Where's The Smoking Gun? A Study of Underwriting Standards For US Subprime Mortgages
Where's The Smoking Gun? A Study of Underwriting Standards For US Subprime Mortgages
A Study of
Underwriting Standards for US Subprime
Mortgages
Geetesh Bhardwaj
Vanguard
and
Rajdeep Sengupta
Federal Reserve Bank of St. Louis
Bank of Finland/SUERF conference
4–5 June 2009
House of the Estates, Helsinki
The views expressed are those of the individual authors and do not necessarily reflect official positions
of the Federal Reserve Bank of St. Louis, the Federal Reserve System, or the Board of Governors.
Subprime Default Probabilities
0.7
0.6
0.5 2005
Probability of default
2006
0.4
2007 2002
2004
0.3 2001
2000
0.2
2003
0.1
0
4 8 12 16 20 24 28 32 36 40 44 48 52 56
Mortgage Age in Months
Motivation
The President’s Working Group on Financial Markets (March,
2008):
“The turmoil in financial markets was triggered by a dramatic
weakening of underwriting standards for U.S. subprime
mortgages, beginning in late 2004, and extending into early
2007.” –
Emphasis in the original
Study of subprime mortgage originations with a view to
examining and testing this hypothesis.
Dominant explanation:
Decline in Underwriting Standards
Implications:
1. Subprime mortgages of earlier vintages had robust
underwriting
2. Something went wrong after 2004
3. The change occurred within subprime originations (but not
the overall mortgage market).
• Underwriting: Summarizing the risk of default ex ante with the
purpose or approving or denying the loan application
• Borrower's observable characteristics at the time of
origination of the loan
Hard Information?
Decline in underwriting shown with hard information
(Demyanyk and van Hemert, 2008)
Stein (2002):
Data and Coverage
We use the data from LoanPerformance
Securitized subprime mortgages only
More than 9 million originations securitized as
subprime
Covers almost the entire market for subprime
mortgages that have been securitized, especially the
later vintages
Summary Trends: 1998‐2006
• Increase in the proportion of ARMs
• Increase in the proportion of Low‐doc loans
• Increase in the proportion of high LTV loans
• Increase in average FICO scores.
Evidence from Summary Statistics
Despite exposing themselves to more credit risk on
some borrower attributes (for example, by lowering
documentation requirements) …
lenders seem to have attempted to offset this by
increasing the average quality of borrowers (by raising
credit score requirements) to whom such loans were
made.
FICO and Low Documentation
70
60
50
40
30
20
Low Documentation
10 Low Doc with FICO < 620
0
2000 2001 2002 2003 2004 2005 2006 2007
FICO and Full Documentation
90
80
70
60
50
40
Full Documentation
30
Full Doc with FICO < 620
20
10
0
2000 2001 2002 2003 2004 2005 2006 2007
FICO and Default
Why did lenders choose higher FICO Scores?
Ex post, some industry experts have even faulted originators on
this account:
“... the crucial mistake many lenders made was relying on
FICO credit scores to gauge default risk, regardless of the
size of the down payment or the type of loan.”
The woman who called Wall Street's meltdown‐ Fortune Magazine, Aug. 4,
2008
Results: Determinants of Default
Our estimates show that a higher FICO score at
origination significantly lowers the probability of (ex
post) default.
The magnitude of this relationship is not significantly
different across vintages
Why? Because we account for the endogeneity
problem of including loan terms in a default regression
(1) Endogeneity of Mortgage Terms: Theory
Asymmetric Information:
Akerlof (1970), Rothschild and Stiglitz (1976), Brueckner (2000)
Advances in empirical contract theory:
Chiappori and Salanie, (2000); Chiappori et al. (2006)
Under both adverse selection and moral hazard, one
should observe a positive correlation conditional on
observables between risk (ex‐post default) and
coverage (LTV)
Theoretical Framework
Adverse selection: risk is exogenous and unobservable
Positive correlation: high‐risk agents are more likely to opt for
the mortgage contract with the lower downpayment but a
higher interest rate (Brueckner, 2000)
Moral Hazard: the reverse causality would generate the same
correlation
Borrowers buying into mortgages with higher LTV for any
unspecified or exogenous reasons are likely to exert less effort
to repay the loan and therefore become riskier
Endogeneity: Anecdotal Evidence
Mortgage Pricing Sheet, Option One Mortgage Corp.
