Credit supply and the housing boom

Transcription

1 Alejandro Jus5niano Federal Reserve Bank of Chicago Giorgio Primiceri Northwestern University Andrea Tambalo: Federal Reserve Bank of New York Anacapri June 26, 2014

2 1. Decline in real mortgage rates +"!!# *"!!# )"!!# ("!!# '"!!# &"!!# %"!!# $"!!# -./01213#/203#4#567#89:#;<=8$!>/?# -./01213#/203#4#567#89:#;<=8$>/?# $,,!# $,,(# %!!!# %!!(# %!$!#

3 2. Unprecedented boom and bust in (real) house price %!"# %+"#,-,.# / # %*"# %)"# %("# %'"# %&"# %%"# %""# $"#!"# %$!)# %$$"# %$$)# &"""# &"")# &"%"#

4 3. Boom and bust in household debt!")!$!"(%$!"(!$!"'%$!"'!$!"%%$!"%!$!"&%$!"&!$!"#%$!"#!$!!"#$%&'(')*+&$+,-."%(/$"012$3"$4"1567*8" *++!$ *++%$,!!!$,!!%$,!*!$

5 4. Debt- to- collateral ra5o: constant and then spikes!"(!%!!"#$%&'(')*+&$+%)(,")*&(&)"%(-$"./,$0"$1"/234*5"!"$$%!"$!%!"'$%!"'!%!"&$%!"&!%!"#$% )**!% )**$% #!!!% #!!$% #!)!%

6 The US economy in the 2000s: Four stylized facts Decline in mortgage rates Unprecedented boom- bust cycle in house prices Massive HH debt accumula5on, and then deleveraging Debt- to- collateral ra5o constant, and then spikes

7 This paper Ques%on: What is the fundamental driver behind these facts?

8 This paper Ques%on: What is the fundamental driver behind these facts? Approach: Model of HH borrowing as laboratory borrowing constraints, houses as collateral lending constraints

9 Summary of the results Ques%on: What is the fundamental driver behind these facts? Answer: An increase in credit supply brought about by looser lending constraints. Consistent with decline in mortgage rates large increase in house prices constant debt- to- collateral ra5o during the boom

10 Summary of the results Ques%on: What is the fundamental driver behind these facts? Answer: An increase in credit supply brought about by looser lending constraints. Consistent with decline in mortgage rates large increase in house prices constant debt- to- collateral ra5o during the boom Looser collateral requirements not an important driving force of the boom. At odds with the behavior of mortgage rates, house prices, household leverage

11 Summary of the results Ques%on: What is the fundamental driver behind these facts? Answer: An increase in credit supply brought about by looser lending constraints. Consistent with decline in mortgage rates large increase in house prices constant debt- to- collateral ra5o during the boom Looser collateral requirements not an important driving force of the boom. At odds with the behavior of mortgage rates, house prices, household leverage Excessive loosening of collateral requirements can explain why house prices started to fall, even if liberaliza5on was in full swing

12 Outline Model Parameteriza5on Quan5ta5ve results Expansion in credit supply Loosening of collateral requirements

13 Simplest model Build on Kiyotaki and Moore (1997) Iacoviello (2005) Campbell and Hercowitz (2006) 2 groups of households Pa5ent Lenders Impa5ent Borrowers

14 Simplest model Build on Kiyotaki and Moore (1997) Iacoviello (2005) Campbell and Hercowitz (2006) 2 groups of households Pa5ent Lenders Impa5ent Borrowers No produc5on income is exogenous Fixed supply of houses

15 The problem of the borrowers max E 0 t β b t =0 [ u( c ) b,t + v( h )] b,t c b,t + p t [ h b,t +1 ( 1 δ)h ] b,t + R t 1 D b,t 1 y b,t + D b,t

16 The problem of the borrowers max E 0 t β b t =0 [ u( c ) b,t + v( h )] b,t c b,t + p t [ h b,t +1 ( 1 δ)h ] b,t + R t 1 D b,t 1 y b,t + D b,t Borrowing is limited by a collateral constraint D b,t θ p t h b,t +1

17 The problem of the borrowers max E 0 t β b t =0 [ u( c ) b,t + v( h )] b,t c b,t + p t [ h b,t +1 ( 1 δ)h ] b,t + R t 1 D b,t 1 y b,t + D b,t Borrowing is limited by a collateral constraint Associated mul5plier: μ 0 D b,t θ p t h b,t +1

18 The problem of the lenders ( β l > β b ) max E 0 t β l t =0 [ u( c ) l,t + v( h )] l,t c l,t + p t [ h l,t +1 ( 1 δ)h ] l,t + R t 1 D l,t 1 y l,t + D l,t

