量化技術(shù)：10年低點的技術(shù)分散提供了機會

1、February 13, 2019 08:00 PM GMTFebruary 13, 2019 08:00 PM GMTAsia Technology | Asia QUANTECH:TechDispersionat Decade Low Offers OpportunityThesharprallyinhasgeneratedconsiderable inefficiencies.Inthisreport,weexplainreasonsthislow dispersion and identify opportunities to exploit these inconsistencies

2、.Fundamentalstomostin HardwareandJapan.TechdispersionreacheslowlevelpostGFC.Thepositioningrallythe AsiaTechsectorup11%YTD,outperformingtheoverallmarket(MSCIACAsia) by497bpshasresultedstockdispersioncollapsingforthelastsixmonths, decliningto20%atpresentfrom72%sixmonthsago.Thesectorhastradedlike agian

3、tETFYTDandataninflectionpoint,ourview.Thisgenerating inefficienciesthatwecanexploitandwethinkfuturestockperformancewill divergeasthemarketbeginstoreflectactualstockfundamentals.Acrossall techsubsectors,thecurrentdispersionlevelshavefallentoextremelevels,TechHardwareandJapanTechparticular.What is cau

4、sing this low dispersion? Macro issues have dominated stock performanceandweidentifythreereasonsforthelowdispersion:1)fundflow drivenbyETFswithlargecapconstituentsconcentratedtechnology;2)market callingthebottomagainstlowexpectationsDecemberandcheeredby2H recovery;3)shortcoveringsurgefollowinglastDe

5、c.ssell-off.However,low dispersionrarelytendstolastandaspositioningplateauswiththeprobabilityof further macro/trade surprises diminishing, bottom-up stock picking should normalize dispersion. Corporate earnings season has been disappointing and guidanceweak,however,manyinvestorshaveviewedasthetrough

6、ofthe current cycle and a positive for future estimate achievability. We think once activeflowscomebacktothemarket,themarketwillonceagainmakethe distinctionbetweenqualitynamesandstockswithweakeningfundamentals.Apickersmarketbuylaggardstocks,hardwareandJapan Wehavesearchedforstockspecificopportunitie

7、sandfocusonstocks whose share prices appear to have detached from fundamentals. In previous instancesofdispersionlows/troughssuchasOctober2008,May2012,and March2016laggardstockshaveledandoutperformedothertechnamesby 8%the3monthsfollowinganinflectiondispersion.Ourquantscreenof laggard tech stocks wit

8、h improving fundamentals a low dispersion environment, combined with our sector analyst views, suggests investors accumulate Ennoconn, TDK, Rohm and Sharp. Stocks that have re-rated despite weakenedfundamentalsareAirtac,Hiwin,SMIC,WinSKHynix,Japan, TSMC, Mediatek, Asmedia, Aspeed, ASM Pacific and Su

9、nny Optical.MORGAN STANLEY ASIA LIMITED+Shawn KimEQUITY ANALYST HYPERLINK mailto:Shawn.Kim Shawn.Kim+852 3963-1005Yinan Zhang, Ph.D.QUANTITATIVE ANALYST HYPERLINK mailto:Yinan.Zhang Yinan.Zhang+852 3963-3507Daisy DaiRESEARCH ASSOCIATE HYPERLINK mailto:Daisy.Dai Daisy.Dai+852 2848-7310Morgan Stanley

10、appreciates your support in the 2019 Institutional Investor All-Asia Research Team Survey. HYPERLINK /ranking-assistance Request your ballot.S. Korea TechnologyAsia PacificIndustryViewCautiousIndustryViewCautiousMorgan Stanley does and to do business with companiesinMorganStanleyResearch.Asa result,

11、investorsshouldbeawarethatthefirmmayhavea conflict of that could the objectivity of Morgan Stanley Research. should consider MorganStanleyasonlyasinglefactorinmaking their investmentdecision.ForanalystandotherimportantrefertotheSection,locatedattheendofthis report.+=Analystsemployedbynon-U.S.affilia

12、tesarenotregisteredwith FINRA,maynotbeassociatedpersonsofthememberandmaynot besubjectNASD/NYSErestrictionsoncommunicationswitha subjectcompany,publicappearancesandtradingsecuritiesheldby a research analystaccount.QUANTECH Stock ScreensExhibit1: LaggardStockstoOwn1)LatestEPSBreadthBetterthanAverage,2

