What can we Learn from the Financial Flows of the Crisis?

Transcription

1 What can we Learn from the Financial Flows of the Crisis? Juliane Begenau Saki Bigio Jeremy Majerovitz HBS UCLA Stanford Barcelona GSE Summer Forum June 9, / 77

2 Introduction Financial Crisis Crisis in macro theory Quick Response Army of macro models now 2 / 77

3 Introduction Financial Crisis Crisis in macro theory Quick Response Army of macro models now...but are shocks and frictions consistent w/ financial flows? 3 / 77

4 Introduction Financial Crisis Crisis in macro theory Quick Response Army of macro models now...but are shocks and frictions consistent w/ financial flows? Here use financial flows to learn about: 1. shocks affecting bank activities 2. frictions on banks 4 / 77

5 Frictions in Macro-Finance Constraints on external financing 5 / 77

6 Frictions in Macro-Finance Constraints on external financing Market value constraints: banks can divert assets Agency conflicts or asymmetric information Gertler & Kiyotaki 2010, Jerman & Quadrini 2012, Christiano & Ikeda 2013, Bigio 2014, Martinez-Miera & Suarez / 77

7 Frictions in Macro-Finance Constraints on external financing Market value constraints: banks can divert assets Agency conflicts or asymmetric information Gertler & Kiyotaki 2010, Jerman & Quadrini 2012, Christiano & Ikeda 2013, Bigio 2014, Martinez-Miera & Suarez 2014 Regulatory book value constraints Adrian & Shin 2011, Begenau / 77

8 Frictions in Macro-Finance Constraints on external financing Market value constraints: banks can divert assets Agency conflicts or asymmetric information Gertler & Kiyotaki 2010, Jerman & Quadrini 2012, Christiano & Ikeda 2013, Bigio 2014, Martinez-Miera & Suarez 2014 Regulatory book value constraints Adrian & Shin 2011, Begenau 2015 Costly asset/balance sheet adjustment Fire sales: e.g. Brunnermeier & Sannikov 2014 Debt overhang: e.g. Admati, DeMarzo, Hellwig & Pfleiderer / 77

9 Spirit of Paper Focus on Crisis and its effects on financial sector through lens of macro-theorist Interpretation of flows à la Friedman-Schwartz: Traditional banks: originators of loans and creators of deposits Shadow banks: accounting artifact More precisely: Traditional banks (TB): depository institutions Shadow asset banks (SAB): SBD, GSE, Mortgage Pools, REITs, finance companies, asset-backed security issuers Shadow liability banks (SLB): Money market mutual funds 9 / 77

10 View Two key events (shocks): 1. Implosion of shadow banks & absorption by traditional banks fire-sale shock 2. Equity losses on traditional banks wealth shock Did events impair ability to issue loans and deposits? Question fundamental for macro-finance models 10 / 77

11 Answers depend on 1. Magnitude of shocks: Magnitude of shadow-bank implosion? Magnitude of bank losses? 2. Frictions affecting traditional banking activities: Constraints on banks? Lessons: How to setup and calibrate models? 11 / 77

12 This paper (Tedious) accounting with theory in mind Steps: 1. Quantify flows from shadow to traditional banks (shock 1) 2. Quantify traditional bank losses (shock 2) 3. Analyze cross-sectional responses (constraints) 12 / 77

13 Step 1: Aggregate Financial Flows to Quantify Implosion of Shadow Banks 13 / 77

14 Traditional Financial System Saver Deposits Equity Bank Reserves Deposits Borrower Loans Equity Goods Loans Equity 14 / 77

15 Modern Financial System: SB grew to 150% of TB Saver Shadow liability Deposits MMMF Equity CP/Repo MMMF Shadow asset Bank ABS, MBS and Co CP/Repo Reserves Loans Deposits Equity Borrower Loans Goods ABS etc Equity 15 / 77

16 Shock: SBD alone report $42 B write-downs, valuation losses? Saver Shadow liability Deposits MMMF Equity CP/Repo MMMF Shadow asset Bank ABS, MBS and Co CP/Repo Reserves Deposits Borrower Loans Equity Goods Loans ABS etc Non performing Equity 16 / 77

