《高盛银行业不可能三角报告英文版-79页-WN7.pdf》由会员分享，可在线阅读，更多相关《高盛银行业不可能三角报告英文版-79页-WN7.pdf（79页珍藏版）》请在taowenge.com淘文阁网|工程机械CAD图纸|机械工程制图|CAD装配图下载|SolidWorks_CaTia_CAD_UG_PROE_设计图分享下载上搜索。
1、We introduce the first report in our 3-part Testing the Impossible Trinity series,which looks at the idea that banks cannot maintain a good balance of provisions,capital and dividends at the same time owing to squeezed earnings.In this report,we assume local gov.debt default risk will be limited as
2、long as debt rollover is permitted and net balances continue to increase,and assess the potential multi-year margin loss of banks on the back of local gov.debt rollover due to lowering rates.We look at nine questions that we think will be top of mind for investors and group these questions under thr
3、ee broad themes.The first theme concerns the local gov.debt exposure on bank balance sheets,in terms of size,mix and distribution.We conclude non-covered banks with 48%local gov.debt to total assets(vs.18%for covered banks)would face more challenges with potential tail risk.Inferring from this,we ex
4、pect six large banks with larger balance sheets to step up and take on more local gov.debt.The second theme is bank earnings risk,due to margin loss on local gov.debt.We use a loss assumption of Rmb 30tn(USD 4.5tn)gross addition of local gov.debt and a 30bps effective rate decrease each year,with ba
5、nking system ROE decreasing by 100bps,and non-covered banks by 150bps each year in 2023E-25E.Based on this,we stress-test that a 60bps rate cut per year on local gov.debt would trigger non-covered banks to face recapitalization risk with ROE declining to 1.7%.Our third theme assesses risk factors of
6、 local gov.debt widening the divergence of individual banks.We highlight two sets of banks to see how these factors affect divergence:1)(downgrade to Sell,from Neutral)and(upgrade to Buy,ABCPSBC from Sell)on the size of local gov.debt,and 2)(downgrade to Sell from Buy)ICBC and (maintain Buy)on the%o
7、f local gov.bonds.CCBFor our investment views on each bank,please see the third report in this series,Testing the Impossible Trinity III:Increasing dividend risk;up to Buy,PSBC ICBC/ABC/Industrial down to Sell.Shuo Yang,Ph.D.+852-2978-0701 |Goldman Sachs(Asia)L.L.C.China Financial Services Testing t
8、he Impossible Trinity Part I:Nine questions around local gov.debt4 July 2023|11:21PM HKTGoldman Sachs does and seeks to do business with companies covered in its research reports.As a result,investors should be aware that the firm may have a conflict of interest that could affect the objectivity of
9、this report.Investors should consider this report as only a single factor in making their investment decision.For Reg AC certification and other important disclosures,see the Disclosure Appendix,or go to employed by non-US affiliates are not registered/qualified as research analysts with FINRA in th
10、e U.S._Nine Questions around local gov.debt on bank balance sheets 1.Whats the size and mix of local gov.debt on bank balance sheets?Based on our economists numbers,we estimate Rmb 34tn(USD 4.9tn)of local gov.debt sits on the balance sheets of covered banks,which account for 61%of banking assets.The
11、re are two elements to local gov.debt,namely loans(including shadow credit)of Rmb 18tn(USD 2.6tn)and bonds of Rmb 16tn(USD 2.3tn),on our estimates.We arrive at this by summing up loans with a tenor of more than 5 years(excluding mortgages),and use the average portfolio share of 48%for local gov.bond
12、s held by banks.We also include shadow credit in local gov.debt and summarize it as part of loans.2.What about the local gov.debt exposure of non-covered banks?We estimate non-covered banks,which account for 39%of banking assets,have Rmb 59tn(USD 8.5tn)of local gov.debt,making up 63%of the total bas
13、ed on GS economists estimates of total local gov.debt.We also estimate that non-covered banks account for 48%of local gov.debt as a%of total IEA,vs.18%for covered banks,suggesting more earnings risk on potential margin loss on local gov.debt given the larger size.Exhibit 1:We assess Rmb 34tn(USD 4.9
14、tn)local gov.debt on balance sheets of covered banks,accouting for 61%of banking assets,with loans(including shadow credit)of Rmb 18tn(USD 2.6tn)and bonds of Rmb 16tn(USD 2.3tn).As of 2022,GSe Exhibit 2:BoCom/Huaxia have largest/smallest local gov.debt exposure(as%of IEA)at 29%/9%,with 16%avg.of cov