13、)UnderperformedtheLast3and 6monthsversusMedian,and3)OW/EWRatingTickerNamePricePrice TargetUp/downsideMS RatingFY2 EPS Revision Breadth(3MMA)Latest3M BeforePerf.Percentile6414.TWEnnoconn25438050%OW-3%-32%436762.TTDK889092003%OW-12%22%126770.TAlps Electric237225005%OW-2%5%136971.TKyocera6176730018%OW1

14、%13%456963.TRohm8040960019%OW-11%-8%316753.TSharp12781200-6%EW6%1%16758.TSony4980620024%EW24%28%142409.TWAU Optronics10.9511.55%EW14%19%416740.TJapan Display769018%EW-13%4%1009150.KSSEMCO115500100000-13%EW5%67%246981.TMurata1778015300-14%EW-2%21%486997.TNippon Chemi-Con22952000-13%EW-6%-6%6000977.SZ

15、Inspur19.33203%EW-3%-1%23300017.SZWangsu8.348.52%EW-14%-2%39300136.SZShenzhen Sunway24.7321-15%EW-10%-14%25Source: Morgan Stanley Research. Data as of Feb 13, 2019. Stock performance percentile based on absolute performance.Exhibit2: StockstoAvoid1)LatestEPSBreadthWorsethanAverage,2)OutperformedtheL

16、ast3and6months versusMedian,and3)UW/EWRatingTickerNamePricePrice TargetUp/downsideMS RatingFY2 EPS Revision Breadth(3MMA)Latest3M BeforePerf.Percentile1590.TWAirtac386200-48%UW-34%-44%992049.TWHiwin Tech250180-28%UW-27%-15%740981.HKSMIC8.26-29%UW-18%-4%693105.TWOWIN Semi166.587-48%UW-38%-45%94000660

17、.KSSK Hynix7620063000-17%UW-61%12%716871.TMicronics Japan854600-30%UW-18%-28%743008.TWLargan Precision39902600-35%UW-19%-20%75034220.KSLG DisplaEW-19%-2%632330.TWTSMC229199-13%EW-23%-5%612454.TWMediaTek259236-9%EW-26%-32%765269.TWAsmedia668535-20%EW-15%54%995274.TWOAspeed706566-20%EW-22

18、%-45%812455.TWVPEC6766-1%EW-16%-43%582408.TWNanya Tech60.549-19%EW-31%-29%56005930.KSSamsung Electronics4620040000-13%EW-60%-19%713396.HKLegend Holdings21.752829%EW-25%-2%532357.TWAsustek22726015%EW-34%-32%513231.TWWistron21.15211%EW-16%-29%790522.HKASM Pacific89.3585-5%EW-32%-12%73018260.KSSamsung

19、SDS2245002300002%EW-17%20%85002236.SZDahua15.0114-10%EW-21%-28%68300207.SZSunwoda Electronic C10.889-17%EW-17%-13%952382.HKSunny Optical93.3560-36%EW-21%-36%56Source: Morgan Stanley Research. Data as of Feb 13, 2019. Stock performance percentile based on absolute performance.Quant:UniquePatternandIt

20、sImplicationsforStock PickingWhat is Interesting Now?Stock dispersion measures the variation of cross-sectional stock returns. Stock dispersion measures the variation of cross-sectional stock returns. Dispersion is most commonly applied to assess the potential added value of stock selection capacity

21、. Namely, a market with low stock dispersion makes it difficult for active investors to differentiate their performance from the index portfolio, and vice versa.AsiaTechdispersionhascollapsedinthelast6andhave attheendofJanuaryafterreachingalowlevelpostGFC.Exhibit3 shows Asia Tech stock dispersion ca

22、lculated with trailing 12M returns (relative to benchmark).Overallstockdispersionhasdeclinedto20%atpresentfrom72%six monthsago.Comparedtohistoricalprecedents,decliningdispersionacommontrend afterthemarketcorrectsfromaprolongedbullcycle,asstockstendtofalltogether whenmarketsentimentturnsfromriskontor

23、iskoff.However,themagnitudeofpeak- to-troughdeclinethelastsix-monthcyclestrikinglylarge,evencomparedtothe 2007-08cycle.Exhibit 3: Asia Tech Stock Dispersion: Largest Peak-to-Trough Decline post GFC100%StandardDeviationBandStock Dispersion(12-month)50%25%13%Jan-03Jan-05Jan-07Jan-09Jan-11Jan-13Jan-15J

24、an-17Jan-19Source:ThomsonReuters,Rimes,MorganStanleyResearch.WeeklydataasofFeb8,2019.Herethedispersionismeasuredoveraequally-weighted.samedispersiondefinitionisappliedtotherestofthereport,unlessspecificallymentionedotherwise.have&ininChinaWe demonstrate in Exhibit 4 the stock dispersion by country a

