtime.Logicsuggeststhattheeffectshouldbearbitragedaway.Ifthebestperformersofthepast12monthscontinuetodowell,smartinvestorswillbuythemafter11monthshaveelapsed,reducingthereturnsonoffertothosewhowaittheextramonth.Inturn,otherswillbuyaftertenmonths,thennine,eightandsoonuntiltheeffectdisappears.Whenefficient-markettheoristscomeacrossamarketanomaly,theytendtodismissitinoneofthreeways.Thefirstargumentisthattheanomalyisastatisticalquirkobtainedbytorturingthedata;itwillnotpersist.Butthemomentumeffectwasnoticedin1985(byWernerdeBondt,aBelgianeconomistnowatDePaulUniversityinChicago,andRichardThaler,oftheUniversityofChicagoBoothSchoolofBusiness)andhasnotgoneaway.Thesecondisthatanygainsfromthestrategywillbedissipatedinhighertradingcosts.Clearly,theLBSteam'sstrategyofrebalancingaportfolioeverymonthwouldbeexpensivebutMrMarshsaysthesewouldnotoffsetanannualperformancegapofovertenpercentagepoints.Thethirdisthathigherreturnssimplyreflectthehigherrisksofthestrategy.Thishasbeenusedtoexplainawaytwoothernotableanomalies:thesizeandvalueeffects.Smallcompaniestendtodobetterthanbiggeronesinthelongterm,buttheytendtobelessdiversifiedandthereforemorerisky.Andsharesthatlookcheaponconventionalmeasures(assetvalue,dividendyield,price-earningsratio)alsotendtodeliverabove-averagereturns,butbelongtofirmsthatarelikeliertogobust.AccordingtoapaperbyCliffAsness,whoco-foundedAQR,thebetterperformanceofmomentumstocksisnotmerelyareflectionofhigherrisk.Hefindsthatthemomentumeffectpersistedevenwhenthedatawerecontrolledforcompanysizeandvalue(definedasprice-to-book)criteria.Anotherexplanationisneeded.Onepossibilityrelatestotiming.Theefficient-markethypothesisassumesthatnewdevelopmentsareinstantlyassimilatedintoassetprices.However,investorsmaybeslowtoadjusttheiropinionstofreshinformation.Iftheyviewacompanyunfavourably,theymaydismissanimprovementinquarterlyprofitsasablip,ratherthanachangeintrend.Somomentummaysimplyrepresentthelagbetweenbeliefsandthenewreality.Onceatrendisestablished,asharemaybenefitfromabandwagoneffect.Professionalfundmanagershavetoprepareregularreportsforclientsontheprogressoftheirportfolios.Theywillnaturallywanttodemonstratetheirskillsbyowningsharesthathavebeenrisinginpriceandsellingthosethathavebeenfalling.Thiswindow-dressingmayaddtomomentum.PaulWoolleyoftheLondonSchoolofEconomicshassuggestedthatmomentummightresultfromanagencyproblem.Inv