高盛-房地產(chǎn)融資支持的影響

12June2024|9:40PMHKT

ChinaBanks

Thepathforward:Theimpactofpropertyfundingsupport

ShuoYang,Ph.D.

+852-2978-0701|shuo.yang@GoldmanSachs(Asia)L.L.C.

ClaireOuyang

+852-2978-6686|

claire.x.ouyang@

GoldmanSachs(Asia)L.L.C.

Aswatheofnewpropertymeasuresinthepastmonthwilllikelystimulatefurtherlocalgovernmentdebt?nancingbothtofundre?nancingneedsandfacilitatea

reductioninhousinginventory.Consequently,weupdateourFY24-26EPPOP/net

pro?testimatestore?ecthighergovernmentlendinginbankportfolios,whichhasapositiveimpactoncapitalthroughanexpansionoflowerrisk-weightedassets,

reducingRWAdensity,offsetbyafurtherdampeningofmarginsfromloweryieldsandlargerreservesonloweredpropertycollateral.

Inall,ourupdatedestimatesre?ect:

1.AnongoingreductioninNIMto1.4%in2026(from1.7%in2023),re?ecting

higherlocalgovernmentdebtgrowthof13%forourcoveredbanks,3pptshigherthanourpriorestimates(seeTrinityI).WeexpectlargeSOEbankstotakea

disproportionateshareofthisincrementallending.

2.AhigherbalanceofRmb4.5tninpropertyloanstofurtherfacilitatethe

reductioninhousinginventory.Weraisereservestore?ectloweredpropertycollateralvaluesandestimateaRmb0.3tnlossontheRmb4.5tninnew

propertyloans.

3.AnannualcapitalreleaseofRmb0.7tnforourcoveredbanks,withapositive18bpsimpacttotheCET1ratioeachyearthrough2026.

_

WeestimatethatourcoveredbankswillrequireRmb0.3tninreplenishmentcapitaltomaintaindividends,whichhassofarlargelybeenfacilitatedviaconvertiblebonds(60%oftotalnewcapitalraisinginthepast?veyears).AcapitalscenarioanalysisofRWAdensityforlargeSOEbanksvs.MUFG(coveredbyJapanbanksanalyst

MakotoKuroda)suggeststhereisstillsigni?cantroomtoraisebalancesheet

leverageby3turns,assumingRWAgrowthathalfthepaceofassetgrowthof13%pa(to17xthrough2033).

BanksharepricesremainvolatileatthelowerendofP/PPOPhistoricalranges,andsensitivetonear-termshiftsinmacroeasing.WeresetourtargetP/PPOPmultiplestothecyclemedian,andcross-checkwithimplieddividendyieldsgiventheongoingprioritizationofsteadydividendstreamsamidstadeclininginterestrate

GoldmanSachsdoesandseekstodobusinesswithcompaniescoveredinitsresearchreports.Asaresult,investorsshouldbeawarethatthe?rmmayhaveacon?ictofinterestthatcouldaffecttheobjectivityofthis

report.Investorsshouldconsiderthisreportasonlyasinglefactorinmakingtheirinvestmentdecision.ForRegACcerti?cationandotherimportantdisclosures,seetheDisclosureAppendix,orgoto

/research/hedge.html.Analystsemployedbynon-USaf?liatesarenotregistered/quali?edasresearchanalystswithFINRAintheU.S.

GoldmanSachsChinaBanks

environment.AssumingDPSismaintainedatcurrentlevels(GSforecast),theTP-implied

dividendyieldis5.5%/7.1%forA/Hshare.However,ifbankswereto?nancetheRmb0.3tncapitalshortfallin2025-26Ewithareductionindividends,thiswouldlowertheTP-implieddividendyieldto2.8/3.6%forA/Hshareson25-26Eavg.,comparedtoa

2.3%10-yrgovernmentbondand1.5%1-yrtimedeposit.Accordingly,wetweakourFY24-26EPPOP/netpro?testimatesslightly,withournewTPsimplying+1%/0%

upsideforA/Hsharesonaverage.Ourratingsareunchanged.WehighlightBuy-ratedBONBandCCB,Sell-ratedICBCandABC,andNeutral-ratedCMBgivenfullvaluation.

