1 DYNAMICS OF HOUSING DEBT IN THE RECENT BOOM AND BUST Manuel Adelino (Duke) Antoinette Schoar (MIT Sloan and NBER) Felipe Severino (Dartmouth)
2 Motivation Lasting impact of the 2008 mortgage crisis on US the economy Households significantly increased their stock of mortgage debt over the boom period Once the housing market slowed down, large increases in defaults particularly for prime borrowers These losses impaired banks and the shadow banking system, and led to the Great Recession Key question: What led to an increase in mortgage debt?
3 What drove credit expansion? Subprime view Financial innovation and deregulation created perverse incentives in the financial sector, which provided unsustainable credit, especially to low income, low credit score borrowers These distorted credit flows led to higher house prices and mortgage crisis (e.g., Mian and Sufi, 2009, 2016) Expectations view Inflated house prices (expectations) led banks to underestimate the potential for losses. Banks lent against high house values Borrowers increase demand for housing, exploit increased supply Savings glut and lower interest rates might be triggers for original increase in house prices
CREDIT EXPANSION Credit Expansion Housing Boom Real Economic Activity
MULTIPLE CHANNELS OF CREDIT EXPANSION Credit Expansion Expectation Shock Housing Boom Real Economic Activity Credit expansions and expectations are endogenous variables
6 Empirical Challenge Difficulty in testing the two views: Explanations are not mutually exclusive, may reinforce each other Credit screening might go down as a result of high house price expectations, since lenders feel protected by increased prices. But more access to credit can then feed into higher prices Proper diagnosis of the origins of the mortgage crisis has different implications for policy Fixing incentives and monitoring in banks vs. macro-prudential regulation
5 Summary of Results I House price expectations and increased collateral values played an important role in the credit expansion Flow (new purchase mortgages) and stock of credit increased across all income groups over the boom no cross sectional dislocation of credit flows towards poor households Credit expansion was significantly larger in areas with high house price growth CLTVs stayed stable over the boom = lending against V
6 Summary of Results II Quicker buying and selling of homes in the boom Attempt of households to buy into increasing house prices Stock of mortgage debt resets up when household buys new home Homeownership rates increase during boom for all households except for the poorest Credit did not expand to allow poor borrowers to become homeowners Large increase in cost of owning relative to renting during the boom Share of defaults by middle income and middle credit score borrowers increases in the bust.
9 Data Household-level survey data from American Community Survey (public use microdata sample) Housing costs for owners and renters Homeownership (confirmed with data from American Housing Survey and Current Population Survey) Value-to-income Home Mortgage Disclosure Act (HMDA) Balance and location of individual mortgages originated in the US Corelogic (formerly Dataquick) Ownership transfers of residential properties from deeds and assessors records (1996-2012) Allows us to compute LTVs Mortgage performance from LPS (formerly McDash) 5% random sample IRS income at the zip code level. House prices from Zillow.
