Default Rates on Prime and Subprime Mortgages: Differences & Similarities - Published by the Consumer and Community Affairs Division
←
→
Page content transcription
If your browser does not render page correctly, please read the page content below
Published by the Consumer and Community Affairs Division September 2010
Default Rates on
Prime and Subprime
Mortgages:
Differences & Similarities
Profitwise News and Views welcomes article proposals and comments from bankers, com- munity organizations, and other readers. It is mailed (either electronically or via U.S. mail) at no charge to state member banks, financial holding companies, bank holding companies, government agencies, nonprofit organizations, academics, and community economic develop- ment professionals. You may submit comments, proposals, or request a subscription by writing to: Profitwise News and Views Consumer and Community Affairs Division Federal Reserve Bank of Chicago 230 South LaSalle Street Chicago, IL 60604-1413 or CCA-PUBS@chi.frb.org The material in Profitwise News and Views is not necessarily endorsed by and does not nec- essarily represent views of the Board of Governors of the Federal Reserve System or the Federal Reserve Bank of Chicago. © 2010 Federal Reserve Bank of Chicago Profitwise News and Views articles may be reproduced in whole or in part, provided the articles are not reproduced or distributed for commercial gain and provided the source is appropriately credited. Prior written permis- sion must be obtained for any other reproduc- tion, distribution, republication, or creation of derivative works of Profitwise News and Views articles. To request permission, please e-mail or write to the address indicated above. Advisor Alicia Williams Managing Editor Michael V. Berry Assistant Editor/Production Manager Mary Jo Cannistra Contributing Editor Jeremiah Boyle Compliance Editor Steven W. Kuehl Economic Research Editor Robin Newberger Economic Development Editor Harry Pestine Designer Katherine Theoharopoulos Production Specialist Edwina Davis Visit the Web site of the Federal Reserve Bank of Chicago at: www.chicagofed.org
RESEARCH REVIEW
Default rates on prime and subprime
mortgages: differences and similarities
by Gene Amromin and Anna L. Paulson
Introduction and summary historically high default rates? How have Home prices clearly play a key role in
recent trends in home prices and households’ ability and desire to honor
For the past several years, the news economic conditions affected mortgage their mortgage commitments. One of the
media have carried countless stories markets? One of the things we want to things we consider in this article is
about soaring defaults among subprime consider, specifically, is whether prime and whether performance of prime and
mortgage borrowers. Although concern subprime loans responded similarly to subprime loans responded similarly to
over this segment of the mortgage home price dynamics. rapid home price appreciation from 2002
market is certainly justified, subprime to 2005, and the sharp reversal in home
mortgages only account for about one- Figure 1, panel A summarizes default prices beginning in 2006.
quarter of the total outstanding patterns for prime loans; panel B reports
mortgages in the United States. The similar trends for subprime loans using In this article, we make use of loan-
remaining 75 percent are prime loans loan-level data from LPS Applied level data on individual prime and
that are made to borrowers with good Analytics. Each line in the figure shows subprime loans made between January
credit, who fully document their income the cumulative default experience for 1, 2004, and December 31, 2007, to do
and make traditional down payments. loans originated in a given year as a two things: 1) analyze loan (and borrower)
While default rates on prime loans are function of how many months it has been characteristics and the default
significantly lower than those on since the loan was made. Several patterns experience for prime and subprime loans;
subprime loans, they are also increasing are worth noting. First, the performance of and 2) estimate empirical relationships
rapidly. For example, among prime loans both prime and subprime loans has gotten between home price appreciation, loan
made in 2005, 2.2 percent were 60 substantially worse, with loans made in and borrower characteristics, and the
days or more overdue 12 months after 2006 and 2007 defaulting at much higher likelihood of default. These estimates
the loan was made (our definition of rates. The default experience among allow us to quantify which factors make
default). For loans made in 2006, this subprime loans started deteriorating default more or less likely and to examine
percentage nearly doubled to 4.2 earlier, with rates being higher for loans how default sensitivity varies over time
percent, and for loans made in 2007 it made in 2005 than in 2004. Defaults and across prime and subprime loans.