Exhibit1:AssumingDPSismaintainedatcurrentlevels(GS

forecast),TPimplieddividendyieldis5.5%/7.1%forA/Hshare.

However,ifbankswereto?nancetheRmb0.3tncapitalshortfallin2025-26Ewithareductionindividend,thiswouldlowerTPimplieddividendyieldto2.8%/3.6%forA/Hshareon25-26Eavg.,comparedwitha2.3%10-yeargovernmentbondanda1.5%1-yeartime

deposit

Exhibit2:ThespreadbetweenlargeSOEbankdividendyieldand10-yeargovernmentbondyieldrangesfrom1.5%to6.3%

8%7%6%5%4%3%2%1%0%

7.1%

5.5%

3.6%

2.8%

H-shareA-shareH-shareA-share

Implieddiv.yieldassumingDPSconstant(25-26avg.)

Implieddiv.yieldassumingdividendcuttofundcapitalreplenishment(25-26avg.)

2.3%

10yrgovbondyield

1.5%

1yrtime

depositrate

10yrgovbondyield

ImplieddivyieldassumingdividendcuttofundcapitalreplenishmentImplieddivyieldassumingDPSconstant

10%9%8%7%6%5%4%3%2%1%0%

6% targetspread

6.3%max

3.2%

5.1%

1.5%min

median

2.5%

0.1%

2015201620172018201920202021202220232024E2025E2026E

Source:Wind,GoldmanSachsGlobalInvestmentResearchSource:Wind,GoldmanSachsGlobalInvestmentResearch

Part1:Moregrowthinlocalgovernmentdebt

_

InourpreviousTrinityI,wesawlimitedlocalgovernmentdebtdefaultriskifdebtsroll

overbutnotedthatlowerratescouldleadtoNIMlossesforbanks.BankreportedNIMscameinbelowGSe,andweexpectNIMrisktocontinue,asevidencedby1)greater

thanexpectedof?cialgovernmentlow-ratedebtgrowth.Totallocalgovernmentdebt

grewby8%vs.priorinTrinityI,mainlydrivenbyof?cialgovernmentdebtgrowth,whileLGFV,thehigherratepartoflocalgovernmentdebt,stayedlargely?at;2)ourcoveredbanksgainedmoresharefromnon-coveredbanksonlocalgovernmentdebtgrowth

drivenbytheirlargerbalancesheetcapacity,whilenon-coveredbankswilllikelyfaceacapitalshortfallifratesweretomovelower,allelseequal,accordingtoourstresstest.Moreover,weexpectthatiflocalgovernments/SOEsweretostepupthehousing

inventoryreductionprograms(seePart2),moredebtwouldbeaddedtolocal

government/SOE,furtherdrivingratesandNIMlower,whileriskweightingwouldalsomovelowertoreducethecapitalrequirement(seePart3).

12June20242

GoldmanSachsChinaBanks

1.Whathaschangedinlocalgovernmentdebtsizeandmix,andwhataretheimplicationsforbanks?

OureconomistsestimateRmb102tninonshorelocalgovernmentdebt,+8%vs.theirpriorestimatesofanetincreaseofRmb6tninof?cialgovernmentdebt,andanLGFVdebtbalanceofRmb2tn.Thismeanslowerratesonlocalgovernmentdebtgivena

higherproportionofof?cialgovernmentdebt,whichhasalowerrate.

SinceourTrinityreports(I,II,III),thereportedNIMofcoveredbankshasfallenby

-23bpsonaveragevs.GSeof-19bps.Weexpectafurtherdeclineinratesonlocalgovernmentdebtonalikelylargerincreaseofof?cialgovernmentdebttodrivetotallocalgovernmentdebtgrowth.

Hence,weestimatea~30bpsratedecreaseinlocalgovernmentdebteachyear,and

togetherwithrepricingofotherinterest-bearingassetsandliabilities,weexpectNIMforourcoveredbankstofurtherdecreaseby16/7/1bpin24E/25E/26E.