17 Unraveling between 2007q3-2009q2 Saver Run on MMMF Shadow liability Deposits CP/Repo MMMF MMMF Equity +Deposits Bank Run on the REPO Collapse in ABCP Shadow asset ABS etc CP/Repo Reserves Loans ABS etc Deposits +Deposits Equity Repurchase of MBS by Banks Goods Borrower Loans ABS etc Equity Shadow Banks: $541 billion net outflow $499 billion increase in traditional banking assets 17 / 77

18 Accounting Principles Where are these numbers coming from? Obviously, not-trivial Consolidate shadow bank balance sheets... Balance sheets shrinking, (by t-accounts) either: Sales of assets and payout of liabilities Write-down of assets, equity losses Assets sold, someone has to buy them... Assets lost, someone else has to lose also / 77

19 Data Most of the time focus on quarterly Flow of Funds Bank level data on commercial banks and bank holding companies Data on security broker and dealers (FOCUS) filed by SEC Annual reports 19 / 77

20 Rise and Fall of Shadow Banks 20 Total Assets $ Trillion Shadow Asset Banks Shadow Liab Banks Traditional Banks 2 0 Jun98 Jun00 Jun02 Jun04 Jun06 Jun08 Jun10 Jun12 Jun14 20 / 77

21 Assets Total Assets Shadow Asset Banks Shadow Liab Banks Traditional Banks $ Trillion Change Mar07 Sep07 Mar08 Sep08 Mar09 Sep09 Mar10 Sep10 Mar11 Sep11 21 / 77

22 Asset Flows: Diff 2007 Q3 and 2009 Q2 SAB: $879 billion SLB: $805 billion / Caveat: mixes prime with gov funds TB: $912 billion TARP: $426.4 billion of which $ billion at least benefited TB What about redefinitions? Goldman Sachs and Morgan Stanley now traditional banks Holding companies with few deposits Broker dealer subsidiaries stay in SBD section of the FoF But GS and MS in SBD assets absorbed within holding company adjust SAB asset for GS and MS Effect of Adjustment: SAB outflow $1346 billion Shadow sector: net outflow of $541 billion 22 / 77

23 Upper bound of what TB did absorb? 2007q3-2009q2 Adjusting TB inflows by CB in GS or MS holdings $912 billion increase in TB assets of which $413 GS and MS TB: absorbed at most $499 billion from SB What securities needed to be absorbed and from whom? SBD, borrowed securities Detail GSE & Mortgage-pools, mortgages Detail Asset-backed security issuers Detail 23 / 77

24 SBD Gains and Losses Annual reports from large SBD Reported writedowns $46.1 Billion 24 / 77

25 Back of the Envelope Shadow banking institution: net outflow of $541 billion Write-downs: $46.1B TB assets increased by $499B, only $495B could come from the shadow Puzzle $-4.1 billion? Natural growth in TB activity (TBD compute detrended flows) Gross ammounts (double counting)? Central bank Institutional Investors Realization of losses Book versus market value accounting 25 / 77

26 Step 2: Traditional Sector Equity Losses 26 / 77

27 Data This section... BHC level data from FR Y-9C and CRSP Aggregate series corrected by dropping new entrants (e.g. MS, GS, AXP, etc.) 27 / 77

28 Major Drivers of Losses and Profits Major Drivers of Losses and Profits q1 2002q1 2004q1 2006q1 2008q1 2010q1 2012q1 2014q1 USD Billion Net Income Net Gains from Securities ( ) Provision for Loan Losses Trading Revenue 28 / 77

29 Trading Revenue Components Trading Revenue: Total, Interest Rate Exposures, and Credit Exposures q1 2002q1 2004q1 2006q1 2008q1 2010q1 2012q1 2014q1 USD Billion Trading Revenue Credit Exposures Interest Rate Exposures 29 / 77

30 Decomposition of Charge-Offs Net Charge Offs USD Billion 2000q1 2002q1 2004q1 2006q1 2008q1 2010q1 2012q1 2014q1 Other CandI IndividualLoans RELoans 30 / 77