15、ered banks.As of 2022,GSe 642211433121219.56.84.63.63.12.21.81.00.80.40.30.2012345678910Rmb tnLocal gov.debt exposure Local gov.loansLocal gov.bondsShadow bankingCovered banksRmb 34tn in totalRmb 16tn bondsRmb 18tn loans&shadow credits29%24%24%20%16%13%13%12%11%11%10%9%Avg.16%0%5%10%15%20%25%30%Loca
16、l gov.debt as%of IEA(2022)Local gov.loan as%of IEALocal gov.bond as%of IEAShadow banking as%of IEASource:Goldman Sachs Global Investment Research,Company dataSource:Goldman Sachs Global Investment Research,Company data4 July 2023 2Goldman SachsChina Financial Services_9XdU9YnW8VyWaQcM8OnPrRpNnOjMrRq
17、PeRoOrP8OnMsOwMnOpMwMmNqR3.Do large banks have larger exposure to local gov.debt?The six large banks(ICBC,CCB,ABC,BOC,BoCom,PSBC)account for Rmb 29tn(USD 4.2tn)of local gov.debt,31%of the estimated system total,based on GS economists estimates.And we assume larger banks can support more local gov.de
18、bt growth due to their sizable balance sheets,with local gov debt as a%of IEA at 18%,compared with 29%for the system.With a 1%increase in exposure,we believe these six banks could support additional local gov.debt of Rmb 1.6tn(USD 0.2tn),potentially offsetting the slowdown in growth of local gov.deb
19、t on non-covered bank balance sheets,as we assess the exposure would be 48%.4.Whats the gross and net addition of local gov.debt each year?We estimate Rmb 30tn(USD 4.5tn)of gross addition of local gov.debt each year to 2025E,with Rmb 20tn(USD 3.0tn)rollover balance and Rmb 10tn(USD 1.5tn)net additio
20、n.Exhibit 3:Covered banks have Rmb 34tn(USD 4.9tn)local gov.debt,18%of total assets,vs.Rmb 59tn(USD 8.5tn)and 48%of total assets for non-covered banks.As of 2022,GSe Exhibit 4:System local gov.debt exposure breakdown As of 2022,GSe 45182748163294345929%18%48%0%10%20%30%40%50%60%020406080100120 Syste
21、mCovered banksNon-covered banksRmb tnLocal gov.debt exposure Local gov.loans+shadow bankingLocal gov.bondsAs%of total assets(RHS)Rmb 16tn,17%Rmb 18tn,19%Rmb 27tn,29%Rmb 32tn,35%System local gov.debt exposure breakdown(2022)Covered banks-bondsCovered banks-loans Non-covered banks-loansNon-covered ban
22、ks-bondsCovered banks:Rmb 34tn,37%of total exposureNon-covered banks:Rmb 59tn,63%of total exposureSource:Goldman Sachs Global Investment Research,Company dataSource:Goldman Sachs Global Investment Research,Company dataExhibit 5:Six large banks total Rmb 29tn(USD 4.2tn)local gov debt,31%of estimated
23、system total.As of 2022,GSe Exhibit 6:We expect large banks can support more local gov.debt growth on their large balance sheet,with local gov.debt as 18%of IEA,vs.29%of system.As of 2022,GSe Non-covered banks,Rmb 59tn,64%Large banks,Rmb 29tn,31%Smaller banks,Rmb 5tn,5%System local gov.debt exposure
24、 breakdown(2022)Non-covered banksLarge banksSmaller banks48%29%18%16%0%10%20%30%40%50%60%Non-covered banksSystemSix large banks(covered)Smaller banks(covered)Local gov.debt as%of total assetsSource:Goldman Sachs Global Investment Research,Company dataSource:Goldman Sachs Global Investment Research,C
25、ompany data4 July 2023 3Goldman SachsChina Financial Services_For the purposes of this analysis,we assume an average 5 year duration of local gov.debt,and unchanged weights of local gov.debt on bank balance sheets.This would mean 20%of local gov.debt rollover and 10%balance growth each year.5.Whats
26、the potential margin loss on local gov.debt?We estimate system ROE will decrease by 100bps each year in 2023E-25E,due to margin loss on local gov.debt,else being equal.Looking at covered banks,we arrive at a 60bps ROE decrease,and non-covered banks at 150bps.We quantify the margin risk of both stock
27、 rollover and increased flows of local gov.debt,assuming default risk is limited as long as rollover is allowed and net balances increase.We expect lowering rates on both stock rollover and increases in local gov.debt flows given the increasing size and difficulty of preventing default risk.We assum
28、e a 1%rate decrease on rollover debt each year and 150 bps spread loss between local gov.debt and the overall bank rate as opportunity cost to model the margin loss of local gov.debt.This implies the effective interest rate on local gov.debt would decrease by 30bps each year on average in 2023E-25E.