Exhibit3:WeexpectNIMofourcoveredbankstofurtherdecreasein2024E/25E/26E

Exhibit4:OureconomistsestimateRmb102tninonshorelocal

governmentdebt(+8%vs.theirpriorestimate)onanetincreaseofRmb6tninof?cialgovernmentdebtandanLGFVdebtbalanceofRmb2tn

Rmbtn

2.2

2.0

1.8

%

1.6

1.4

1.2

1.0

Coveredbanksavg.

2.07

2.01

1.87

202020212022

NIM

1.68

1.64

2023

"trinity"estimate

120

100

80

1.63

Rmbtn

1.59

60

40

1.47

1.39

1.40

20

0

2025E

2026E

2024E

OfficiallocalgovernmentdebtImplicitlocalgovernmentdebt

102

yoy:8%

94

62

60

41

35

New

Old

Source:GoldmanSachsGlobalInvestmentResearchSource:GoldmanSachsGlobalInvestmentResearch

2.Whathaschangedinlocalgovernmentdebtownedbybanks?

Morelocalgovernmentdebthasbeentakenonbybanksunderourcoverage,especiallylargeSOEbanks.WeupdateourlocalgovernmentdebtestimationfromthebanklevelandnotethatourcoveredbanksgrewlocalgovernmentdebtbyRmb4.4tn(up13%vs.ourpriorestimate)whilenon-coveredbanksgrewtheirlocalgovernmentdebtbyRmb3.0tn(up6%vs.prior).

nSixlargeSOEbanksgrewtheirlocalgovernmentdebtbyRmb3.8tnvs.prior,

comparedwithourothercoveredbanksthatgrewtheirlocalgovernmentdebtbyRmb0.6tnvs.prior.

nSixlargeSOEbankshavetakenonmorelocalgovernmentdebt,withtheirshareincreasingby1pptvs.priorto33%,whilenon-coveredbanksreducedtheir

exposureby1pptvs.priorto62%.

12June20243

GoldmanSachsChinaBanks

Exhibit5:Banks’totallocalgovernmentdebtgrewbyRmb8tnvs.prior

Rmbtn

Exhibit6:ThesixlargeSOEbankshavetakenonmorelocal

governmentdebt,withtheirshareincreasingby1pptvs.priorto33%,whilenon-coveredbanksreducedtheirexposureby1pptvs.priorto62%

Asof2023

65

55

45

Rmbtn

35

25

15

5

-5

LocalgovernmentdebtAs%oftotal(RHS)

62%

63%

63

33%

59

31%

33

29

5%5%

56

Old

OldNewNoncoveredbanks

New

OldNew

LargeSOEbanks

Smallercoveredbanks

70%60%50%40%30%20%10%0%

-10%

Source:GoldmanSachsGlobalInvestmentResearch

Oldestimatesfrom2022

Source:GoldmanSachsGlobalInvestmentResearch

3.Whatdoesourstresstestonlowerratesonlocalgovernmentdebtsuggest?

A~60bpsratedecreaseinlocalgovernmentdebteachyearwouldseeourcoveredbanksstay‘safe’intermsofcapitalization,butnon-coveredbankswouldrequire

recapitalization,allelseequal.Thismeansthat:

nNIMlossescouldbeacapitalriskfornon-coveredbanks,evenwithoutconsideringcreditlosses.

nMorelocalgovernmentdebtcouldbetakenonbyourcoveredbanks,especiallythelargeSOEbanks,giventheirlargerbalancesheetcapacity.

