Why Are Banks Not Recapitalized During Crises?

Transcription

1 Why Are Banks Not Recapitalized During Crises? Matteo Crosignani NYU Stern September 2014

2 Outline 1 THREE FACTS 2 INTUITION 3 MODEL 4 EMPIRICAL EVIDENCE

3 FACT 1: RELIANCE ON DOMESTIC BANKS Levels Non-GIIPS

4 FACT 2: PRIVATE AND PUBLIC LENDING Non-GIIPS

5 FACT 3: RELUCTANCE TO RECAPITALIZE EU BANKS EBA (Oct11): banks need $146 bn to meet new capital requirements Acharya, Engle, Pierret (2014): EBA underestimates capital shortfall reliance on risk-weight Basel III: delays in implementing CRD IV (Dec12 Jan14) Difficult to intervene during a sovereign crisis What I would admit is that maybe it s not the best moment in the middle of the crisis to change the rules [...] Danièle Nouy, Chair of the Suvervisory Board of the SSM

6 Outline 1 THREE FACTS 2 INTUITION 3 MODEL 4 EMPIRICAL EVIDENCE

7 THIS PAPER Fact 1. Highly levered banks in the periphery risk-shift buying domestic govt bonds High payoff in the good state High correlation with other sources of revenues LL binds in bad state Zero risk weight EAD

8 THIS PAPER Fact 1. Highly levered banks in the periphery risk-shift buying domestic govt bonds High payoff in the good state High correlation with other sources of revenues LL binds in bad state Zero risk weight Fact 2. Govt bonds become more attractive relative to lending to private sector crowding-out EAD

9 THIS PAPER Fact 1. Highly levered banks in the periphery risk-shift buying domestic govt bonds High payoff in the good state High correlation with other sources of revenues LL binds in bad state Zero risk weight Fact 2. Govt bonds become more attractive relative to lending to private sector crowding-out Fact 3. Govt faces a trade-off when setting capital requirements: distortion in private lending vs. higher demand for bonds Myopic govt keeps domestic banks undercapitalized buyers of last resort Race to the bottom in regulation EAD

10 LITERATURE REVIEW Theory: Acharya, Rajan (2013); Gennaioli, Martin, Rossi (2014); Acharya, Drechsler, Schnabl (2014); Bolton, Jeanne (2011); Broner, Erce, Martin, Ventura (2013); Broner, Martin, Ventura (2010). Empirical: Acharya, Steffen (2013); Becker and Ivashina (2014); Arslanalp, Tsuda (2012); Brutti, Saure (2013); Acharya, Engle, Pierret (2013); Drechsel, Drechsler, Marques-Ibanez, Schnabl (2013).

11 WELL CAPITALIZED BANKS Italy Spain IT Firms Unicredit Santander ES Firms

12 F1: HOME BIAS AND CROWDING-OUT Italy Spain IT Firms Unicredit Santander ES Firms

13 F2: BANKS NOT RECAPITALIZED DURING CRISES Italy Spain IT Firms Unicredit Santander ES Firms

14 F3: RACE TO THE BOTTOM Italy Spain IT Firms Unicredit Santander ES Firms

15 Outline 1 THREE FACTS 2 INTUITION 3 MODEL 4 EMPIRICAL EVIDENCE

16 ENVIRONMENT Two-periods No discounting Risk neutrality Two countries with a banking sector and a government Countries only differ in domestic banks initial capital Shock hits w.p. 1 θ (0, 1) at t = 1.5

17 α(1 k) INVESTMENT OPPORTUNITY SET Italy Spain (1 α)(1 k) IT Firms k Bank I Invest k in lending α home bias in govt bond market

18 PRIVATE LENDING Invest k at t = 0. Revenues f(k) at t = 1 and t = 2 Shock destroys t = 2 revenues Govt levies tax τ in both periods k (1 τ)f(k) θ (1 τ)f(k) t=0 t=1 t=2 1 θ 0