rose by another 20 percent, reaching among subprime loans are, of course,
By looking at prime and subprime
4.8 percent. By comparison, the much higher than defaults among prime
loans together, we hope to refine the
percentage of subprime loans that had loans – note the difference in scales of
possible explanations for the ongoing
defaulted after 12 months was 14.6 the two panels. However, the deterioration
mortgage crisis.2 Both prime and
percent for loans made in 2005, 20.5 in the performance of prime loans
happened more rapidly than it did for subprime loans have seen rising defaults
percent for loans made in 2006, and
subprime loans. For example, the in recent years, as well as very similar
21.9 percent for loans made in 2007. To
percentage of prime loans in default patterns of defaults, with loans made in
put these figures in perspective, only
during their first 12 months grew by 95 more recent years defaulting at higher
1.4 percent of prime loans and less than
percent between 2005 and 2006. Among rates. Because of these similarities, it
7 percent of subprime originated in
subprime loans it grew by a relatively seems reasonable to expect that a
2002 defaulted within their first 12
modest 53 percent. successful explanation of the subprime
months.1 How do we account for these
crisis – the focus of most research to
Profitwise News and Views September 2010 1
RESEARCH REVIEW
date – should also explain the patterns of
defaults we observe in prime mortgages. Figure 1.A: Cumulative mortgage default rate of prime first-lien loans
(as a function of loan age, by year of origination)
Loan and borrower characteristics 0.12
In this section, we discuss trends in 0.10
loan and borrower characteristics, as well
0.08
Default Rate
as the default experience for prime and
subprime loans for each year from 2004
0.06
through 2007.
0.04
Data
0.02
The loan-level data we use come from
LPS Applied Analytics (LPS), which
0
gathers information from a number of
loan servicing companies.3 The most Months since mortgage origination
recent data include information on 30
million loans, with smaller (but still very SOURCE: LPS Applied Analytics.
large) numbers of loans going back in
time. The data cover prime, subprime, and Figure 1.B: Cumulative mortgage default rate of subprime first-lien loans
Alt-A loans,4 and include loans that are (as a function of loan age, by year of origination)
privately securitized, loans that are sold
to the GSEs, and loans that banks hold 0.40
on their balance sheets. 0.35
The total number of loans originated 0.30
in the LPS data in each year of the
Default Rate
0.25
period we study ranges from a high of
0.20
6.2 million in 2005 to a low of 4.3
million in 2007. The mortgage servicers 0.15
reporting to LPS Applied Analytics give 0.10
each loan a grade of A, B, or C, based
0.05
on the servicer’s assessment of whether
the loan is prime or subprime. We treat 0
A loans as prime loans, and B and C
Months since mortgage origination
loans as subprime. To make the analysis
tractable, we work with a 1 percent SOURCE: LPS Applied Analytics.
random sample of prime loans made
between January 1, 2004, and score. 5 Some of the loan characteristics in the 12 months following origination.
December 31, 2007, for a total of that we analyze include the loan amount; We focus on the first 12 months, rather
68,000 prime loans, and a 10 percent whether the loan is a fixed-rate or than a longer period, so that loans made
random sample of subprime loans made variable-rate mortgage; whether the in 2007 can be analyzed the same way
during the same time period, for a total loan was fully documented; the ratio of as earlier loans, as our data are
of 62,000 subprime loans. the loan amount to the value of the complete through the end of 2008.6
home at origination (LTV); whether the
The LPS data include a wide array of We augment the loan-level data with
loan was intended for home purchase or
variables that capture borrower and loan information on local economic trends and
refinancing and, in case of the latter,
characteristics, as well as monthly loan trends in local home prices. The
whether it involved equity extraction (a
performance status. In terms of economic variable we focus on is the
“cash-out refinance”); and whether the
borrower characteristics, important local unemployment rate that comes from
loan was held on the originating bank’s
variables include the debt-to-income the U.S. Bureau of Labor Statistics
portfolio, sold to one of the GSEs, or
ratio of the borrower (DTI) and the monthly zip-code-level statistics.
privately securitized. The outcome
borrower’s creditworthiness, measured Quarterly data on home prices is
variable that we focus on is whether the
by their FICO (Fair Isaac Corporation) available by metropolitan statistical area
loan becomes 60 days or more past due
2 Profitwise News and Views September 2010RESEARCH REVIEW
(MSA) from the Federal Housing Finance bigger cushion against future price loans very dependent on the ability to
Agency (FHFA)—an independent federal declines. In this way, greater prevalence refinance.9 In contrast, loans to prime
agency that is the successor to the of cash-out refinancing transactions borrowers are predominantly made as
Office of Federal Housing Enterprise may be indicative of the increasing risk fixed-rate contracts (about 75 percent of
Oversight (OFHEO) and other in the universe of existing loans. all prime loans), and the majority of prime
government entities.7 We use the FHFA ARMs have introductory periods of five to
As indicated in table 1, mortgage
all transactions House Price Index (HPI) seven years.
servicers assign many refinancing
that is based on repeat sales information.
