Sovereign Distress, Bank Strength and Performance:

Size: px

Start display at page:

Download "Sovereign Distress, Bank Strength and Performance:"

Shannon Farmer
6 years ago
Views:

Sovereign Distress, Bank Strength and Performance: Evidence from the European Debt Crisis Yifei Cao, Francesc Rodriguez-Tous and Matthew Willison 29

1 Sovereign Distress, Bank Strength and Performance: Evidence from the European Debt Crisis Yifei Cao, Francesc Rodriguez-Tous and Matthew Willison 29 November 2016, Sheffield *The views expressed in this paper are those of the authors and do not necessarily reflect those of the Bank of England or its committees

2 Outline Motivation Literature EBA Sovereign Exposure Data Econometric Method Data & Sample Results Conclusion and Policy Implications 2

3 Motivation Banks are substantially exposed to sovereign debt EBA stress test 2010: in early 2010, the total net position on sovereign debts for the 90 European banks that participated in the test was 750 bn. On average 7% of GIIPS banking assets and 2-3% of Banking assets in core countries The latest EBA stress test result: total net position has increased to 2589 bn., as of December 2015 Zero-risk weight on sovereign exposure Bank capital was not required for investment in sovereign debts Sovereign debts were assumed to be risk-free assets by policy makers However, they are not risk-free 3

4 Motivation European sovereign debt crisis in 2010, followed by a credit crunch and a recession in periphery countries Is there a connection between sovereign crisis and banking crisis? If so what is the transmission channel? Can higher bank capital limit the impact of sovereign shocks? Is that necessary to have a higher capital requirement on bank s sovereign exposure? 4

5 Literature Motivation of the Increase in Sovereing Exposure during the Euopean Debt Crisis Moral Suasion Uhlig GER Becker & Ivashina De Marco & Macchiavelli Ongena et al Carry Trade Crosignani Acharya and Steffen JFE Buch et al JFS Horvath et al

6 Literature Effect of a Sovereign Shock on Bank Lending Theories 1. Capital channel 2. Funding channel Evidence Gennaioli et al. 2014a. JF Gennaioli et al. 2014b Popov and Van Horen RF De Marco Altavilla et al

7 Research Question 1. Effect of Sovereign Shocks on Bank Lending Whether banks sovereign exposure would have a negative effect on bank s lending activities during sovereign distress period? 2. Effect of Bank Strength Whether better capitalised bank would be more resilient to sovereign shocks? Sovereign Distress 3. Decomposition of Sovereign Exposure Sovereign Exposure Bank Strength Bank Performance What kind of sovereign exposure matter? AFS or HTM? Short-term or long-term? Domestic or Foreign? 7

8 EBA Sovereign Exposure Data First multi-notch downgrading, start of the crisis LTRO2, Greek Debt Swap LTRO Mario Draghi Whatever it takes Q1 Q2 Q3 Q4 Q1 Q2 Q3 Q4 Q1 Q2 Q3 Q4 Q1 Q2 Q3 Q4 Q1 Q2 Q3 Q4 Q1 Q2 Q3 Q4 Exposure Date 2010 Q Q Q Q Q Q Q Q Q Q Q4 Exercise Name ST 2010 ST 2011 CX2011 CX2012 TX2013 ST2014 TX2015 ST2016 Availability No Yes Yes Yes Yes Yes Yes Yes No. of Banks Exposure to: 30 European Countries 30 European Countries 30 European Countries 30 European Countries 30 European Countries + HK JP CH US AU CA 30 European Countries + HK JP CH US AU CA CN HR 30 European Countries + HK JP CH US AU CA CN HR 30 European Countries + HK JP CH US AU CA CN HR 8

9 Sovereign Distress, Bank Strength and Performance Example Erste Banking Group, as of 31/12/2010, mln EUR Country Austria Belgium Residual Maturity Gross Long Position Net Position Exposure in Derivatives AFS FVO HFT L&A HTM 3M Y Y Y 1, Y 1,314 1, Y 1,584 1, Y 1,471 1, , Total 5,964 5,899 2, , M Y Y Y Y Y Y Total