19 DOMESTIC GOVERNMENT BONDS Banks invest q D = α(1 k) in domestic govt bonds Govt repays debt with proceeds from tax collection default in bad state q D θ q D (1+r) t=0 t=1 t=2 1 θ 0

20 FOREIGN GOVERNMENT BONDS q F = (1 α)(1 k) invested in foreign govt bonds Foreign govt defaults w.p. 1 θ q F θ θq F (1+r ) t=0 t=1 t=2 1 θ θq F (1+r )

21 GOVERNMENT DEBT CAPACITY Govt maximizes spending using debt τ is exogenous Issues debt D at t = 1. Chooses whether to default on D at t = 2 Cost of default(1+r)c(α, k) > 0 where C 1 > 0 and C 2 > 0

22 GOVERNMENT DEBT CAPACITY Govt maximizes spending using debt τ is exogenous Issues debt D at t = 1. Chooses whether to default on D at t = 2 Cost of default(1+r)c(α, k) > 0 where C 1 > 0 and C 2 > 0 Govt defaults if not able and/or not willing to pay Not able if τf(k) < D(1+ r) (AP) Not willing if C(α, k) < D (WP)

23 GOVERNMENT DEBT CAPACITY Govt maximizes spending using debt τ is exogenous Issues debt D at t = 1. Chooses whether to default on D at t = 2 Cost of default(1+r)c(α, k) > 0 where C 1 > 0 and C 2 > 0 Govt defaults if not able and/or not willing to pay Not able if τf(k) < D(1+ r) (AP) Not willing if C(α, k) < D (WP) investors are willing to buy govt bonds if { D min C(α, k), τθf(k) } 1+r

24 GOVERNMENT SPENDING τf(k )+τβθf(k ) }{{} tax collection + D βθ(1+r)d }{{} govt debt

25 GOVERNMENT SPENDING where τf(k )+τβθf(k ) }{{} tax collection + D βθ(1+r)d }{{} govt debt { D = min C(α, k), τθf(k) } 1+r

26 BANKS PROBLEM θ 2(1 τ)f(k) t=1 1 θ (1 τ)f(k) Good state: lending (t = 1, 2) Bad state: lending (t = 1)

27 BANKS PROBLEM θ 2(1 τ)f(k) +q D (1+r) t=1 1 θ (1 τ)f(k) Good state: lending (t = 1, 2) + dom. bonds Bad state: lending (t = 1)

28 BANKS PROBLEM θ 2(1 τ)f(k) +q D (1+r) +q F θ(1+r ) t=1 1 θ (1 τ)f(k) +q F θ(1+r ) Good state: lending (t = 1, 2) + dom. bonds + for. bonds Bad state: lending (t = 1) + for. bonds

29 BANKS PROBLEM θ [ 2(1 τ)f(k) +q D (1+r) +q F θ(1+r ) L] + t=1 1 θ [ (1 τ)f(k) +q F θ(1+r ) L] + Good state: lending (t = 1, 2) + dom. bonds + for. bonds - debt Bad state: lending (t = 1) + for. bonds - debt

30 BANKS PROBLEM θ 2(1 τ)f(k) +q D (1+r) +q F θ(1+r ) L t=1 1 θ (1 τ)f(k) +q F θ(1+r ) L Suppose L is low enough (W) LL does not bind Invest in domestic bonds iff θ(1+r) θ(1+r )

31 BANKS PROBLEM θ 2(1 τ)f(k) +q D (1+r) +q F θ(1+r ) L t=1 1 θ 0 Suppose L is high enough (U) LL binds in the bad state of the world Invest in domestic bonds iff (1+r) θ(1+r ) EAD

32 BENCHMARK CASE: WELL CAPITALIZED BANKS Assume L low enough identical: k, D, r Two types of equilibria { D = min C(α, k), τθf(k) } 1+r α(z)

33 BENCHMARK CASE: WELL CAPITALIZED BANKS Assume L low enough identical: k, D, r Two types of equilibria α < α (WP) binds { D = min C(α, k), τθf(k) } 1+r Italy Spain Bank I Bank S α(z)