_

Exhibit7:Ourstresstestsuggeststhatwiththeeffectiverateon

localgovernmentdebtdecreasingby60bpsperyear(vs.our30bpsbasecase),non-coveredbankswouldhaveaCET1ratiobelowtheminimumrequirement

Adj.CET-1ratio(stresscase)

Exhibit8:Morelocalgovernmentdebtcouldbetakenonbyour

coveredbanks,especiallythelargeSOEbanks.Withoutpacing

growthoflocalgovernmentdebt,localgovernmentdebtasa%oftotalassetsforlargeSOEbanksshouldremain?atvs.prior

Asof2023

14.012.0

%

10.08.06.0

4.02.00.0

Coveredbanks

Requiredmin.2.85%

11.9

10.4

8.3

2024E

System

Effectiverate(RHS)

2.26%

Requiredmin.7.5

11.8

10.2

7.9

2025E

Non-coveredbanks

1.68%

11.6

9.9

7.5

2026E

3.0%2.5%2.0%1.5%1.0%0.5%0.0%-0.5%-1.0%

50%45%40%35%30%25%20%15%10%

47%

48%

18%18%

15%

15%

Old

OldNewNoncoveredbanks

New

Smallercoveredbanks

OldNewLargeSOEbanks

Source:GoldmanSachsGlobalInvestmentResearchSource:Companydata,GoldmanSachsGlobalInvestmentResearch

12June20244

GoldmanSachsChinaBanks

4.CanbanksdivergeonNIMsagainstabackdropoflowerratesonlocalgovernmentdebt?

Yes.BankswithashortermaturityofbothassetsandliabilitiesshouldbebetterpositionedandseeNIMsbottomoutsooner.

Thisisbecauseshorterassetmaturitiesmeanlessexposuretolocalgovernmentdebt,whichgenerallyhasalongerduration(weestimatelocalgovernmentloansbyassessingthe5-yearloanportfolioexcludingmortgages).Inaddition,ashorterliabilitymaturity

meansfasterrepricingatlowerrates,whichcansavecosts.

Thesebankswillhavelikelyfactoredinagreaterimpactfrompolicyratecuts(which

doesnotfullyrepricebanks’backbooksimmediatelybutrepricesbankloanswhenthey

arerenewed),andseebetterNIMsvs.otherbankswithalongermaturityandamismatchofassetsandliabilities.

BONBisacaseinpoint:givenitsshorterdurationofassetsandliabilities,weestimateonlya2bpsyoyfallinNIMin24E-26Eonaveragevs.-12bps/-10bpsforICBC/ABC,thetwolargeSOEbanks,whichhavelongerdurationin2024E-26E.

Exhibit9:Bankassetandliabilitydurationvs.NIMchange

Totalloanduration(yrs)Totaldepositduration(yrs)NIM(%)NIMyoy(bps)

2H22

1H23

2H23

yoy

hoh

2H22

1H23

2H23

yoy

hoh

2023

2024E

2025E

2026E

2023

2024E

2025E

2026E

ICBC

4.43

4.33

4.27

-0.16

-0.06

ICBC

0.79

0.87

0.92

0.14

0.06

ICBC

1.57

1.32

1.22

1.21

(32)

(26)

(10)

(1)

CCB

3.69

3.52

3.45

-0.24

-0.06

CCB

0.77

0.83

0.90

0.13

0.07

CCB

1.70

1.51

1.45

1.44

(30)

(19)

(6)

(1)

BOC

3.65

3.52

3.48

-0.17

-0.03

BOC

0.66

0.73

0.81

0.15

0.08

BOC

1.43

1.25

1.18

1.15

(14)

(18)

(7)

(3)

ABC

4.05

3.99

3.94

-0.10

-0.05

ABC

0.77

0.84

0.89

0.11

0.05

ABC

1.54

1.32

1.23

1.23

(32)

(23)

(9)

(0)

BoCom

3.94

3.80

3.70

-0.23

-0.10

BoCom

0.85

0.87

0.99

0.13

0.11

BoCom

1.24

1.14

1.04

1.04

(17)

(10)

(10)

1

PSBC

3.41

3.37

3.34

-0.06

-0.03

PSBC

0.56

0.62

0.54

-0.02

-0.08

PSBC

1.89

1.76

1.69

1.66

(16)

(14)

(7)

(3)

CMB

3.08

3.34

3.00

-0.08

-0.34

CMB

0.56

0.68

0.64

0.09

-0.04

CMB

2.05

1.84

1.77

1.77

(23)

(21)