19 The problem of the lenders ( β l > β b ) max E 0 t β l t =0 [ u( c ) l,t + v( h )] l,t c l,t + p t [ h l,t +1 ( 1 δ)h ] l,t + R t 1 D l,t 1 y l,t + D l,t Mortgage lending is limited by a lending constraint D l,t L

20 The lending constraint D l,t L In reduced form, captures all factors hampering the free flow of funds from the savers to mortgage financing Implicit or explicit, regulatory and technological constraints on mortgage lending

21 The lending constraint D l,t L In reduced form, captures all factors hampering the free flow of funds from the savers to mortgage financing Implicit or explicit, regulatory and technological constraints on mortgage lending Example: Money- market funds, pension funds and insurance companies are restricted by regula5ons to holding only the safest securi5es

22 The lending constraint D l,t L In reduced form, captures all factors hampering the free flow of funds from the savers to mortgage financing Implicit or explicit, regulatory and technological constraints on mortgage lending Example: Money- market funds, pension funds and insurance companies are restricted by regula5ons to holding only the safest securi5es Can be interpreted as stemming from leverage restric5on or regulatory- capital requirement of financial intermediaries

23 Two addi5onal simplifying assump5ons Rigid demand for houses by the lenders Linear u5lity in consump5on p t = β b [ h,c + ( 1 δ)p ] t +1 1 µ t θ mrs b,t +1 Implica5ons Borrowers are marginal buyers of houses Varia5on in house prices only due to varia5on in discoun5ng

24 Two addi5onal simplifying assump5ons Rigid demand for houses by the lenders Linear u5lity in consump5on p t = β b [ h,c + ( 1 δ)p ] t +1 1 µ t θ mrs b,t +1 Implica5ons Borrowers are marginal buyers of houses Varia5on in house prices only due to varia5on in discoun5ng When collateral constraint binds ( μ > 0 ), θ p

25 Interac5on of borrowing and lending constraints Borrowing constraint: D b,t θ p t h b,t +1 Lending constraint: D l,t L D b,t L

26 Interac5on of borrowing and lending constraints Borrowing constraint: D b,t θ p t h b,t +1 Lending constraint: D l,t L D b,t L Which constraint binds is exogenous: L and θ endogenous: p t = β b 1 µ t θ mrs + ( 1 δ )p t +1 [ ]

27 Standard model without lending constraint R 1/β b Demand of funds 1/β l Supply of funds D b θ p h b

28 Non- binding lending constraint R 1/β b Demand of funds - R = 1/β l - Collateral constraint binding (µ>0) - p t = β b [ mrs + 1 δ 1 µ t θ ( )p t +1] 1/β l Supply of funds L D b θ p h b

29 Binding lending constraint R - R = 1/β b - Collateral constraint not binding - p t = β b mrs + ( 1 δ)p t +1 [ ] 1/β b Demand of funds 1/β l Supply of funds L D b θ p h b

30 Relaxing the lending constraint R 1/β b Demand of funds - R = 1/β l - Collateral constraint binding (µ>0) - p t = β b [ mrs + 1 δ 1 µ t θ ( )p t +1] 1/β l Supply of funds L D b θ p h b

31 Relaxing the lending constraint R 1/β b Demand of funds 1/β l Supply of funds L D b θ p h b

32 Relaxing the lending constraint R 1/β b Demand of funds 1/β l Supply of funds L D b θ p h b

33 Relaxing the lending constraint R 1/β b Demand of funds 1/β l Supply of funds L D b θ p h b

34 Relaxing the lending constraint R 1/β b Demand of funds 1/β l Supply of funds L D b θ p h b

35 Relaxing the lending constraint R 1/β b Demand of funds 1/β l Supply of funds L D b θ p h b

36 Relaxing the lending constraint R 1/β b Demand of funds 1/β l Supply of funds L D b θ p h b

37 Relaxing the lending constraint R 1/β b Demand of funds 1/β l Supply of funds L D b θ p h b

38 Relaxing the lending constraint R 1/β b Demand of funds 1/β l Supply of funds L D b θ p h b

39 Relaxing the lending constraint R 1/β b Demand of funds 1/β l Supply of funds L D b θ p h b

40 Outline Model Parameteriza5on Quan5ta5ve results Expansion in credit supply Loosening of collateral requirements

41 Parameter values Calibrate parameters to match Micro data: Survey of Consumer Finances Triennial detailed survey data of US households balance sheet

42 Taking the model to the data: Challenges In the data, many HHs have both mortgages and assets Iden5fy borrowers as agents with likle liquid financial assets in SCF Kaplan and Violante (2012) Standard mortgage contracts specify accumula5on of equity Replace simple collateral constraint with D b,t θ p t d t d t = 1 ρ ( )d t 1 + h b,t +1 1 δ ( )h b,t ρ = δ D b,t θ p t h b,t +1 ρ > δ HHs accumulate equity over 5me

