ÝÐ Consumption and House Prices in the Great Recession: Model Meets Evidence Greg Kaplan Chicago Kurt Mitman IIES - Stockholm Gianluca Violante Princeton Ò Å Ø Ò Ó Ø ÓÒÓÑ ØÖ ËÓ ØÝ
The QuestionyÝÐ Relative House Price 0.3 Boom Logs (1997:Q1 = 0) 0.2 0.1 0 Bust -0.1-0.2 1995 2000 2005 2010 2015 Year
The QuestionyÝÐ Relative House Price 0.3 Boom Logs (1997:Q1 = 0) 0.2 0.1 0 Bust -0.1-0.2 1995 2000 2005 2010 2015 Year What caused the boom and bust in house prices?
Two ViewsyÝÐ 1. Credit view Availability of credit to marginal borrowers determines demand for housing and house prices Financial deregulation in early 2000s (i.e., PLS) led to unsustainable lending to subprime low-income borrowers
Two ViewsyÝÐ 1. Credit view Availability of credit to marginal borrowers determines demand for housing and house prices Financial deregulation in early 2000s (i.e., PLS) led to unsustainable lending to subprime low-income borrowers 2. Expectations view Waves of optimism and pessimism affect desire to borrow, housing demand and house prices Middle class (i.e., prime borrowers) crucial to the story
Two ViewsyÝÐ 1. Credit view Availability of credit to marginal borrowers determines demand for housing and house prices Financial deregulation in early 2000s (i.e., PLS) led to unsustainable lending to subprime low-income borrowers 2. Expectations view Waves of optimism and pessimism affect desire to borrow, housing demand and house prices Middle class (i.e., prime borrowers) crucial to the story What do the microdata say?
Micro DatayÝÐ Credit view: credit growth (in boom) and defaults (in bust) concentrated among marginal borrowers Main-Sufi: influential body of work supporting this hypothesis
Micro DatayÝÐ Credit view: credit growth (in boom) and defaults (in bust) concentrated among marginal borrowers Main-Sufi: influential body of work supporting this hypothesis Recently, evidence in favor of credit supply view has been challenged by Albanesi et al., Adelino et al., Foote et al.!&& '& #& "& (( (% (% (% *& (& )& %& $# %& %& $' $&!& &!"!#!"!# $&&% $&&) $&&( $&&* +,-./0/**& **&/1/+,-./0/"$& +,-./2/"$& Share of originations Share of delinquencies
Micro DatayÝÐ Credit view: credit growth (in boom) and defaults (in bust) concentrated among marginal borrowers Main-Sufi: influential body of work supporting this hypothesis Recently, evidence in favor of credit supply view has been challenged by Albanesi et al., Adelino et al., Foote et al.!&& '& #& "& (( (% (% (% *& (& )& %& $# %& %& $' $&!& &!"!#!"!# $&&% $&&) $&&( $&&* +,-./0/**& **&/1/+,-./0/"$& +,-./2/"$& Share of originations Share of delinquencies Suggestive evidence: need measurement through structural models
Equilibrium Models of the Credit ViewyÝÐ Favilukis-Ludvigson-van Nieuwerburgh (2015); Justiniano-Primiceri-Tambalotti (2015); Greenwald (2016) Successful in generating large house price movements
Equilibrium Models of the Credit ViewyÝÐ Favilukis-Ludvigson-van Nieuwerburgh (2015); Justiniano-Primiceri-Tambalotti (2015); Greenwald (2016) Successful in generating large house price movements What does it take for looser credit to push up house prices? 1. Large effect of credit shocks on housing risk premium 2. Many households constrained in their housing choice
Equilibrium Models of the Credit ViewyÝÐ Favilukis-Ludvigson-van Nieuwerburgh (2015); Justiniano-Primiceri-Tambalotti (2015); Greenwald (2016) Successful in generating large house price movements What does it take for looser credit to push up house prices? 1. Large effect of credit shocks on housing risk premium 2. Many households constrained in their housing choice Model features that deliver these outcomes: 1. Short-term debt & no default: makes housing very risky 2. No rental market: many households that want to consume more housing, but can t
Our PaperyÝÐ Equilibrium model with rental market and long-term mortgages Aggregate shocks: income, credit, and beliefs Parameterize to cross-sectional and life-cycle facts Compare to time-series on: house prices, rent-price ratio, home ownership, leverage, and foreclosures Decompose the role of each shock Compare with new micro evidence Study transmission of house prices to consumption