(7)

0

PAB

2.66

2.60

2.51

-0.15

-0.09

PAB

0.86

0.95

0.97

0.12

0.02

Industrial

1.54

1.45

1.38

1.38

(12)

(9)

(7)

(0)

Industrial

3.69

3.81

3.79

0.10

-0.02

Industrial

0.95

0.95

0.95

0.00

0.00

HuaXia

1.74

1.50

1.48

1.47

(22)

(24)

(2)

(1)

HuaXia

2.53

2.54

2.43

-0.10

-0.11

HuaXia

0.72

0.63

0.69

-0.02

0.06

PAB

2.20

1.77

1.72

1.71

(39)

(43)

(5)

(1)

BONB

2.20

2.19

2.19

-0.01

0.00

BONB

0.83

0.88

0.88

0.05

0.00

BONB

1.62

1.65

1.59

1.57

(11)

3

(6)

(2)

BONJ

2.37

2.38

1.88

-0.48

-0.49

BONJ

1.09

1.15

1.25

0.16

0.10

BONJ

1.18

1.11

1.05

1.05

(25)

(7)

(6)

0

Avg(yrs)

3.31

3.28

3.17

-0.14

-0.12

Avg(yrs)

0.78

0.83

0.87

0.09

0.04

Avg

1.64

1.47

1.40

1.39

(23)

(17)

(7)

(1)

Source:Companydata,GoldmanSachsGlobalInvestmentResearch

_

Part2:Threescenariosfornewpropertyloans

WeestimatetotalpropertyloangrowthofRmb2.2/4.5/6.5tnin2024E-26E,againstthebackdropofahousinginventoryreduction.Ourthreescenariosarebuiltonreplacing

propertybondsandshadowbankingwithloanstorolloverdebtandreducehousinginventory.

nInScenario1,weassume50%replacementtocompleteinthreeyears.ThiswouldrequireRmb2.2tninnewpropertyloans,butrolloverriskwouldremainonshrinkingtotalproperty?nancing.

nInScenario2,weassumefullreplacementtocompleteinthreeyears,whichwouldrequireRmb4.5tninnewpropertyloans.Thisisalikelyscenarioiflocal

governments/SOEsweretostepin,togetherwithpropertydevelopers,totakeonbankloans.

nInScenario3,weassumeafurtherRmb2tninnewpropertyloanstoreduce

12June20245

GoldmanSachsChinaBanks

housinginventory,asperourChinapropertyteam(seehere).TotalnewpropertyloanswouldbeRmb6.5tn.

Therangeofthesenewpropertyloanestimatescouldbeimpactedby1)collateral

availability;and2)whetherlocalgovernments/SOEsstepintore?nancetheproperty

sector.WeestimateaRmb0.3tnlossbasedontheRmb4.5tninnewpropertyloansinScenario2.

1)Scenario1–assumesRmb2.2tninnewpropertyloanstoreplace50%of

propertybondsandshadowbankinginthreeyears:Intermsofnewproperty

?nancing,weviewpropertyloansastheonlyavailablesourcetothepropertysector

frombanks.Thisisbecausewehaveseenfewerpropertybondsandtrustproductsonincreasingcreditlosses,andthesethereforedonotappeartobeservingasasourceofnew?nancing,atleastintheshortterm.InScenario1,weexpectthelossratioon

propertycredittostayhighonshrinkingtotalproperty?nancing.Westresstested

propertycreditlossesbasedonanexpectedcreditloss(ECL)model.MoredetailscanbefoundinourPropertyRiskMonitorreportsI,II,III,IV.