34 BENCHMARK CASE: WELL CAPITALIZED BANKS Assume L low enough identical: k, D, r Two types of equilibria { D = min C(α, k), τθf(k) } 1+r α < α (WP) binds α α (AP) binds Italy Spain Italy Spain Bank I Bank S Bank I Bank S α(z)

35 HOME BIAS AND CROWDING-OUT Assume f(k) = ǫ k and C(α, k) = z (1 α)(1 k) Suppose one country has undercapitalized banks e.g., Italy has U banks and Spain has W banks

36 HOME BIAS AND CROWDING-OUT Assume f(k) = ǫ k and C(α, k) = z (1 α)(1 k) Suppose one country has undercapitalized banks e.g., Italy has U banks and Spain has W banks R.S. Home Bias in Italy

37 HOME BIAS AND CROWDING-OUT Assume f(k) = ǫ k and C(α, k) = z (1 α)(1 k) Suppose one country has undercapitalized banks e.g., Italy has U banks and Spain has W banks R.S. Home Bias in Italy Home bias in Spain: α I = α S = 1 r I r S needed to attract Spanish banks α I = 1 Crowding-out in Italy lower tax collection Italy cannot credibly commit to repay its debt

38 HOME BIAS AND CROWDING-OUT Assume f(k) = ǫ k and C(α, k) = z (1 α)(1 k) Suppose one country has undercapitalized banks e.g., Italy has U banks and Spain has W banks R.S. Home Bias in Italy Home bias in Spain: α I = α S = 1 r I r S needed to attract Spanish banks α I = 1 Crowding-out in Italy lower tax collection Italy cannot credibly commit to repay its debt 1+ r U = ǫ 2θ 2θ (1 τ)(1+ θ)((1 τ)(1+ θ)+ 2τθ 2 ) θ(1 τ)(2(1 τ)+θ(1+τ(1+2θ)) ) }{{}}{{} Fundamental Term Risk-Shifting Term

39 BANKS CAPITALIZATION W 0 L I L I

40 BANKS CAPITALIZATION W U 0 L I L I L I

41 BANKS CAPITALIZATION L S WU UU L S Multiple Equilibria L S WW UW 0 L I L I L I

42 EQUILIBRIUM MORAL SUASION Govt recapitalizes banks iff τǫ(1+ βθ)( k W k U ) }{{} increased tax collection D U D W }{{} lower debt issuance + βθ((1+r W )D W (1+r U )D U ) }{{} higher payments to bondholders

43 EQUILIBRIUM MORAL SUASION Govt recapitalizes banks iff τǫ(1+ βθ)( k W k U ) }{{} increased tax collection Govt faces a trade-off: D U D W }{{} lower debt issuance Collects taxes in both periods Issues debt at t = 1 + βθ((1+r W )D W (1+r U )D U ) }{{} higher payments to bondholders

44 EQUILIBRIUM MORAL SUASION Govt recapitalizes banks iff τǫ(1+ βθ)( k W k U ) }{{} increased tax collection Govt faces a trade-off: D U D W }{{} lower debt issuance Collects taxes in both periods Issues debt at t = 1 + βθ((1+r W )D W (1+r U )D U ) }{{} higher payments to bondholders! β such that govt recapitalizes domestic banks iff β j > β Might be that β 1 Myopia not needed

45 RACE TO THE BOTTOM IN CAPITAL REGULATION Suppose economy is in WW region and govts are sufficiently myopic continuum of eqm with α I = α S = α if α α, (AP) binds race to the bottom in capital regulation Trembling reasons to select high home bias equilibria: access to LOLR, information frictions,...