transactions to the ambiguous category One oft-mentioned culprit for the
Trends in loan and borrower characteristics of “refinancing with unknown cash-out.” subprime crisis is the growth of lenders
Nevertheless, among prime loans made who followed the “originate-to-distribute
Many commentators (see, for in 2004, 12 percent were known to model” (see, for example, Keys et al.,
example, Demyanyk and Van Hemert, involve cash-outs. By 2005, this 2010, and Calomiris, 2008). These
2009) have noted that subprime percentage had risen to about 21 lenders sold virtually all of the mortgages
lending standards became more lax percent, and it remained at this level they made, typically to private
during the period we study, meaning through 2007. For subprime loans made securitizers. Because such lenders do
that the typical borrower may have in 2004, 35 percent involved cash-outs; not face a financial loss if these
received less scrutiny over time and it for those made in 2005, 43 percent; for mortgages eventually default, they have
became easier for borrowers to get those made in 2006, 47 percent; and for relatively little incentive to screen and
loans overall, as well as to get larger those made in 2007, a staggering 57 monitor borrowers. In addition to selling
loans. Table 1 summarizes loan percent. Put differently, cash-out loans loans to private securitizers, the lenders
characteristics for each year from accounted for at least 82 percent can hold loans on their own portfolios or
2004 through 2007 for both prime and
(0.575/0.7) of all subprime refinancing sell them to one of the GSEs. However,
subprime mortgages.
transactions in 2007! Another loan only loans that meet certain criteria
Consistent with prior work, we also characteristic that might be an important (borrower with a FICO score of at least
document declining borrower quality determinant of subsequent defaults is 620, loan value of less than $417,000,
over time in the subprime sector. For whether the interest rate is fixed for the and a loan-to-value ratio [LTV] of 80
example, whereas the average FICO life of the contract or allowed to adjust percent or less) can generally be sold to
score for subprime borrowers in 2004 periodically (as in adjustable-rate the GSEs. Most subprime loans cannot
was 617, it had declined to 597 by mortgages, or ARMs). When an ARM be sold to GSEs and must be either
2007. 8 By contrast, when we look at resets after the initial defined period privately securitized or held in portfolio.
prime loans, the decline in lending (which may be as short as one year or as
The extent of loan securitization is one
standards is less obvious. The average long as seven), the interest rate and,
of the striking facts in table 1. Recall that
FICO score among prime borrowers consequently, the monthly mortgage
the LPS data comprised loans serviced
was 710 in 2004 and 706 in 2007, a payment, may go up substantially. Higher
by the large mortgage servicers. As a
decline of less than 1 percent. payments may put enough stress on
result, LPS overstates the actual extent
some families that they fall behind on
Our data also allow us to look at the of securitization somewhat, as it is more
their mortgages. While these loans seem
prevalence of different mortgage common for smaller banks to hold loans
attractive at first because of low
transactions, such as purchases or in portfolio and also to service them
introductory interest rates (and low initial
refinancings. We are particularly internally. That being said, the LPS data
payments), they expose borrowers to
interested in refinancings that extract indicate that within the first month of
additional risk if interest rates go up or if
home equity (a cash-out refinance). By origination, about half of prime
credit becomes less available in general.
taking out equity in a refinancing, a mortgages made in 2004 remained in
household may end up being more Among subprime mortgages, ARMs their originators’ portfolios. This figure
vulnerable to future home price accounted for 73 percent in 2004, 69 declined to about 40 percent in each of
declines, especially if its new mortgage percent in 2005, and 62 percent in the subsequent years in our data. By
has a high loan-to-value ratio. To the 2006. By 2007, the ARM share had comparison, many fewer subprime loans
extent that the practice of cash-out fallen to 39 percent, since the availability were retained by their originators even
refinancing was common over the of these types of loans declined in the for the first month: just over 40 percent
period in our study, the increases in second half of the year. Importantly, of loans made in 2004 and less than 30
home prices may be associated with nearly all subprime ARMs have percent made in the following years.
constant or even increasing leverage introductory periods of three years or However, by the end of the first year
rather than with safer loans and a less, which makes borrowers with these since origination, the share of originated
Profitwise News and Views September 2010 3RESEARCH REVIEW
Table 1: Loan characteristics at origination
Prime Loans Subprime Loans
2004 2005 2006 2007 2004 2005 2006 2007
% default in first 12 months 2.43% 2.39% 4.33% 4.93% 11.19% 16.22% 23.79% 25.48%
% default in first 18 months 3.90% 3.74% 7.67% 6.86% 15.92% 23.35% 34.91% 33.87%
% default in first 21 months 5.11% 4.91% 10.51% 6.40% 23.35% 31.72% 43.75% 32.15%
HPI growth (12 months since 13.44% 9.10% 1.94% -4.19% 13.99% 9.70% 1.52% -3.94%
origination)
HPI growth (21 months since 20.98% 10.89% -1.39% 18.85% 11.06% -2.80%
origination)
Unemployment rate (12 5.15% 4.70% 4.45% 4.80% 5.28% 4.83% 4.55% 4.81%
months following orig.)