Exhibit10:Scenario1-assumesRmb2.2tninpropertyloansto

replace50%ofpropertybondsandshadowbankinginthreeyears

Exhibit11:...rolloverriskremainsonshrinkingtotalproperty?nancing

PropertybondShadowbankingPropertyloanPropertyloanyoy(RHS)PropertybondShadowbankingPropertyloanTotalfundingyoy(RHS)

4

23%

2.2tn

3

10%

10%

2

4%

4%

5%

Rmbtn

1

1%

0%

0

-1

-2

-3

2018201920202021202220232024E2025E

2%

2026E

25

20

Rmbtn

15

10

5

0

20

20

24%

18

10%

3%

2020

20182019

20

19

-2%

-6%

20212022

18

17

-3%

-7%

20232024E

1616

-3%-2%

2025E2026E

30%25%20%15%10%5%

0%

-5%-10%

_

Source:GoldmanSachsGlobalInvestmentResearch,PBOCSource:GoldmanSachsGlobalInvestmentResearch,PBOC

2)Scenario2–assumesRmb4.5tnofnewpropertyloanstofullyreplacepropertybondsandshadowbankinginthreeyears:Thiscouldresultin?attotalproperty

?nancing,whichmayreducerolloverriskandthustheprobabilityofdefault(PD)of

banks.However,wenotethereisthequestionofcollateralavailabilityonthisscaleof

propertyloangrowth.Assuming20%loantovalue(LTV),collateralrequiredbybanks

couldbeRmb22.5tn,comparedwiththeRmb20tnnetvalueestimatedbyproperty

developers,asperourChinapropertyteam(seehere),andifbanksweretooriginatealloftheseloans,theycouldseelosses,ornewpropertyloanswouldbebelowRmb4.5tnduetoinsuf?cientcollateral.

Thus,tofullyreplacethesedebtswithpropertyloans,inthisscenario,weassumelocalgovernments/SOEswouldneedtostepintotakeonloansandsharelossestogether

withpropertydevelopers.

12June20246

GoldmanSachsChinaBanks

Wehighlightthat1)ourestimateisfornewpropertyloanstogrowgraduallyas

necessaryfundingtothepropertysector,withatotalofRmb4.5tninnewproperty

loansbyend-2026,comparedwiththeannualRmb~3-5tnfundingshortfallestimatedbyourChinapropertyteam.Thissuggeststhattailriskremains.

Furthermore,wenowestimateatotalpropertyloanlossofRmb0.4tn(Rmb0.3tnlossontheRmb4.5tnofnewpropertyloans)vs.Rmb0.09tnpriorinourTrinityIIreport.OurlossestimationisbasedontheassumptionthatthenetNPLformationratehasnot

peakedout,andthatmoreNPLsonpropertyloanswouldneedtobetakenonbybanks.TheimpliedNPLratioonnewpropertyloansis~6%,suggestinga10%probabilityof

default(PD)and60%lossgivendefault(LGD).Weexpectthatmostnewpropertyloanscouldbeoriginatedbyourcoveredbanksgiventheirlargerbalancesheetcapacity(seehere).Forillustrativepurposes,withthislosstaking,weestimateourcoveredbanks’

CET1ratiowoulddecreaseby-22bpsonaverage.

Exhibit12:Scenario2-assumesRmb4.5tnofnewpropertyloanstofullyreplacepropertybondsandshadowbankinginthreeyears

Exhibit13:Thiswouldresultin?attotalpropertyfunding,which

mayreducerolloverriskandthustheprobabilityofdefault(PD)forbanks

4

3

2

Rmbtn

1

0

-1

-2

-3

PropertybondShadowbanking

23%

10%

5%

0%

2018201920202021

Propertyloan

4%

1%

20222023

Propertyloanyoy(RHS)

4.5tn

20%

7%

4%

2024E2025E2026E

25

20

Rmbtn

15

10

5

0

PropertybondShadowbankingPropertyloanTotalfundingyoy(RHS)

24%

18

20

20

10%

3%

20

-2%

1918181818

0%0%0%

-3%

-6%

30%25%20%15%10%5%

0%

-5%-10%

2018201920202021202220232024E2025E2026E

Source:GoldmanSachsGlobalInvestmentResearch,PBOC

_

Exhibit14:WeestimatetotalpropertyloanlossesofRmb0.4tn(a

Rmb0.3tnlossontheRmb4.5tnofnewpropertyloans)vs.ourpriorestimateofRmb0.09tninourTrinityIIreport