29、This would translate into a NIM decrease of covered banks by an average 3-10bps each year in 2023E-25E.We believe ICBC/ABC/Industrial bank would have a larger NIM contraction of 7.3/6.6/9.4bps vs.the 6.0bps average for covered banks each year 2023-25E,while and would have a smaller NIM CMBBOCdecreas
30、e of 4.0/3.3bps,mainly due to the size of local gov.debt.The more local gov.debt held by banks,the more NIM dilution.Based on these assumptions,and with the multi-year margin contraction led by lowering rates of increasing local gov.debt,banking system CET1 ratio could be 9.9%,with that of non-cover
31、ed banks remaining above the minimum requirement at 7.9%by end-2025E,although the CET1 ratio would decrease by 80bps cumulatively in 2023E-25E.Exhibit 7:We estimate Rmb 30/31/34tn(USD 4.2/4.4/4.8tn)gross local gov.debt increase,with Rmb 19/21/23tn(USD 2.7/3.0/3.3tn)of rollover and Rmb 11/10/11tn(USD
32、 1.5/1.4/1.5tn)of net increase in 2023E/24E/25E.Exhibit 8:.compared with new system credit growth addition of Rmb 25/23/25tn(USD 3.5/3.3/3.5tn),assuming annual balance growth of 10%.94105115126192123111011303134020406080100120140 20222023E2024E2025ERmb tnGross increase of local gov.debt(system)Local
33、 gov.debtRolloverNew increaseGross increase2523253031341110110510152025303540 2023E2024E2025ERmb tnGross/net increase of local gov.debt vs.system credit Net increase of system creditGross increase of local gov.debtNet increase of local gov.debtSource:Goldman Sachs Global Investment Research,Company
34、dataSource:Goldman Sachs Global Investment Research,Company data4 July 2023 4Goldman SachsChina Financial Services_6.How low could the rates on local gov.debt go?Our stress-test shows that,if the effective interest rate on local gov.debt decreases by 183bps cumulatively in 2023E-25E,or 60bps each ye
35、ar,non-covered banks would face recapitalization risk with the CET1 ratio dipping below 7.5%,the min.requirement,by end-2025E.And the system ROE would be 6.6%vs.7.4%in our base case,with covered/non-covered banks ROE at 10.3%/1.7%vs.10.8%/3.1%of our base case in 2025E.This suggests that,before reach
36、ing the point of non-covered banks(39%of banking assets)requiring recapitalization:1)the effective interest rate on local gov.debt could go lower,as our stress test suggests 30bps more rate cuts each year or 100bps lower in total for 2023E-25E;or 2)more time could be allowed for debt rollover,assumi
37、ng a fixed setup of rate cuts on local gov.debt in our base case,else being equal.Exhibit 9:We estimate system ROE decrease by 100bps each year in 2023E-25E,due to margin loss on local gov.debt,else being equal.Compared with covered banks which at a 60bps ROE decrease,non-covered banks would decreas
38、e c.150bps.Exhibit 10:The effective rate cut each year would be 30bps,with 1%rate cut for 20%balance rollover and 150bps opportunity cost for 10%new growth,which would translate to a NIM decrease of covered banks by 3-10bps each year in 2023E-25E.11.311.110.87.87.67.43.63.43.10.720.610.551.100.960.8
39、91.571.421.3802468101214 2023E2024E2025E2023E2024E2025E2023E2024E2025E Covered banksSystemNon-covered banks%Margin loss impact to ROE Adj.ROEvs.before adj.350316286256-18-16-18-12-18-11200220240260280300320340360 2022BalancerolloverNewgrowth 2023E BalancerolloverNewgrowth 2024E BalancerolloverNewgro