10 Credit Flows, Stock, and DTI No significant cross-sectional dislocations in credit flows or stock of household debt across income or FICO bins Flow of (purchase) mortgages increases across all income groups (HMDA data) Stock of mortgage debt also increases proportionally (SCF data) Cost of owning a house increases predominantly in areas with high house price growth (ACS data) Cost of owning relative tor renting goes up dramatically in the boom and collapses after the crisis, especially in boom states
11 Flow of new purchase mortgages $ bn, HMDA data, IRS income quintiles 1,000 900 800 700 600 500 400 300 200 100 0 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 Quintile 1 (lowest) Q2 Q3 Q4 Quintile 5
12 Flow of new purchase mortgages Shares by quintile, HMDA data, IRS income quintiles 100% 90% 80% 34.6 34.8 34.5 33.2 31.4 30.3 33.5 32.6 32.8 35.5 37.2 38.7 38.7 36.9 35.3 70% 60% 50% 25.5 25.3 25.6 25.2 24.9 24.4 24.1 24.9 25.4 25.2 24.8 24.9 25.0 25.1 25.4 40% 30% 17.2 17.1 17.4 17.6 17.8 18.0 17.6 18.2 18.2 17.4 16.9 16.5 16.5 17.0 17.4 20% 10% 0% 13.2 13.1 13.0 13.5 14.4 15.0 14.1 14.2 13.9 12.9 12.6 11.9 11.8 12.5 13.0 9.5 9.6 9.6 10.6 11.5 12.3 10.7 10.1 9.8 8.9 8.5 8.0 8.0 8.6 8.9 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 Quintile 1 (lowest) Q2 Q3 Q4 Quintile 5
13 Debt-to-income Owners with positive debt, SCF data 4.0 3.5 3.0 2.5 2.0 1.5 1.0 0.5 0.0 2001 2004 2007 Bottom quintile 2 3 4 Top quintile
14 Cost of owning, boom and non-boom states Recent movers, ACS data Boom states (above median HPA) Non-boom states 10 10 8 8 6 6 4 4 2 2 0 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 0 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015-2 -2-4 -4-6 -6 Quintile 2 (second lowest) Q3 Q4 Quintile 5 Quintile 2 (second lowest) Q3 Q4 Quintile 5
15 Difference between cost of owning and renting Recent movers, ACS data 8 6 4 2 0 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015-2 -4-6 -8 Quintile 2 (second lowest) Q3 Q4 Quintile 5
16 Homeownership rates Increases in homeownership for all except the lowest income households, ACS data Confirm the fact with other Census sources (AHS, CPS) Similar pattern holds across high and low house price states
17 Homeownership Rates (all states) ACS data 0.08 0.06 0.04 0.02 0 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015-0.02-0.04-0.06-0.08 Quintile 1 (lowest) Q2 Q3 Q4 Quintile 5
Homeownership Rates ( Boom states ) ACS data
19 Homeownership rate (historical) Census and ACS data 0.05 0.04 0.03 0.02 0.01 0-0.01 1980-1990 1990-2000 2000-2005 2005-2010 2010-2015 -0.02-0.03-0.04-0.05 Quintile 1 (lowest) Q2 Q3 Q4 Quintile 5
20 Speed of Flipping Homes (Churn) Speed at which homeowners sold and bought homes Was high in early 2000s and peaked in 2006: Approximately 8% of homeowners had bought their home in the previous 12 months Rate drops to 4% in the crisis, only increases after 2012 Churn rates are higher for high and middle vs. low income households
21 Churn Share of owners moving in last 12 months, ACS data 0.1 0.08 0.06 0.04 0.02 0 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 Quintile 1 (lowest) Q2 Q3 Q4 Quintile 5
22 Loan-to-value Distribution of Combined LTV at origination Very stable over the boom period, even when we condition on mortgage size. No pronounced differences in areas with high and low house price growth.
Loan-to-value New home purchases, Corelogic (formerly Dataquick) 23
Loan-to-value in boom states New home purchases, Corelogic (formerly Dataquick) 24
25 Defaults Middle-income and prime borrowers sharply increase their share of total delinquencies in the crisis Particularly in areas with large house price increases and subsequent drops
Share of Delinquent Mortgages 3 Years Out, by Cohort Value Weighted, LPS data 26
Share of Delinquent Mortgages 3 Years Out, by house price growth in 2002-06
Who are the marginal borrowers?
29 Stability of employment Households with full-time employment, ACS data 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 Quintile 1 (lowest) Q2 Q3 Q4 Quintile 5
Loan Approval Rates 1.00 0.90 0.80 0.70 0.60 0.50 0.40 0.30 0.20 0.10 0.00 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 Q1 Q2 Q3 Q4 Q5
31 Conclusion House prices played a major role in the expansion of credit and delinquencies Credit expansion and subsequent defaults significantly stronger in areas with high house price growth Stable CLTV levels at origination over the boom suggest lenders lent mechanically against increasing house price values (V). Post crisis policy response Mortgage credit and homeownership rates to lower income borrowers drops dramatically after the crisis Changes in policy targeted at lower income households may have prevented them buying when prices were low, without improving the stability of the mortgage market.