46 EFFICIENCY Recapitalization is always efficient... ((1 k W )(1+ r W ) (1 k U )(1+r U )) + ǫ(1 τ)((1+ θ) k W 2θ k U ) 0 }{{}}{{} higher profits from no crowding-out holding govt bonds

47 EFFICIENCY Recapitalization is always efficient... ((1 k W )(1+ r W ) (1 k U )(1+r U )) + ǫ(1 τ)((1+ θ) k W 2θ k U ) 0 }{{}}{{} higher profits from no crowding-out holding govt bonds...but high levered banks want to recapitalize if and only if ((1 k W )(1+r W ) (1 k U )(1+r U )) + ǫ(1 τ)((1+ θ) k W 2θ k U ) L(1 θ) }{{}}{{} higher profits from no crowding-out holding govt bonds

48 TO DO Aggregate shocks θ D > θ I Contagion Partial write-downs of debt Allow foreign private lending

49 Outline 1 THREE FACTS 2 INTUITION 3 MODEL 4 EMPIRICAL EVIDENCE

50 SUPPORTING EMPIRICAL EVIDENCE Model prediction: highly levered and local banks drive the purchases of domestic bonds in crisis

51 SUPPORTING EMPIRICAL EVIDENCE Model prediction: highly levered and local banks drive the purchases of domestic bonds in crisis Data: EBA stress tests and Bankscope

52 SUMMARY STATISTICS 10Q1 10Q4 11Q3 11Q4 12Q2 12Q4 13Q2 Tot. GIIPS Assets Core Exposure GIIPS to GIIPS Core Tot. Sov. GIIPS Exposure Core Dom. Sov. GIIPS Exposure Core E/A GIIPS Core Quantities in m EUR

53 GIIPS HIGH VS. LOW LEVERAGE BANKS

54 GIIPS LOCAL VS. INTERNATIONAL BANKS Core Banks

55 ALTERNATIVE CHANNELS Moral suasion Consistent as long as what we observe is an equilibrium outcome Regulatory arbitrage Inconsistent with lower holdings of GIIPS non-domestic bonds Information advantage Capitalization should not matter

56 CONCLUSION Shown that during crises govts rely on domestic bond buyers banks increase holdings of domestic govt bonds policy makers are reluctant to recapitalize domestic banks Developed a tractable GE model that rationalizes these facts and yields additional empirical implications Shown that stress test data support the proposed channel

57 Outline 5 Appendix

58 Appendix FACT 1: INCREASING HOME BIAS (LEVELS) Back

59 Appendix FACT 1: HOME BIAS (NON-GIIPS) Back

60 Appendix FACT 1: HOME BIAS (NON-GIIPS) Back

61 Appendix FACT 2: PRIVATE AND PUBLIC LENDING (NON-GIIPS) Back

62 Appendix FACT 2: PRIVATE AND PUBLIC LENDING (NON-GIIPS) Back

63 Appendix GIIPS LOCAL & HIGH LEV. VS. INTERN. & LOW LEV. Back

64 Appendix CORE HIGH VS. LOW LEVERAGE BANKS Back

65 Appendix EXPOSURES AT DEFAULT Introduction Banks Problem Bank Country GIIPS Bank Name Exposure to EAD (e m) EAD/E GR EFG Eurobank Ergasias GR 53, GR Alpha Bank GR 46, IT Banca Monte Paschi Siena IT 205, ES Banco Popular Espanol ES 120, IT Banco Popolare IT 122, ES Caixa ES 259, IE Irish Life and Permanent IE 36, ES BBVA ES 378, IT Intesa Sanpaolo IT 418, IT Unicredit IT 382, IE Bank of Ireland IE 68, GR EFG Eurobank Ergasias PL 5, IE Bank of Ireland GB 64, IE Allied Irish Banks IE 85, GR EFG Eurobank Ergasias RO 4, ES Banco Santander ES 355, GR EFG Eurobank Ergasias BG 3, ES Banco Santander GB 292, GR EFG Eurobank Ergasias DE 2, IT Unicredit DE 151, GR Alpha Bank CY 4, IE Irish Life and Permanent GB 8, IE Allied Irish Banks GB 32, GR Alpha Bank RO 4, GR Alpha Bank GB 3, IT Unicredit AT 74,

66 Appendix HHB AND LHB REGIONS Back α 1 HHB LHB α 0 z z