25、nd subsector. First, we divide the entire Asia Tech universe into 8 buckets:Japan(3buckets):TechHardware&Semis,Internet&Software,Consumer ElectronicsChina(2buckets):TechHardware&Semis,Internet&Software (2buckets):TechHardware&Semis,Internet&Software Taiwan(1bucket):TechHardware&SemisNext,wecompareth

26、ecurrentdispersionlevelswithwheretheystood1-yearago, overlaid with its 3-year standard deviation band. The country subsectors where dispersionshavefallenthemostthelastsixmonthsare:TechHardware&SemisChina,Taiwan,andJapan;andInternetandSoftwareChinaandJapan.Wealsonote thatacrossallsubsectors,thecurren

27、tdispersionlevelmateriallybelowitshistorical averagelevel.Exhibit4: Intra-bucketStockCurrentvs.One-yearAgovs.3-yearStandardDeviation Band1StandardDeviationBand(Last3-year)CurrentOne-yearagoCHINA: INTERNET &SOFTWARECHINA: HARDWARE &SEMISTAIWAN: HARDWARE &SEMISJAPAN: INTERNET &SOFTWAREJAPAN: HARDWARE

28、&SEMISKOREA: INTERNET &SOFTWAREKOREA: HARDWARE &SEMISJAPAN: CONSUMER ELECTRONICS5%20%35%50%65%Source: Thomson Reuters, MSCI, Rimes, Morgan Stanley Research. Weekly data as of Feb. 8, 2019.Wecanalsodissecttheoverallstockdispersionintotwocomponents:1)cross-bucketdispersion,whichmeasuresthevariationofc

29、ross-sectionalsectorreturns;and2)intra- bucketstockdispersion,whichmeasuresthevariationofstockreturnswithintheir respectivesectors,asshownExhibit5.Nowweareatanunusualmixwhereboth dispersionshavecollapsedtohistoricallows,similartowhatwehadseenOctober 2008,May2012,andMarch2016.Exhibit5: TotalStockinto

30、Cross-Bucket+Intra-BucketStock 80%Cross-subsectorDispersionIntra-subsector StockDispersion40%20%10%5%3%Jan-03Jan-05Jan-07Jan-09Jan-11Jan-13Jan-15Jan-17Jan-19Source: Thomson Reuters, MSCI, Rimes, Morgan Stanley Research. Weekly data as of Feb 8, 2019.Laggard Stocks Outperform Post Dispersion TroughsI

31、n previous instances of dispersion lows/troughs in October 2008, May 2012, and March 2016, Laggard stocks (defined as stocks that have underperformed their peers in the last 3 months and 6 months) have outperformed the market, while others have underperformed. Exhibit 6 shows that on average, Laggar

32、d stocks have outperformed others by 8% in the 3 months following a dispersion trough.Asaresult,investinglaggardstockswithfundamentalsimprovingalowdispersion environment has potential for outperformance. Please see our Quantamental Stock ScreensforTechHardware&SemisExhibit1andExhibit2.Alsopleaseseet

33、he completelistofAsiaTechstocksandtheirperformancethelast6monthsthe Appendix:HowMuchStocksHaveCorrectedandReboundedtheLast6Months.Exhibit 6: Laggard Stocks Tend to Outperform After Dispersion Troughs1-month2-month1-month2-month3-month10%5%0%-5%LaggardOthersLaggardOthersLaggardOthersPost Oct-2008 Dis

34、persionPost May-2012 DispersionPost Mar-2016 Dispersion TroughSource: Thomson Reuters, MSCI, Rimes, Morgan Stanley Research.Exploiting Dispersion at the Stock LevelAbnormallylowdispersionoffersanintheAsiaTechspace.Thetech sectorsstrongrallyof11%YTD,outperformingMSCIACAsiaby497bpssharp contrastvs.slu

35、ggishperformance2018,whenthesectorfell-21%absoluteterms and underperformed the market by 564bps. The tactical risk-reward favoring EM exposurehasresultedliquiditydynamicshavingabiggerinfluenceontechstocksYTD. Inefficiencies have built as a result and we expect fundamentals to eventually reassert the

36、mselves. In this context, we have searched for stock-specific opportunities within the AsiaTechspacetoidentifywhereinvestorsmayhaveoverlookedfundamentalsa rushtobuytechstocks,whichwereoversoldDecemberlastyear.Wehighlightkey attractivenamesExhibit1,andstockswiththemostdisconnectduetoweakening fundame