43 Calibra5on R Economy in the 1990s 1/β b Demand of funds 1/β l Supply of funds L D b θ p h b

44 Quarterly calibra5on Parameter Value Source/Target Discount factor borrower (β b ) % real mortgage rate Discount factor lender (β l ) % decline in real mortgage rates ~ Krusell and Smith (1998) ~ Carroll et al. (2013) Depreciation (δ) Fixed Asset Tables Maximum LTV (θ) 0.80 Amortization (ρ) Median LTV of new or recently refinanced mortgages of liquidity constrained HHs in the SCF Evidence from Duca et al. (2012) Collateral constraint close to binding Mortgage-to-RE ratio of liquidity constrained HHs in the SCF (43%)

45 Outline Model Parameteriza5on Quan5ta5ve results Expansion in credit supply Loosening of collateral requirements

46 Experiment 1: Loosening of lending constraints Star5ng in 2000, simulate the effects of a gradual relaxa5on of mortgage lending constraints

47 Experiment 1: Loosening of lending constraints Star5ng in 2000, simulate the effects of a gradual relaxa5on of mortgage lending constraints Securi%za%on and tranching pension and money market funds gain access to mortgage lending Securi%za%on and tranching reduce banks capital requirements for mortgage lending

48 Securi5za5on over 5me Value of outstanding RMBSs relative to GDP '!" &#" )*+,-." /0,1)*+,-." &!" percent %#" %!" $#" $!" #"!" $((!" $((#" %!!!" %!!#" %!$!"

49 Experiment 1: Loosening of lending constraints Star5ng in 2000, simulate the effects of a gradual relaxa5on of mortgage lending constraints Securi5za5on and tranching pension and money market funds gain access to mortgage lending Securi5za5on and tranching reduce banks capital requirements for mortgage lending SIVs allow banks to move assets off balance sheets not coun5ng them towards risk- weighted capital Around 2003 regulators disregarded recommenda5ons to apply to them the same risk- weighted capital requirements as other types of assets, thereby facilita5ng massive regulatory arbitrage (Acharya and Schnabel, 2009)

50 Experiment 1: Loosening of lending constraints Star5ng in 2000, simulate the effects of a gradual relaxa5on of mortgage lending constraints Securi5za5on and tranching pension and money market funds gain access to mortgage lending Securi5za5on and tranching reduce banks capital requirements for mortgage lending SIVs allow banks to move assets off balance sheets not coun5ng them towards risk- weighted capital Around 2003 regulators disregarded recommenda5ons to apply to them the same risk- weighted capital requirements as other types of assets, thereby facilita5ng massive regulatory arbitrage (Acharya and Schnabel, 2009) Foreign capital inflows from Emerging Asia + oil producing countries mostly directed towards Government bonds and Agency MBSs

51 Experiment 1: Loosening of lending constraints Star5ng in 2000, simulate the effects of a gradual relaxa5on of mortgage lending constraints Securi5za5on and tranching pension and money market funds gain access to mortgage lending Securi5za5on and tranching reduce banks capital requirements for mortgage lending SIVs allow banks to move assets off balance sheets not coun5ng them towards risk- weighted capital Around 2003 regulators disregarded recommenda5ons to apply to them the same risk- weighted capital requirements as other types of assets, thereby facilita5ng massive regulatory arbitrage (Acharya and Schnabel, 2009) Foreign capital inflows from Emerging Asia + oil producing countries mostly directed towards Government bonds and Agency MBSs Experiment 5med to complete the transi5on in 2006

52 Experiment 1: Loosening of lending constraints L relative to income

53 Experiment 1: Loosening of lending constraints Annualized mortgage rate House prices Debt to GDP ratio Debt to real estate ratio

54 Experiment 2: Loosening of collateral requirements Standard model without lending constraints Simulate the effects of a gradual relaxa5on of collateral requirements

55 Standard model without lending constraint R 1/β b Demand of funds 1/β l Supply of funds θ p h b D b

56 Standard model without lending constraint R 1/β b Demand of funds 1/β l Supply of funds θ p h b D b

57 Experiment 2: Loosening of collateral requirements Standard model without lending constraints Simulate the effects of a gradual relaxa5on of collateral requirements θ from 0.8 to 1.02, to match the increase in HH debt of experiment 1

58 Experiment 2: Loosening of collateral requirements (θ) Annualized mortgage rate Lending constraints Collateral constraints: House prices Debt to GDP ratio Debt to real estate ratio