31 Within Real-Estate Charge-Offs Real Estate Net Charge Offs USD Billion 2000q1 2002q1 2004q1 2006q1 2008q1 2010q1 2012q1 2014q1 Farm+Foreign Residential Commercial Construction 31 / 77

32 How did the losses show up in equity? 32 / 77

33 Book Equity vs. Market Capitalization Book Equity and Market Capitalization of BHCs USD Billion q1 2002q1 2004q1 2006q1 2008q1 2010q1 2012q1 2014q1 Equity Market Capitalization Equity (Public BHCs) 33 / 77

34 Book vs. Market Equity of Big Four USD Billion Bank of America J.P. Morgan Citigroup Wells Fargo 2000q1 2004q1 2008q1 2012q1 2000q1 2004q1 2008q1 2012q1 Equity Market Capitalization 34 / 77

35 Facts Two Notions... Losses Accumulated Over Recession: 1. Book Value Gains of 173 Billion 2. Market Value Losses of 656 Billion 3. Equity Issuances of 147 Billion Very little action in books / 77

36 Differences in Book and Market? From cross-section...more than only price of risk 36 / 77

37 Differences in Book and Market? From cross-section...more than only price of risk Bank incentives to delay acknowledging losses Regulation: better capital ratio Corporate Control: better numbers to shareholders + flexibility to delay losses (Laux and Leuz JEP 2010) 37 / 77

38 Are books missing something? Test whether market cap is picking up important information log(marketcap/bookequity) = α + βx + ɛ β reflects effect on market cap beyond effect on book equity Test regressors that should provide information about bank s value Return on Book Equity Return on Book Equity One Year Later Delinquent Loans to Book Equity Total Dividends to Book Equity Compare results before (2006 Q1) and after crisis (2009 Q1) 38 / 77

39 Tobin s Q and RoE (Cross-Sectional) Tobin s Q and Return on Book Equity Tobin s Q (Logged) Return on Book Equity (Trimmed) 2006 Q1 Fitted values 2009 Q1 Fitted values 39 / 77

40 Tobin s Q and RoE Next Year Tobin s Q and Return on Book Equity Next Year Tobin s Q (Logged) Return on Book Equity Next Year (Trimmed) 2006 Q1 Fitted values 2009 Q1 Fitted values 40 / 77

41 Tobin s Q and Delinquency Rates (Cross-Section) Tobin s Q and Delinquent Loans to Equity Tobin s Q (Logged) Delinquent Loans to Equity (Trimmed) 2006 Q1 Fitted values 2009 Q1 Fitted values 41 / 77

42 Tobin s Q and Dividend Rate (Cross-Section) Tobin s Q and Total Dividend Rate Tobin s Q (Logged) Total Dividend Rate (Trimmed) 2006 Q1 Fitted values 2009 Q1 Fitted values 42 / 77

43 Market data seems better Market data driven by: Profits & Future Profits Loan Delinquency Dividends Oaxaca Decomposition: 43% to 45% of Drop in Logged Tobin s Q can be explained by these four regressors Takeaway: info in market cap not in books 43 / 77

44 Why are book/market values relevant? Market Cross-section: tells us who lost wealth Book Cross-section: who acknowledged losses...relevant to learn bank constraints 44 / 77

45 Step 3: Constraints on Bank Activities 45 / 77

46 Traditional Sector Fire-Sale Shock Outflow from shadow banking system to traditional banks: $495B Wealth Shock Market value losses of 656 Billion (803 Billion Lost and 147 Billion New Equity) Effects on bank activities? Answer depends on constraints...(this section) 46 / 77

47 Illustrate different constraints Two period bank optimization model Prices and demands exogenous Choose flow: equity e, dividend div, riskless assets c, loans l, whether to acknowledge losses Net-worth N t, deposits D t, loans L t, riskless assets C t Constraints: Regulatory Market value Assets: e.g. fire sale Liabilities: e.g. debt overhang 47 / 77