37、ntals Exhibit2Asoneofthemostvolatilesectorshistorically,techstockstendtoexhibithigher idiosyncrasy than other sectors. However, stocks have been highly correlated among themselvesthecurrentYTDrallyandthishasgeneratedinefficiencieswecanexploit, as we think future stock performance will normalize, per

38、formance between quality namesandstockswithweakeningfundamentalsshoulddiverge.Theearningsenvironmentstilldifficultandwedoexpectearningsdownwardrevisions togetmuchworsefromhere.Anewroundofupgradeslooksdistantbuttheunderlying trendEPSshouldstarttomovehigheronceweseemoreevidencethattheglobal economicre

39、coverysustainablefrom2H19.Furthermeaningfulupsidefromhere, however,becomesmorechallenginggivenstocksarenowpre-tradingthe2Hrecovery. Asaresult,wewouldfocusonqualitynameswithimprovingfundamentals,andsee risingrisksstocksthatovershotthisrallybutwithoutsolidforwardearnings support.Exhibit7: EarningsBein

40、gDownwardlywhile Share Exhibit 8: Fwd 12M PE Valuation Sharply Rebounded in Jan 19151413121110-40%9 Source: Thomson Reuters, Morgan Stanley ResearchAsiaAvg Source: Thomson Reuters, Morgan Stanley ResearchTechapickersmarketorclosetindexing?TheYTDperformanceoftheAsia TechspacelikeonebigETF;Exhibit3sho

41、wsthattechstockdispersionhasreachedan all-timelow.Dispersionthemeasurementofthevariationindividualstockprices,or broadlyspeakingalpha,whichhasbeenirrelevanttherecentpositioningrally.We observestockshavemovedtandemregardlessofunderlyingfundamentals.Wehaveseenstockstradedflattouponanearningsmissornear

42、-termguidedown.(SKHynixup 6%onearningsmissby12%4Q18;Nidectradedflatonasubstantialearningsmiss; Samsungalsorose2%on4Q18earningmiss).Thisnotsustainableandweexpectthe techsectorwillreturntoastockpickersmarket,dispersiontomovehigherwith more variation of returns explained by idiosyncratic risk and compa

43、ny specifics, where stock selection and fundamental analysis become more important drivers of outperformance.52%55%36%3%BeatMissBeatMissRevNPExhibit9: 4Q18MonitorLargeExhibit10: 52%55%36%3%BeatMissBeatMissRevNP60%50%40%30%20%4%3%2%1%0%-1%3%2%0%-1%0.2%0.3%0.1%0%10%0%-2%BeatIn-lineMissBeatIn-lineMissB

44、eatIn-lineRevOPNP4Q18 3Q18Source: Bloomberg, Morgan Stanley ResearchSource: Bloomberg, Morgan Stanley ResearchWhatiscausingthislowdispersion?Weidentifiedthreereasonsforthelowdispersion, namely 1) macro positioning triggering fund flows driven by ETFs with EM highly concentratedtechstocks;backedby2)t

45、hecallingofacyclicalbottomandoptimism on2Hdemandrecovery;andtriggering3)shortcoveringsurgeafterlastDec.ssell-off.FundflowsbyETFs:FortheweekendedFeb.6,2019,EMequityfunds(ex- ChinaA)reportedlargeinflowsofUS$3.28bn,drivenbyinflowsofUS$3.22bnpassivefunds.EMETF(passive)fundsreportedinflowsofUS$3.22bnandE

46、Mnon- ETF(active)fundsreportedinflowsofUS$0.06bn.Wethinkthelowdispersionof therecentrallymainlydrivenbyfundflowsofETFs,asExhibit12showsthe cumulativenon-ETFflowshaveremainedflat2019,whilecumulativeETFflows surged 2019.Risingbutfundamentalweaknessisreal:TheFederalReservewasable tomeethighexpectations

47、foradovishshift,boostingpricesandsentimentacross assetclasseswithhighbetatechakeybeneficiary.Butagainstrenewedoptimism thererealfundamentalweakness,bothgrowthandearningstrends.InflationarypressureswillalsomakeharderfortheFedtoholditsline.Wewait tomakealargerdirectionalupgradeandlookforatroughearning

48、srevisions (whichcontinuetofall),aclearerbottominggrowthmomentum,orareturnof moreinvestorfeartoturnmorebullishontechandraiseexposure.Headinginto thistechearningsseason,mostcompaniestalkedabouta2Hrecovery(MU, Samsung,TSMC,WDC,LRCX,ASML,SKHynix),eventhoughinvestorsareskeptical aboutthegrowthtrajectory