10 Sovereign Exposure Categories Category Loan & Advances Held to Maturity Available for Sale Fair Value Option Valuation Book Value Regulatory Treatment Banking Book Notes Book Value Banking Book Banks hold these assets till maturity, rather than trading them for profit. Fair Value Fair Value Banking Book Trading book Banks hold these liquid assets as liquidity reserves ; typically it would be used for large holdings of liquid assets that may need to be sold to provide liquidity for the bank to meet deposit outflows, yet are not actively traded; Interest from AFS assets would be realised through P&L; A change in market value would be realised through a change in AFS reserves, which is part of bank capital. FVO assets may be used for liquid assets that are not actively traded, but act as a first tranche of such assets that could be used to meet outflows; A change in the market value of FVO assets would be realised through P&L; Thus FVO assets are usually treated as trading book assets where market capital requirement is needed. Held for Trading Fair Value Trading Book A change in the market value of HFT assets would be realised through P&L. 10

11 Non GIIPS Banks Exposure to GIIPS Sovereign Debts Category Breakdown, mln euro AFS HFT FVO Derivative Total 11

12 Non GIIPS Banks Exposure to GIIPS Sovereign Debts Maturity Breakdown, mln euro Short-term Exposure Medium-term Exposure Long-term Exposure 12

13 German Banks Total Sovereign Exposure Category Breakdown, mln euro AFS HTM TB Derivative 13

14 Empirical Strategy Sovereign Shock Indicators Downgrading notches Increase in government bonds yield Increase in CDS Spread Exposure to the Sovereign Shocks Expo i,t = Holding Position i,c,t 1 Crisis c,t Total Assets i,t 1 c where Holding Position i,c,t can be decomposed according to its category, maturity and counterparty 14

15 Empirical Strategy Example: Suppose a bank only have sovereign exposure to country A and B Crisis happens in both countries The bank s overall exposure to the sovereign distress: Expo t = Position A,t 1 Crisis A,t + Position B,t 1 Crisis B,t Total Assets t 1 If crisis only happens in country A then: Expo t = Position A,t 1 Crisis A,t Total Assets t 1 15

16 1 Sovereign Risk Project Econometric Method Diff-in-Diff Method Hypotheses 1 st step: 2 nd step: L i,c,t = β 0 + β 1 Strength i,c,t 1 + β 2 Expo i,t 1 +β 3 X i,c,t 1 + β 4 X c,t 1 + γ t + γ c + ε i,c,t L i,c,t = β 0 + β 1 Strength i,c,t 1 + β 2 Expo i,c,t 1 +β 3 Strength i,c,t 1 Expo i,c,t 1 +β 4 X i,c,t 1 + β 5 X c,t 1 + γ t + γ c + ε i,c,t β 2 < 0 β 3 > 0 where Expo = { AFS, HTM ST, MT Domestic } Strength = {CET1 Ratio, Tier1 Capital Ratio } 16

17 Data & Sample Sovereign Exposure Data EBA Disclosure Sovereign Shock Indicator Downgrading Notches (Moody s) CDS Spread (Datastream) Bond Yield (Datastream) Bank Balance Sheet and Income Statement Data SNL database Sample Period: 2010Q4 2016Q2, Quarterly Sample Size: 1,248 observations for 78 banks from 18 European countries 17

18 Summary Statistics Variables (1) (2) (3) (4) (5) N Mean St.D. Min Max Bank Level Loan Growth 1, Size 1, CET1 Ratio 1, Liquidity 1, Interbank Ratio 1, Profitability 1, Sovereign Holding Position Total Position 1, HTM Position 1, AFS Position 1, TB Position 1, Derivative Position 1, Short-term Position 1, Medium-term Position 1, Long-term Position 1, Domestic Position 1, Foreign Position 1, Country Level GDP Growth 1, Unemployment Rate 1, Inflation Rate 1, Depreciation 1, Gov. Bond Yield 1, Number of Banks