Source:GoldmanSachsGlobalInvestmentResearch,PBOC

Exhibit15:Forillustrativepurposes,withthislosstaking,we

estimateourcoveredbanks’CET1ratiowoulddecreaseby-22bpsonaverage

0.27%

0.26%

14.0%

13.9%

15%14%13%12%11%10%9%8%7%

Adj.CET-1

0.22%

0.43%

0.24%

13.5%

11.1%

11.5%

ratioLossimpact

0.32%

0.09%

0.19%0.12%

0.18%0.16%0.19%

10.2%

9.3%

9.1%9.1%

9.5%9.4%9.4%

6%5%

Source:GoldmanSachsGlobalInvestmentResearchSource:GoldmanSachsGlobalInvestmentResearch

12June20247

GoldmanSachsChinaBanks

3)Scenario3–assumesRmb6.5tnofnewpropertyloanstorolloverdebtandreducehousinginventorybyRmb2tn(seehere).Inthisscenario,weassumelocalgovernment/SOEstepinmoretotakeoverfrompropertydevelopersfornewloan

growth.Ourpropertyteamestimatelocalgovernment/SOEtakeonRmb2tnofnewpropertyloansata20%loantovalue(LTV),giventheirbalancesheetcapability(Rmb+10tnnetvaluefornewpropertyloans)(seehere).

Wethinkthatwithlocalgovernment/SOEtakingonmoreloansfrombankstore?nancethepropertysector,theprobabilityofdefault(PD)ofbankloansshoulddecreaseto

lowerthelossratio.Thatsaid,otherfactorscouldposeuncertaintiestonewbankloangrowthandprogresstoreducethehousinginventory.Forexample:

nTherequiredreturnoflocalgovernment/SOEonhousinginventoryreductionwouldbe~1.8%-2.5%,representedbytherentalyieldand10-yeargovernmentbondrate.Thehigherthebarofthisreturnrequirement,themorediscountrequiredbylocalgovernments/SOEstotakeoverthere?nancingofpropertydevelopers’assets.

nTheabilityoflocalgovernments/SOEstoleverupandtakeonbankcredit.Local

governmentswithhighleveragemayonlyhavelimitedroomtoborrowmorefrombanks.

InScenario3,wethinkbanks’lossratioscoulddecreaseaslongasdebtisrolledover,andhousinginventoryreductiongoessmoothlywithstablepropertyprices.However,NIMcoulddecreaseduetolowrates,whiletheriskweightingofthesecreditscould

declinetoreducethecapitalrequirement.

Exhibit16:Scenario3-assumesRmb6.5tnofnewpropertyloanstorolloverdebtandreducehousinginventorybyRmb2tn

Exhibit17:Inscenario3,weassumelocalgovernments/SOEs

wouldstepinmoretotakeoverfrompropertydevelopersfornewloangrowth

4

_

3

2

Rmbtn

1

0

-1

-2

-3

PropertybondShadowbanking

23%

10%

5%

0%

2018201920202021

PropertyloanPropertyloanyoy(RHS)

6.5tn

20%19%

4%

4%

1%

202220232024E2025E2026E

25

20

Rmbtn

15

10

5

0

PropertybondShadowbankingPropertyloanTotalfundingyoy(RHS)

20

30%25%20%15%10%5%

0%

-5%-10%

20

20

2020

24%

18

18

19

18

10%

11%

3%

0%

0%

-3%

-2%

-6%

2018201920202021202220232024E2025E2026E

Source:GoldmanSachsGlobalInvestmentResearch,PBOCSource:GoldmanSachsGlobalInvestmentResearch,PBOC

12June20248

GoldmanSachsChinaBanks

Part3:CET1capitalissettorelease

LargeSOEChinabanksaresettoreleaseCET1capitalonnewregulationsandanassetmixchange,withlowerRWAdensityandhigherleverage.Weestimateannualcapital

releaseofRmb0.7/0.4tnforourcovered/largeSOEbanks.Wealsorunacapital

scenarioanalysistoseewhatwouldhappeniflargeSOEbanksconvergewithMUFG

(coveredbyMa