48 Bank Problem Normalized by stock of loans L 0 W ( D 0, ε 0 ; F 0, p, R L) = max Q (e) + U (div) +... {div,c,l,e} R 4 + +βe [ max ( 0, (1 ɛ 1 ) ( R L l + 1 ) )] + C 1 D 1 subject to and λ (l) l + pc 1 D 1 = W 1 W 1 = pe div D 0 48 / 77

49 Frictions Regulatory constraints (can choose to acknowledge losses) L 1 κ (1 κ) ((C 1 D 1 ) + max {ε 0, 0}) Liability constraints (downward stickiness) D 1 δd 0 Market-based constraints: [ ( )] N M E V 1 ε 1 L 1 V ( ξ λ (l) ) R L. Fire-sale function: λ (l) Equity adjustment costs: Q (e) 49 / 77

50 This Section Book-based constraints irrelevant: Banks can slow down accounting Sticky assets and liabilities? Wealth shocks and market-based constraints: Induce decrease in leverage Shocks Wealth, Market-Leverage 50 / 77

51 Leverage Leverage Ratios q1 2002q1 2004q1 2006q1 2008q1 2010q1 2012q1 2014q1 Book Leverage Market Leverage 51 / 77

52 Frictions in Equity Issuance? Banks issued lots of equity, but not enough to undo the losses Anecdotal: hard to issue enough equity after hit by losses 52 / 77

53 Equity Issuances (Levels Since 2000 Q1) USD Billion Levels of Market Capitalization and Equity Issuances 2000q1 2002q1 2004q1 2006q1 2008q1 2010q1 2012q1 2014q1 Market Capitalization Summed Equity Issuances 53 / 77

54 Equity Issuances Big Four USD Billion Bank of America J.P. Morgan Citigroup Wells Fargo 2000q1 2004q1 2008q1 2012q1 2000q1 2004q1 2008q1 2012q1 Market Capitalization Summed Equity Issuances 54 / 77

55 Irreversibility of Leverage in the Cross-Section Can banks de-lever in the cross-section? Estimate a dynamic market-equity elasticity of liabilities Idea: Market-equity describes wealth Poorer, more market-constrained banks should reduce liabilities Did we see this? Similarly: did we see a reduction in loans, liquid assets such as cash and liquid liabilities such as repos? 55 / 77

56 Liabilities vs. Market Capitalization (Log Differences) Liabilities and Market Capitalization Change in Logs Over Past Three Years Liabilities (Log Differences) Market Capitalization (Log Differences) 2006 Q1 Fitted values 2009 Q1 Fitted values 56 / 77

57 Dynamic Equity Elasticities OLS (three-year diffs). (1) (2) (3) (4) Liabilities (3 Year) Loans (3 Year) Repo (3 Year) Cash (3 Year) Log Market Cap (0.0296) (0.0310) (0.183) (0.0790) Post * Log Market Cap (0.0331) (0.0347) (0.202) (0.0883) Post (0.0289) (0.0304) (0.160) (0.0773) Constant (0.0206) (0.0216) (0.115) (0.0551) Observations R Standard errors in parentheses p < 0.05, p < 0.01, p < / 77

58 Discussion: illiquid liabilities a constraint? Illiquid liabilities: (can t turn away depositors) Some highly liquid (e.g. Repo) If illiquid liabilities constraint: Repo elasticity should be higher than non-repo during the crisis Result: Repo elasticity also zero Constrained banks should be picking a corner solution But: Most banks have some liquid liabilities during the crisis 58 / 77

59 Composition of Liabilities Percent Deposits Fed Funds Purchased Trading Liabilities Other BHC Liabilities Composition Liabilities 10 Trillion $ / 77

60 Deposits as a Share of Liabilities Deposits as a Share of Liabilities Density Deposits as a Share of Liabilities 2006 Q Q1 60 / 77

61 Deposits / Liabilities (Weighted by Liabilities) Deposits as a Share of Liabilities (Weighted by Total Liabilities) Density Deposits as a Share of Liabilities 2006 Q Q1 61 / 77

62 Repo as a Share of Liabilities Repo as a Share of Liabilities Density Repo as a Share of Liabilities (Trimmed) 2006 Q Q1 62 / 77