19 Aggregate Sovereign Shocks Downgrade Notches Yield Increase CDS Increase 19

20 Average Total Exposure to the Crisis Total Expo (Downgrade) Total Expo (CDS Spread) Total Expo (Bond Yield) 20

21 Trend of Loan Growth, breakdown by exposure level First multi-notch downgrading; Start of the crisis LTRO LTRO2 Greek Debt Swap Mario Draghi Whatever it takes Expo Low (44) Expo High (43) 21

22 Baseline Result: Effect of Total Exposure on Lending Dependent Variable: Loan Growth Downgrade CDS Spread Bond Yield (1) (2) (3) (4) (5) (6) (7) (8) (9) Total Exposure t ** (0.310) (0.341) (0.249) Total Exposure t ** ** (0.444) (0.311) (0.368) Total Exposure t (0.402) (0.424) (0.413) R-squared Observations 1,112 1, ,112 1, ,112 1, No. of Banks Country Controls Yes Yes Yes Yes Yes Yes Yes Yes Yes Bank Controls Yes Yes Yes Yes Yes Yes Yes Yes Yes Quarter FE Yes Yes Yes Yes Yes Yes Yes Yes Yes Bank FE Yes Yes Yes Yes Yes Yes Yes Yes Yes Robust standard errors in parentheses. *** p<0.01, ** p<0.05, * p<0.1 22

23 Baseline Result: Effect of Total Exposure on Lending Dependent Variable: Loan Growth Downgrade (1) (2) (3) ** Expo Loan Total Exposure t Growth (0.310) Total Exposure t ** (0.444) Total Exposure t (0.402) t t+1 t+2 Quarter 1 Quarter 2 Quarter 3 R-squared Observations 1,112 1, Interpretation: No. of Banks percent increase in banks Country Controls Yes Yes Yes exposure to the crisis would Bank Controls Yes Yes Yes reduce bank s loan growth rate Quarter FE Yes Yes Yes by 1 percentage point 2 quarters Bank FE Yes Yes Yes later. Robust standard errors in parentheses. *** p<0.01, ** p<0.05, * p<0.1 23

24 2 nd Step Results: Bank Strength and Performance Dependent Variable: Loan Growth % Exposure scaled by Downgrade (1) (2) (3) (4) (5) (6) Total Exposure t ** ** ** * (0.433) (1.576) (1.491) (1.646) (1.601) (1.694) Total Exposure t 2 * CET1 Ratio t *** (1.000) Total Exposure t 2 * Tier1 Capital Ratio t *** (0.870) Total Exposure t 2 * Total Capital Ratio t *** (0.925) Total Exposure t 2 * CET1/Total Assets t * (1.237) Total Exposure t 2 * Equity/Total Assets t ** (0.788) R-squared Observations 1,041 1,041 1,041 1,041 1,041 1,041 No. of Banks Other Interactions Yes Yes Yes Yes Yes Bank Controls Yes Yes Yes Yes Yes Yes Country Controls Yes Yes Yes Yes Yes Yes Quarter FE Yes Yes Yes Yes Yes Yes Bank FE Yes Yes Yes Yes Yes Yes Robust standard errors in parentheses. *** p<0.01, ** p<0.05, * p<0.1 24

25 2 nd Step Results: Bank Strength and Performance Dependent Variable: Loan Growth % (1) (2) Total Exposure t ** ** ** (0.433) (1.576) Expo Loan Total Exposure t 2 * CET1 Ratio t *** Growth (1.000) t t+1 t+2 Quarter 1 Quarter 2 Quarter 3 Interpretation: R-squared Observations 1,041 1,041 No. of Banks Other Interactions Yes Bank Controls Yes Yes Country Controls Yes Yes Quarter FE Yes Yes Bank FE Yes Yes Robust standard errors in parentheses. *** p<0.01, ** p<0.05, * p<0.1 1 percent increase in banks exposure to the crisis would reduce bank s loan growth rate by 1 percentage point 2 quarters later. 25