40、wth2025EbpsLocal gov.debt effective rate trajectory(base case)Source:Goldman Sachs Global Investment Research,Company dataSource:Goldman Sachs Global Investment Research,Company dataExhibit 11:We highlight ICBC/ABC/Industrial bank would have larger NIM contraction of 7.3/6.6/9.4bps vs.6.0bps average
41、 of covered banks each year in 2023-25E,while CMB and BOC would have a smaller NIM decrease of 4.0/3.3bps,mainly due to the size of local gov.debt.Exhibit 12:On our assumptions,with the multi-year margin contraction led by lowering rates of increasing local gov.debts,banking system CET1 ratio could
42、be 9.9%,with that of non-covered banks remaining above the min.requirement at 7.9%by end-2025E,though the CET1 ratio would decrease by 80bps cumulatively in 2023E-25E.-2.9-3.3-4.0-4.2-4.8-5.7-6.6-7.1-7.3-7.4-9.4-9.5Avg.-6.0bps-10.0-9.0-8.0-7.0-6.0-5.0-4.0-3.0-2.0-1.00.0bpsNIM dilution(2023-25E avg.)
43、11.611.511.310.410.19.98.58.27.90.060.110.150.100.180.250.150.280.3902468101214 2023E2024E2025E2023E2024E2025E2023E2024E2025E Covered banksSystemNon-covered banks%Margin loss impact to CET-1 ratio Adj.CET-1 ratiovs.before adj.Source:Goldman Sachs Global Investment Research,Company dataSource:Goldman
44、 Sachs Global Investment Research,Company data4 July 2023 5Goldman SachsChina Financial Services_ 7.What would happen to covered banks if the local gov.debt rate went lower towards the need for recapitilization?If we stress test rates on local gov.debt going lower at 60bps each year in 2023E-25E,we
45、could see covered banks have a NIM decrease of 12bps each year,driving ROE and CET1 ratio decrease by 120/10bps each year,assuming credit cost does not climb up and else being equal.By the end-2025E,the stress tested ROE and CET1 ratio would be 10.3%/11.2%on average,or 52/14bps lower than the base c
46、ase.8.Does the local gov.debt mix matter as an offset to bank earnings?With more bonds in local gov.debt portfolios,and more tax(0%tax rate vs.15%of loans)and capital savings(10%(LGGB)or 20%(LGSB)risk weighting vs.100%of loans),it means banks can better offset the margin loss on local gov.debt.There
47、fore,we believe the%of bonds in local debt portfolios is an important factor in Exhibit 13:We stress-test that with effective interest rate on local gov.debt decreasing by 60bps each year,non-covered banks would face recapitalization risk with the CET1 ratio dipping below 7.5%,the min.requirement by
48、 end-2025E.Exhibit 14:The system ROE would be 6.6%vs.7.4%base case,and covered/non-covered banks ROE would arrive at 10.3%/1.7%vs.10.8%/3.1%of base case in 2025E.11.511.411.210.310.09.78.47.97.52.85%2.25%1.68%-1.0%-0.5%0.0%0.5%1.0%1.5%2.0%2.5%3.0%0.02.04.06.08.010.012.014.0 2023E2024E2025E%Adj.CET-1
49、 ratio(stress case)Covered banksSystemNon-covered banks Required min.Effective rate(RHS)Required min.7.510.810.37.46.63.11.70.02.04.06.08.010.012.0 Base caseStress caseBase caseStress caseBase caseStress case Covered banksSystemNon-covered banks%Adj.ROE in base/stress case(2025E)Source:Goldman Sachs
50、 Global Investment Research,Company data Source:Goldman Sachs Global Investment Research,Company data Exhibit 15:With the stress test to set rates on local gov.debt is to lower at 60bps each year in 2023E-25E vs.base case of 30bps,ROE of covered banks would be 10.3%,or 52bps lower than of base case.