Model: Household and Financial SectorsyÝÐ OLG with two phases in lifecycle: work and retirement CES utility over ND consumption (1 φ) and housing (φ) Idiosyncratic uninsurable earnings shocks Saving in risk-free bonds with exogenously fixed interest rate Housing can be bought at p h (sold s.t. transaction cost) or rented at ρ Long-term mortgages (to be repaid before death), with cash-out refi option, defaultable, competitively priced by financial intermediaries At origination: max LTV and max PTI constraints (λ m, λ π ) and origination cost (proportional to loan size) κ m HELOCs: one-period debt, non defaultable (λ b )
Closing the ModelyÝÐ Final good sector Linear technology in labor with productivity Z w = Z Construction sector Housing permits + labor aggregate housing investments I(p h ) Rental sector Frictionless conversion of rental units into OO units and viceversa Zero-profit condition yields equilibrium rental rate ρ Government Taxes workers (with mortgage interest deduction) and properties, sells land permits, and pays SS benefits to retirees
8.0 6.0 4.0 2.0 1 0 Lifecycle Profiles of Ownership and LeverageyÝÐ 1 0.8 Leverage-Model Leverage-Data 0.6 0.4 0.2 0 Home Ownership - Model Home Ownership - Data 30 40 50 60 70 80 Age 3040 50 60 70 80 Age The model replicates: the steep rise in home ownership from age 25 to 50 the fact that home ownership remains flat during retirement and also the sharp decline in leverage over the life cycle
Aggregate ShocksyÝÐ Aggregate labor income: Z Credit conditions: (λ m, λ b, λ π ) and κ m
Aggregate ShocksyÝÐ Aggregate labor income: Z Credit conditions: (λ m, λ b, λ π ) and κ m Beliefs / News about future housing demand Three regimes for φ (share of housing services in u): µ φ L : low housing share and unlikely transition to φ H µ φ L : low housing share and likely transition to φ H µ φ H : high housing share
Aggregate ShocksyÝÐ Aggregate labor income: Z Credit conditions: (λ m, λ b, λ π ) and κ m Beliefs / News about future housing demand Three regimes for φ (share of housing services in u): µ φ L : low housing share and unlikely transition to φ H µ φ L : low housing share and likely transition to φ H µ φ H : high housing share Boom-Bust: shift from (a) to (b), and back to (a)
Aggregate ShocksyÝÐ Aggregate labor income: Z Credit conditions: (λ m, λ b, λ π ) and κ m Beliefs / News about future housing demand Three regimes for φ (share of housing services in u): µ φ L : low housing share and unlikely transition to φ H µ φ L : low housing share and likely transition to φ H µ φ H : high housing share Boom-Bust: shift from (a) to (b), and back to (a) Calibration of news shock: use data on expectations... but residual
Household Expectations in the ModelyÝÐ 0.07 ExpectedPriceGrowth 0.06 0.05 0.04 0.03 0.02 0.01 2000 2005 2010 2015 Year
Household Expectations in the ModelyÝÐ 0.07 ExpectedPriceGrowth 0.06 0.05 0.04 0.03 0.02 0.01 2000 2005 2010 2015 Year For boom years, survey evidence in Case-Shiller-Thompson shows US households expected house price to grow 5-10 pct per year
House PricesyÝÐ HousePrice 1.3 1.2 1.1 1 0.9 0.8 Benchmark Belief Only Income Only Credit Only 2000 2005 2010 2015 Year
House PricesyÝÐ HousePrice 1.3 1.2 1.1 1 0.9 0.8 Benchmark Belief Only Income Only Credit Only 2000 2005 2010 2015 Year Belief shock accounts for all boom-bust in house prices
House PricesyÝÐ HousePrice 1.3 1.2 1.1 1 0.9 0.8 Benchmark Belief Only Income Only Credit Only 2000 2005 2010 2015 Year Belief shock accounts for all boom-bust in house prices Households unconstrained with respect to housing consumption
Rent-Price RatioyÝÐ Rent-PriceRatio 1.1 1 0.9 0.8 0.7 Bench Belief Inc Credit Data 2000 2005 2010 2015 Year ( ) 1 δh τ ρ = ψ+p h h [ ] 1+r b E ph p h Belief about future appreciation shared by investment company
Home Ownership RateyÝÐ 1.1 1.05 Bench Belief Inc Credit Data 1 0.95 2000 2005 2010 2015 Year Cheap credit drives rise in home ownership Households constrained with respect to their tenure choice
Explaining the Effects of Credit ShocksyÝÐ Why looser/tighter credit does not affect housing demand? Defaultable long-term debt: housing risk premium is small Rental market: buyers are not constrained in housing choice