26 2 nd Step Results: Bank Strength and Performance Dependent Variable: Loan Growth % (1) (2) Total Exposure t ** ** -3.84** (0.433) (1.576) Expo Loan Total Exposure t 2 * CET1 Ratio t *** Growth (1.000) t t+1 t+2 Quarter 1 Quarter 2 Quarter 3 Interpretation: R-squared Observations 1,041 1,041 No. of Banks Other Interactions Yes Bank Controls Yes Yes Country Controls Yes Yes Quarter FE Yes Yes Bank FE Yes Yes Robust standard errors in parentheses. *** p<0.01, ** p<0.05, * p<0.1 For less capitalised banks, 1 percent increase in banks exposure to the crisis would reduce bank s loan growth rate by 3.8 percentage point 2 quarters later. 26

27 2 nd Step Results: Bank Strength and Performance Dependent Variable: Loan Growth % (1) (2) Total Exposure t ** ** -3.84** *** (0.433) (1.576) Expo Loan Total Exposure t 2 * CET1 Ratio t *** Growth (1.000) t t+1 t+2 Quarter 1 Quarter 2 Quarter 3 Interpretation: R-squared Observations 1,041 1,041 No. of Banks Other Interactions Yes Bank Controls Yes Yes Country Controls Yes Yes Quarter FE Yes Yes Bank FE Yes Yes Robust standard errors in parentheses. *** p<0.01, ** p<0.05, * p<0.1 For better capitalised banks, the effect would be insignificant. 27

28 2 nd Step Result: Capital Channel through AFS Exposure Dependent Variable: Loan Growth % Exposure scaled by Downgrade Notches (1) (2) (3) (4) (5) (6) AFS Exposure t * ** * (0.576) (4.600) (3.990) (4.468) (3.928) (3.245) HTM Exposure t ** (0.740) (5.028) (4.069) (5.058) (4.951) (4.497) TB Exposure t (1.160) (20.491) (9.199) (11.203) (13.463) (10.003) Derivative Exposure t ** (6.140) (42.008) (44.384) (47.656) (51.813) (43.244) AFS Exposure t 2 * CET1 Ratio t ** (2.886) AFS Exposure t 2 * Tier1 Capital Ratio t ** (3.058) AFS Exposure t 2 * Total Capital Ratio t * (2.174) AFS Exposure t 2 * CET1/Total Assets t ** (3.639) AFS Exposure t 2 * Equity/Total Assets t (1.441) R-squared Observations 1,041 1,041 1,041 1,041 1,041 1,041 No. of Banks Other Interactions Yes Yes Yes Yes Yes Bank Controls Yes Yes Yes Yes Yes Yes Country Controls Yes Yes Yes Yes Yes Yes Quarter FE Yes Yes Yes Yes Yes Yes Bank FE Yes Yes Yes Yes Yes Yes Robust standard errors in parentheses. *** p<0.01, ** p<0.05, * p<0.1 28

29 2 nd Step Result: Capital Channel through Short-term Exposure Dependent Variable: Loan Growth % Exposure scaled by Downgrade Notches (1) (2) (3) (4) (5) (6) Short_term Exposure t *** ** (1.139) (6.953) (6.750) (8.790) (8.716) (8.357) Medium_term Exposure t *** * (1.325) (7.920) (8.172) (9.763) (10.191) (10.098) Long_term Exposure t (0.914) (5.086) (4.976) (5.357) (5.256) (6.080) Short_term Exposure t 2 * CET1 Ratio t *** (2.997) Short_term Exposure t 2 * Tier1 Capital Ratio t *** (2.393) Short_term Exposure t 2 * Total Capital Ratio t (3.487) Short_term Exposure t 2 * CET1/Total Assets t (5.831) Short_term Exposure t 2 * Equity/Total Assets t (2.804) R-squared Observations 1,041 1,041 1,041 1,041 1,041 1,041 No. of Banks Other Interactions Yes Yes Yes Yes Yes Bank Controls Yes Yes Yes Yes Yes Yes Country Controls Yes Yes Yes Yes Yes Yes Quarter FE Yes Yes Yes Yes Yes Yes Bank FE Yes Yes Yes Yes Yes Yes Robust standard errors in parentheses. *** p<0.01, ** p<0.05, * p<0.1 29