59 Experiment 3: Loosening of collateral requirements (θ) in a model with lending constraints

60 Experiment 3: Loosening of collateral requirements (θ) in a model with lending constraints R Economy in the 1990s 1/β b Demand of funds 1/β l Supply of funds L D b θ p h b

61 Experiment 3: Loosening of collateral requirements (θ) in a model with lending constraints R Economy at the end of /β b Demand of funds 1/β l Supply of funds L D b θ p h b

62 Experiment 3: Loosening of collateral requirements (θ) in a model with lending constraints R 1/β b Demand of funds 1/β l Supply of funds L D b θ p h b

63 Experiment 3: Loosening of collateral requirements (θ) in a model with lending constraints R 1/β b Demand of funds 1/β l Supply of funds L D b θ p h b

64 Experiment 3: Loosening of collateral requirements (θ) in a model with lending constraints R 1/β b Demand of funds 1/β l Supply of funds L D b θ p h b

65 Experiment 3: Loosening of collateral requirements (θ) in a model with lending constraints R 1/β b Demand of funds 1/β l Supply of funds L D b θ p h b

66 Experiment 3: Loosening of collateral requirements (θ) in a model with lending constraints R 1/β b Demand of funds 1/β l Supply of funds L D b θ p h b

67 Experiment 3: Loosening of collateral requirements (θ) in a model with lending constraints Annualized mortgage rate House prices Debt to GDP ratio Debt to real estate ratio

68 Conclusions Increased capacity to lend outward shiq in supply of credit Explains a large frac5on boom in house prices boom in HH debt decline in mortgage rates constant debt- to- collateral ra5o Loosening of collateral requirements not an important driving force. At odds with the behavior of mortgage rates house prices debt- to- collateral ra5o If anything, explains why prices started to fall

69 Conclusions What if collateral requirements depend on some of the factors shiqing also the supply of funds? Not a property of the standard model of collateralized debt Even in this case, our results suggest that credit supply has to shiq more than demand to account for the boom phase of the cycle

70 The story in words The apparent safety of the financial sector s collec5ve balance sheet was akributable to the fact that the biggest global banks had amassed vast quan55es of AAA- rated ( safe ) tranches backed by residen5al mortgages. These assets had historically been safer than similarly rated corporate loans. This was the principal reason behind their lower risk charge (by a factor of five) under the Basel capital requirements that were in place for European banks, for allowing the US commercial banks to park these in off- balance sheet vehicles with likle capital, and le:ng investment banks use internal models for risk management that largely ignored the tail risk of a secular housing collapse.

71 Risk- weighted capital ra5o In the United States, depository ins5tu5ons are subject to risk- based capital guidelines issued by the Fed. These guidelines are used to evaluate capital adequacy based primarily on the perceived credit risk associated with balance sheet assets, as well as certain off- balance sheet exposures such as unfunded loan commitments, lekers of credit, and deriva5ves and foreign exchange contracts. The risk- based capital guidelines are supplemented by a leverage ra5o requirement To be adequately (well) capitalized under federal bank regulatory agency defini5ons, a bank holding company must have a Tier- 1 capital ra5o of at least 4% (6%), a combined Tier- 1 and Tier- 2 capital ra5o of at least 8% (10%), and a leverage ra5o of at least 4% (5%)

72 Non- agency MBSs (Mayer)

73 Share of securi5zed mortgages (Krainer and Laderman, 2011)

74 Securi5za5on rates (Simkovic, 2013)

75 Mortgage spreads (1- year- ARM minus the FFR)

76 Senior Loan Officer Opinion Survey 2007-Q1

77 Experiment 2: Loosening of collateral requirements (a): Maximum LTV 7 x 10 3 (b): Speed of repayment

78 Experiment 2: Loosening of collateral requirements (ρ) Annualized mortgage rate Lending constraints Collateral constraints: Collateral constraints: House prices Debt to GDP ratio Debt to real estate ratio

79 Some literature Importance of borrowing constraints in the boom- bust of the 2000s Boom: Favilukis, Ludvigson, Van Nieuwerburgh (2013), Boz and Mendoza (2012), Garriga, Manuelli and Peralta- Alva (2012), Midrigan and Philippon (2011) Bust: Guerrieri and Lorenzoni (2012), Eggertsson and Krugman (2012), Hall (2012) We concentrate on barriers to lending and their interac5on with collateral constraints Constraints on composi5on of balance sheet of intermediaries Gertler and Kiyotaki (2010), Adrian and Shin (2010), Adrian and Boyarchenko (2012 and 2013), Dewachter and Wouters (2012), He and Krishnamurty (2013), Brunnermeier and Sannikov (2014), etc We concentrate on the link between the availability of credit, household debt and home prices