Explaining the Effects of Credit ShocksyÝÐ Why looser/tighter credit does not affect housing demand? Defaultable long-term debt: housing risk premium is small Rental market: buyers are not constrained in housing choice Why is rise in home ownership disconnected from house prices? Renters buy houses of similar size of those they rented It s the current home owners who upsize and push up demand
Explaining the Effects of Credit ShocksyÝÐ Why looser/tighter credit does not affect housing demand? Defaultable long-term debt: housing risk premium is small Rental market: buyers are not constrained in housing choice Why is rise in home ownership disconnected from house prices? Renters buy houses of similar size of those they rented It s the current home owners who upsize and push up demand If hh s already consume optimal amount of housing, why buy more? Housing is both a consumption good and an asset Many households buy larger houses to realize expected capital gains
Leverage (debt/house value) yýð 1.8 1.6 1.4 1.2 Bench Belief Inc Credit Data Leverage 1 0.8 2000 2005 2010 2015 Year Credit loosening is crucial to maintain constant leverage pre-boom
Endogenous Credit Boom Through BeliefsyÝÐ
Endogenous Credit Boom Through BeliefsyÝÐ 0.16 0.14 Mortgage Rate 0.12 0.1 shift in lender beliefs 0.08 0.06 0.04 0.4 0.5 0.6 0.7 0.8 0.9 1 Loan-to-Value Ratio Lender s optimistic beliefs lower expected default rates lower mortgage rates, especially for subprime borrowers
Foreclosure Rate yýð 0.04 0.03 0.02 Bench Belief Inc Credit Data Foreclosurerate 0.01 0 2000 2005 2010 2015 Year Foreclosure spike due to interaction between optimistic belief and looser credit
An Economy where Credit Matters for House PricesyÝÐ
An Economy where Credit Matters for House PricesyÝÐ No rental mkt & no default: relaxation/tightening of LTV moves prices 1.3 1.2 1.1 1 House Price 1.8 1.6 1.4 1.2 Leverage Credit Data 0.9 1 0.8 2000 2010 Year 0.8 2000 2010 Year Counterfactual surge in leverage during the boom (flat in the data)
Comparison with New EvidenceyÝÐ Fact: credit growth occurred throughout distrib. of FICO scores
Comparison with New EvidenceyÝÐ Fact: credit growth occurred throughout distrib. of FICO scores Model: rank households by default probability at origination
Comparison with New EvidenceyÝÐ Fact: credit growth occurred throughout distrib. of FICO scores Model: rank households by default probability at origination Share of Debt 0.2.4.6.8 Shares of Originated Mortgage Debt 2001 2007 1 2 Default Risk (1=Above Median, 2=Below Median)
Comparison with New EvidenceyÝÐ Fact: credit growth occurred throughout distrib. of FICO scores Model: rank households by default probability at origination Share of Debt 0.2.4.6.8 Shares of Originated Mortgage Debt 2001 2007 1 2 Default Risk (1=Above Median, 2=Below Median) Share of foreclosures in the bottom quartile of default probability at origination falls from 100 pct in 2007 to 40 pct in 2011
ConsumptionyÝÐ 1.1 Consumption 1.05 1 0.95 2000 2005 2010 2015 Year House prices explain 1/2 of boom and bust in C (rest is income)
ConsumptionyÝÐ 1.1 1.05 1 0.95 Consumption 2000 2005 2010 2015 Year.3.2.1 0.1 Change in Log Consumption Renters Owners 0.1.2.3.4 Housing Share of Total Wealth House prices explain 1/2 of boom and bust in C It s because of a wealth effect, i.e. through household balance sheet
Summary: What Did We Learn from the Model?yÝÐ Shift in expected house appreciation drives the boom-bust in p h Credit important for home ownership, leverage, and foreclosures Rental market + long-term mortgages are the key model features Model tells us that micro evidence and aggregate time series agree Changes in p h transmit to C through balance sheet effects
Summary: What Did We Learn from the Model?yÝÐ Shift in expected house appreciation drives the boom-bust in p h Credit important for home ownership, leverage, and foreclosures Rental market + long-term mortgages are the key model features Model tells us that micro evidence and aggregate time series agree Changes in p h transmit to C through balance sheet effects Thanks!