30 2 nd Step Result: Capital Channel through Domestic Exposure Dependent Variable: Loan Growth % Exposure scaled by Downgrade Notches (1) (2) (3) (4) (5) (6) Domestic Exposure t ** ** ** (0.433) (1.470) (1.348) (1.532) (1.521) (1.695) Foreign Exposure t (1.087) (5.765) (6.382) (5.753) (5.750) (6.135) Domestic Exposure t 2 * CET1 Ratio t *** (0.847) Domestic Exposure t 2 * Tier1 Capital Ratio t *** (0.850) Domestic Exposure t 2 * Total Capital Ratio t *** (0.945) Domestic Exposure t 2 * CET1/Total Assets t * (1.114) Domestic Exposure t 2 * Equity/Total Assets t ** (0.860) R-squared Observations 1,041 1,041 1,041 1,041 1,041 1,041 No. of Banks Other Interactions Yes Yes Yes Yes Yes Bank Controls Yes Yes Yes Yes Yes Yes Country Controls Yes Yes Yes Yes Yes Yes Quarter FE Yes Yes Yes Yes Yes Yes Bank FE Yes Yes Yes Yes Yes Yes Robust standard errors in parentheses. *** p<0.01, ** p<0.05, * p<0.1 30

31 Heterogeneity of Capital Effect: GIIPS vs Non-GIIPS Dependent Variable: Loan Growth % (1) (2) (3) (4) Total Exposure t ** ** ** * (0.433) (1.495) (1.545) (1.585) Total Exposure t 2 * CET1 Ratio t *** (0.984) Total Exposure t 2 * CET1 Ratio t 1 * GIIPS 0.883** (0.430) Total Exposure t 2 * Tier1 Capital Ratio t *** (0.964) Total Exposure t 2 * Tier1 Capital Ratio t 1 * GIIPS (0.647) Total Exposure t 2 * Total Capital Ratio t *** (0.875) Total Exposure t 2 * Total Capital Ratio t 1 * GIIPS (0.536) Bank Controls Yes Yes Yes Yes Country Controls Yes Yes Yes Yes Other Interactions Yes Yes Yes Bank FE Yes Yes Yes Yes Quarter FE Yes Yes Yes Yes No. of Observations 1,041 1,041 1,041 1,041 No. of Banks R-squared Robust standard errors in parentheses. *** p<0.01, ** p<0.05, * p<0.1 31

32 2 nd Step Remarks Capital Channel Better capitalised bank tend to perform better High quality capital matters more AFS, short-term and domestic exposure play a role in the capital channel Non-linearity in the effect of regulatory capital No evidence on the heterogeneity of the capital effect among GIIPS and non-giips bank Funding Channel Deposit funding Wholesale funding Interbank Ratio(significant): Net interbank lender tend to perform better 32

33 Robustness Checks Interpolation Transform the EBA sovereign exposure data from a semi-annual frequency to a quarterly frequency Bank-level Time-varying Omitted Variable Bank-Year FE GIIPS Banks Drop GIIPS banks from the sample 33

34 Conclusion Banks sovereign exposure have a negative effect on bank lending during sovereign distress period Evidence for capital channel and interbank funding channel: better capitalised banks and net interbank lenders tend to have a higher lending growth rate during sovereign distress period. More specifically, for less capitalised banks, AFS exposure, short term exposure and domestic exposure have a large negative effect on bank s lending activity. 34

35 Policy Implications Capital matters! Higher risk weights on bank s sovereign exposure would be essential for banks to be more resilient to sovereign shocks. May need an even higher risk weights for bank s AFS exposure, short term exposure and domestic exposure 35

36 Work in Progress Extend the sample period Include observations from pre-crisis period as control group Incorporate the first sovereign exposure dataset (2010Q1) disclosed by EBA 36

37 The End Comments and Questions? 37

Wholesale funding runs

Wholesale funding runs Christophe Pérignon David Thesmar Guillaume Vuillemey HEC Paris The Development of Securities Markets. Trends, risks and policies Bocconi - Consob Feb. 2016 Motivation Wholesale funding growing source