Median income in zip code $50,065 $49,486 $48,417 $48,221 $45,980 $44,965 $43,790 $43,817
(in $100,000)
Origination amount $173,702 $200,383 $211,052 $205,881 $167,742 $172,316 $179,003 $172,667
FICO 710 715 708 706 617 611 607 597
Loan to value ratio 75.92% 74.89% 75.99% 77.75% 79.63% 80.69% 80.40% 80.56%
Debt to income ratio (if 35.95% 37.87% 37.25% 38.74% 39.55% 38.35% 39.78% 40.72%
available)
Debt to income not available 52.8% 32.1% 27.6% 20.8% 41.0% 30.9% 27.2% 8.0%
(fraction of loans)
Interest rate at origination 5.6% 6.0% 6.7% 6.5% 7.1% 7.5% 8.5% 8.4%
Margin rate (rate increase at 2.3% 2.4% 2.9% 2.7% 5.2% 5.4% 5.5% 5.3%
reset for ARMs)
Fraction of loans that are:
Adjustable rate mortgages 26.45% 26.04% 23.16% 12.93% 73.31% 69.49% 61.78% 38.92%
(ARMs)
reset > 3 yrs 14.52% 13.32% 12.11% 10.38% 1.05% 0.96% 1.93% 6.63%
resetRESEARCH REVIEW
loans kept on portfolio drops to low different market segments and different and unemployment rates over the same
single digits for both prime and subprime years. To that end, we estimate the period. The same exercise can be
mortgages. Not surprisingly, nearly all following regressions: performed for loans originated in 2005
subprime mortgages are securitized by at the end of 2006, for loans originated
Prob (default within 12 months)i,j,k =
private investors, whereas GSEs in 2006 at the end of 2007, and so on.
Φ(β1Loani,j,k, β2Borroweri,j,k, β3Econj,k, β4Dk),
dominate the securitization of prime
mortgages. However, by the second half (1) Results
of 2007, the private securitization market where the dependent variable is an The results of the estimation are
had all but disappeared. The GSEs took indicator of whether a loan to borrower i, summarized in table 2. The first four
up much of the slack, accounting for originated in an MSA j in state k columns of data depict estimates for
about 40 percent of all subprime defaulted within the first 12 months. We prime loans originated in each of the four
securitizations in that year.10 model this probability as a function of sample years, and the next four columns
loan and borrower characteristics, MSA- contain the estimates for subprime loans.
Estimates of default level economic variables (unemployment, The juxtaposition of the data for the two
home price appreciation, and income), market segments allows us to easily
In this section, we estimate empirical and a set of state dummy variables (Dk) compare the importance of certain
models of the likelihood that a loan will that capture additional aspects of the factors. The table presents estimates of
default in its first 12 months. This allows economic and regulatory environment. the marginal effects of the explanatory
us to quantify which factors make default We estimate the model as a standard variables, rather than the coefficients
more or less likely and to examine how maximum likelihood probit with state themselves. The marginal effects tell us
the sensitivity to default varies over time fixed effects. how a one unit change in each
and across prime and subprime loans. explanatory variable changes the
To retain maximum flexibility in
Econometric model probability that a loan defaults in its first
evaluating the importance of covariates
12 months, holding fixed the impact of
Mortgages can have multiple sources for prime and subprime defaults, we carry
the other explanatory variables.
of risk—for example, low credit quality, out separate estimations of equation (1)
high loan-to-value ratios, and adjustable for prime and subprime loans. To achieve The defaults of both prime and
rates with short introductory periods and similar flexibility over time, we further subprime loans are strongly associated
high spreads to the reference rate. To subdivide each of the prime and with the FICO score, the LTV, and the
take into account these and other factors subprime samples by year of origination interest rate in every estimation year for
that might influence default rates, we (2004 through 2007). Finally, we attempt each loan type. For instance, an increase
estimate a number of multivariate to account for unobserved heterogeneity of 100 points in the FICO score of prime
regression models that allow us to at the state level by incorporating state borrowers in 2004 and 2005 is
examine the effect of varying one risk fixed effects in our econometric associated with about a 1.2 percentage
factor while holding others fixed. specification. point decrease in default likelihood (the
estimated marginal effect of –0.00012
The analysis sample includes loans The economic variables include both multiplied by 100). To gauge the strength
that do not default and are observed for the realized growth in the FHFA HPI and of this effect, note that in those years the
12 months after origination and loans the average realized unemployment rate. baseline rate of default was about 2.2
that default (become 60 or more days Both of these variables are measured at percent. The point estimates of marginal
past due) within 12 months of origination. the MSA level, and both are computed effects for 2006 and 2007 increase
We drop loans that get refinanced or over the first 12 months since loan about twofold for prime loans, but so
transferred within their first 12 months. origination. Consequently, they match the does the baseline sample default rate.