Positive Correlation: Endogeneity Bias
0.2
Closing Rate Spread
0.16
LTV
0.12
0.08
0.04
0
2000 2001 2002 2003 2004 2005 2006 2007
Determinants of Default: Hazard Ratios
2000 2001 2002 2003 2004 2005 2006 2007
Owner
Occupied 0.8076*** 0.8022*** 0.8127*** 0.7825*** 0.7493*** 0.7725*** 0.7729*** 0.7611***
Refinance (Cash
Out) 0.7625*** 0.6605*** 0.6414*** 0.5419*** 0.5164*** 0.5015*** 0.5558*** 0.5738***
Refinance
(No Cash Out) 0.919*** 0.7927*** 0.7477*** 0.5829*** 0.5341*** 0.539*** 0.5975*** 0.5831***
Closing Rate
Spread 1.2269*** 1.2175*** 1.2453*** 1.2341*** 1.2261*** 1.2481*** 1.1923*** 1.3413***
Refinance
(Cash Out) 0.7821*** 0.7221*** 0.7047*** 0.5934*** 0.554*** 0.5464*** 0.6134*** 0.6855***
Refinance
(No Cash Out) 0.9664*** 0.8657*** 0.8286*** 0.6428*** 0.5852*** 0.6023*** 0.6635*** 0.7166***
Lenders compensate for the increase in the ex ante risk
of one borrower attribute by raising the requirement
standards along another dimension
Need to "aggregate" each borrower characteristic to
build a summary measure that fulfils a variety of
desirable conditions
Solution to this aggregation problem has proved
elusive
Counterfactual Analysis
Getting around the aggregation problem:
How would ex post default rates change if a mortgage
originated to a "representative borrower" in 2005 were
to be given a loan in 2001?
Counterfactual Analysis: Survival Plots
Counterfactual Analysis:
including mortgage terms
Conclusion on Counterfactual
A representative borrower in 2006 (likewise for 2005 and 2007)
had originated mortgages in 2001 and 2002, she would have
performed significantly better than representative borrowers of
vintages 2001 and 2002 respectively
We fail to reject the null hypothesis for 2003 vintages: No
statistically significant differences in the loan performances
between the representative borrowers of 2005 or 2007 vintages
and that of the 2003 vintage
So, what’s behind the high early
default rates on subprime
mortgages?
Subprime Default Probabilities
0.7
0.6
0.5 2005
Probability of default
2006
0.4
2007 2002
2004
0.3 2001
2000
0.2
2003
0.1
0
4 8 12 16 20 24 28 32 36 40 44 48 52 56
Mortgage Age in Months
High Early Prepayments (1)
Post Delinquency
Evidence of the use of prepayments as an exit
option following a delinquency
A sharp drop in prepayments on post‐2004 vintages
accompanied by a sharp rise in foreclosures (and
default)
Post‐delinquency Behavior of Owner Occupied
(up to two ears after origination)
30‐day Delinquency Rate
50 %
Prepayment Rate for delinquent loans
45 %
Foreclosure Rate for delinquent loans
40 %
35 %
Percentage
30 %
25 %
20 %
15 %
10 %
5 %
0 %
2000 2001 2002 2003 2004 2005 2006 2007
Vintage
Post‐delinquency Behavior of Owner Occupied
(up to two years after origination)
60‐day Delinquency Rate
40 %
Prepayment Rate for delinquent loans
35 % Foreclosure Rate for delinquent loans
30 %
Percentage
25 %
20 %
15 %
10 %
5 %
0 %
2000 2001 2002 2003 2004 2005 2006 2007
Vintage
High Early Prepayments (2)
Pre‐ Delinquency
Evidence of the use of prepayments as an exit
option even before registering a delinquency
A sharp drop in prepayments on post‐2004 vintages
accompanied by a sharp rise in delinquencies
Pre‐delinquency Behavior by Product Type
(up to loan age of 18 months )
Fixed
60 %
50 %
40 %
30 %
20 %
10 %
0 %
2000 2001 2002 2003 2004 2005 2006 2007
ARM2 ARM3
60 % 60 %
50 % 50 %
40 % 40 %
30 % 30 %
20 % 20 %
10 % 10 %
0 % 0 %
2000 2001 2002 2003 2004 2005 2006 2007 2000 2001 2002 2003 2004 2005 2006 2007
Pre‐delinquency Behavior by Occupancy
(up to loan age of 18 months )
Owner Occupied
60 %
50 %
40 %
30 %
20 %
10 %
0 %
2000 2001 2002 2003 2004 2005 2006 2007
Second Home Non‐Owner (Investor)
60 % 60 %
50 % 50 %
40 % 40 %
30 % 30 %
20 % 20 %
10 % 10 %
0 % 0 %
2000 2001 2002 2003 2004 2005 2006 2007 2000 2001 2002 2003 2004 2005 2006 2007
Pre‐delinquency Behavior by Purpose
(up to loan age of 18 months )
Purchase
60 %
50 %
40 %
30 %
20 %
10 %
0 %
2000 2001 2002 2003 2004 2005 2006 2007
Refi (Cash Out) Refi (No Cash Out)
60 % 60 %
50 % 50 %
40 % 40 %
30 % 30 %
20 % 20 %
10 % 10 %
0 % 0 %
2000 2001 2002 2003 2004 2005 2006 2007 2000 2001 2002 2003 2004 2005 2006 2007
High Prepayment Rates on Earlier Vintages
Subprime mortgage contracts were designed as “bridge‐
finance” providing temporary credit accommodation
Prepayments were largely sustained by the boom in house
prices in the United States from 1995 to 2006.
Conclusion
Sparse evidence of a dramatic weakening of lending
standards within the subprime market.
Deterioration in underwriting post‐2004 cannot be the
explanation for collapse of subprime mortgage
market
One cannot rule out that underwriting standards for
subprime loans were poor to begin with.
Default Probabilities
Prepayment Probabilities
Growth rate of House Prices
Sustainable?