63 Repo / Liabilities (Weighted by Liabilities) Repo as a Share of Liabilities (Weighted by Total Liabilities) Density Repo as a Share of Liabilities (Trimmed) 2006 Q Q1 63 / 77

64 Discussion: Sticky Leverage Hard to sell assets or unwillingness to sell assets? If illiquid assets constraint: Poorer banks should have: sold off liquid assets, reduce liquid liabilities Data: Dynamic equity elasticity of cash negative: poorer banks increased reserves Caveats: Bigger losses liquidity demand (omitted variable bias) Alternative: dynamic debt-overhang Shareholders: prefer maintain high leverage 64 / 77

65 Discussion: bank constraints during crisis? Regulatory constraints: skirted by slowing-down books Discretionary regulation (e.g. stress-tests) Model book constraints as potentially (random) active constraints Data doesn t suggest market-based constraints Don t see greater deleverage in cross-section Market capturing wealth shocks: Illiquid liabilities not a constraint Illiquid assets might be a constraint asymmetric info, specialization, transaction cost Insufficient equity issuances Agency issues dynamic-debt overhang 65 / 77

66 Conclusion - Takeaway Insights Two numbers: measure size of implosion of shadow-banks and losses to traditional banks. Modeling Constraints: Different class of constraints than pop in macro literature Wealth shocks: illiquid assets (+deposits) + agency frictions 66 / 77

67 End 67 / 77

68 Drivers of Market Losses Dep variable: log market cap growth (2006 Q1 to 2009 Q1) Regressors: (2006 Q1). MBS MarketCap. (1) (2) (3) (4) (5) (6) (0.0841) Loans MarketCap (0.0323) (0.0751) RELoans MarketCap (0.0355) (0.0836) CDSSold MarketCap (0.0856) CDSNet MarketCap (2.789) Constant (0.0677) (0.140) (0.120) (0.139) (0.0491) (0.0490) Observations R Standard errors in parentheses p < 0.05, p < 0.01, p < / 77

69 Security Brokers & Dealers: Total Assets 5 Total Financial Assets US$ Trillion Jun90 Jun94 Jun98 Jun02 Jun06 Jun10 Jun13 69 / 77

70 Lion Share: Other Securities 3 REPO Total Government Securities Total Private Securities Other Securities US$ Trillion Jun90 Jun94 Jun98 Jun02 Jun06 Jun10 Jun13 70 / 77

71 Other Securities: Mostly Borrowed Securities 1.5 Borrow Securities Unidentified Misc 1 US$ Trillion Jun90 Jun94 Jun98 Jun02 Jun06 Jun10 Jun13 Net of receivables due and payables owed to other SBD, largely cash collateral associated with securities lending transactions 71 / 77

72 Security Brokers & Dealers: Liabilities REPO Credit Market Instruments Other Liability Security Credit 2.5 US$ Trillion Jun90 Jun94 Jun98 Jun02 Jun06 Jun10 Jun13 back 72 / 77

73 Asset-backed Security Issuers: Assets 5 Asset-backed Security Issuers - Assets 4.5 Government Sec. Mortgages Consumer Credit Sec. Business Loans US$ Trillion Jun91 Jun94 Jun97 Jun00 Jun03 Jun06 Jun09 Jun12 73 / 77

74 Asset-backed Security Issuers: Liabilities Commercial Paper Corporate Bonds Asset-backed Security Issuers - CP Liabilities US$ Trillion Jun91 Jun94 Jun97 Jun00 Jun03 Jun06 Jun09 Jun12 back 74 / 77

75 Agency-and GSE-backed mortgage pools: Assets 6 #Mortgages 5 Home Multifamily residential Commercial Farm 4 US$ Trillion Jun99 Jun02 Jun05 Jun08 Jun11 75 / 77

76 Government-Sponsored Enterprises (GSEs): Assets 7 #Mortgages 6 Short Mortgages Credit Mkt Instrument other than Mortgages 5 US$ Trillion Jun99 Jun02 Jun05 Jun08 Jun11 76 / 77

77 back 77 / 77