49、,thathardtodisprovethenearterm.ShortafterlastDec.ssell-off:Dec.18wasthesecondworstmonth forAsiaTechstocksperformance2018(-5.5%),duringwhichwesawshort interestratiorisingacrossboard.HeadingintoJan.2019,interestinglywesawstocks tradeflattouponweakearningsandevenwithguidancedowns,asestimatesat buy-side

50、appearedlowerthanthesell-side.Betterthanfearedearningsspurredshortcoveringtrading.Exhibit13andExhibit14bothshowthatshortinterestratio droppedJan.19fromtheDec.peak.Sub-sectorswhichsoldoffmostlastDec. reboundedstrongestJan.2019,asExhibit15suggests,whichloweredthe dispersion of Asias TechsectorDecade-l

51、ow dispersion at an inflection. The case for tactical upside less compellinggiventhestrongrally.Wethinkapolicy-inducedpositioningrallyhaslargely played out, investor expectations have recovered, and tech stocks are no longer tacticallycheap.Themarketmayremainvolatilethenearterm,butwearealso consciou

52、sthatinvestingprimarilyaboutbuyinggoodassetsatattractiveprices.The screensonExhibit1andExhibit2arebasedonquantitativemetricsonlyandshouldbe usedasastartingpointforinvestorslookingforpotentialbuy/sellideas.Inour fundamental picks, we emphasize fundamental upside/downside risk addition to the quantita

53、tive screen overlay. Based on this methodology, we recommend buying Ennoconn,TDK,Sharp,RohmandSharp.LeastprefereedareAirtac,Hiwin,SMIC,Win SK Hynix, Japan, TSMC, Mediatek, Asmedia, Aspeed, ASM PacificandSunnyOptical.Ennoconn(6414.TW,SharonShih):WelikeEnnoconnsclearbusinesspositioningto capture the w

54、orldwide growth opportunity smart manufacturing upgrades. Despite macrovolatility,shouldbeabletocapturethegrowthfromindustrialautomation, industrial networking devices, smart manufacturing upgrade demand within Foxconn group,slotandlotteryprojectwins,andtheseculartrendofbuildingindustrialIoT ecosyst

55、ems.Weestimateits2019revenueatNT$80bn,up20%YoY,andexpectits operatingmargintoimprovefurther,to4.7%.Thestocktradesat16x2019PE,nearthe lowendofitsfive-yearrangeof14-44x.Rohm (6963.T, Kazuo Yoshikawa) - expanding use of power solutions for xEV applicationsfrom2020:Giventhatthelikelihoodofarapidrecovery

56、enddemand lookslow,theprocessofinventorycorrectionandearningsrecoverywillprobablytake time.However,wearestayingwithourOWratingas(1)therenochangethe mid/long-termgrowthstory,thesharepricehasalreadyfallentoP/Bofaround1x, and(3)saleshavemovedintonegativeYoYterritory,anecessarystepforacyclical sharepric

57、erebound.Rohmscurrentsharepriceatattractivelevelsanditsrisk/reward profile hasimproved.Sharp (6753.T, Masahiro Ono): We upgraded the stock to EW from UW back Nov. 2018 give that (1) the share price fell sharply Oct.-Dec. 2018, and with purchasing/retirement of preferred shares underway too, visibili

58、ty on growth numbers ofshareshasincreased;andonreducedcompanyestimatesforN.Americanmaker smartphonesales,estimatesfromtheoutsetdidnotincludetheS/Dplanfornew2018 models,sotherelimitedriskofthisitselfbeingacauseofacutearningsguidance. ValuationsbasedonrecentsharepricescometoP/E7.2-8.7x,EV/EBITDA5.3-5.

59、9xand P/B1.5-1.8x,indicatingthatdilutionriskhasrealisticallyfallencomparedtothepast.Yet, evenweassumefulldilution,P/Eof8xwilllikelyoffersomesupport.WhiletheP/B leveldoeslooksomewhatexpensiveforaLCDpanel-relatedbusiness,Sharpa conglomeratewhichonlyderives30%ofsalesandOPfromLCDbusiness,andoffers ROEof

60、over20%,sowethinkP/Bthe1.5xzonelooksfair.TDK(6762.T,ShojiSato):WereiterateourOWrating,withfourkeyconsiderations:(1) The rechargeable batteries business, which we estimate generated around 70% of 1H consolidatedOP,seemstobeincreasinglycompetitive,itscostcapabilities, adaptabilitytocustomervolumechang