While this may bias our results, keeping period over which we are tracking loan For subprime loans, the estimated
early refinanced and transferred loans in performance. In contrast to all of the marginal effects are a full order of
the sample would understate the share other regressors, this information would magnitude higher, implying that an
of actual defaults, since by definition clearly not be available to the analyst at improvement in FICO scores generates a
these loans are current for the duration the time of loan origination. We can think greater decline in subprime defaults, at
of their (short) presence in the sample. of the model described in equation 1 as least in absolute terms.
the sort of analysis one would be able to
Our goal is to evaluate the relative do at the end of 2005, after all loans Similarly, higher LTV values have a
strength of associations between loan originated in 2004 had gone through strong positive association with defaults
default and observable borrower, loan, their first 12 months, and one is able to for both loan types originated in 2005,
and macroeconomic characteristics in observe what happened to home prices 2006, and 2007. For subprime loans, a
Profitwise News and Views September 2010 5RESEARCH REVIEW
Table 2: Probability of defaulting within 12 months of loan origination (probit regressions with state fixed effects)
Prime Loans Subprime Loans
2004 2005 2006 2007 2004 2005 2006 2007
(1) (2) (3) (4) (5) (6) (7) (8)
VARIABLES default_in12 default_in12 default_in12 default_in12 default_in12 default_in12 default_in12 default_in12
Estimation sample mean 0.0221 0.0217 0.0423 0.0483 0.1076 0.1572 0.2399 0.2539
-0.00166 -0.00494 -0.137*** -0.00356 -0.183* -0.168*** -0.447*** -0.0105
HPI growth
(0.0139) (0.0109) (0.0294) (0.0243) (0.0981) (0.0500) (0.0934) (0.121)
-0.0104 0.222*** -0.0370 0.131 0.218 0.743*** -0.749** -0.476
Unemployment rate
(0.0507) (0.0393) (0.115) (0.0968) (0.324) (0.239) (0.346) (0.503)
-0.00149 -0.00253 -0.00689 -0.0231*** -0.0398 -0.0572*** -0.0942*** -0.0672
Median income in zipcode
(0.00428) (0.00388) (0.00772) (0.00880) (0.0281) (0.0215) (0.0299) (0.0412)
-0.000122 0.000176 0.000830 0.00190** 0.0135*** 0.0160*** 0.0247*** 0.0331***
Origination amount
(0.000387) (0.000425) (0.000806) (0.000758) (0.00436) (0.00341) (0.00598) (0.00627)
-0.000120*** -0.000120*** -0.000262*** -0.000318*** -0.000733*** -0.00122*** -0.00131*** -0.00116***
FICO score
(1.64e-05) (1.16e-05) (1.97e-05) (2.26e-05) (9.12e-05) (6.84e-05) (9.10e-05) (0.000136)
0.00433 0.0144*** 0.0636*** 0.0814*** 0.0523 0.0820*** 0.193*** 0.193***
Loan to value ratio (LTV)
(0.00528) (0.00500) (0.0109) (0.0123) (0.0368) (0.0255) (0.0330) -0.0477
Debt to income ratio (0 if 0.000502 -0.00160 0.00841 0.0343*** 0.0876** 0.143*** 0.0900** 0.106**
missing) (0.00409) (0.00350) (0.00859) (0.00845) (0.0399) (0.0302) (0.0388) (0.0446)
0.00248 0.00148 0.00974** 0.0119** 0.0265 0.0593*** 0.0183 0.000879
Missing DTI dummy
(0.00218) (0.00187) (0.00491) (0.00586) (0.0191) (0.0148) (0.0180) (0.0258)
0.337*** 0.257** 1.351*** 1.653*** 2.487*** 3.092*** 4.692*** 5.384***
Interest rate at origination
(0.103) (0.107) (0.186) (0.218) (0.388) (0.282) (0.345) (0.476)
ARM w/ reset > 3 yrs 0.00151 -0.00366 -0.00112 0.0358** 0.0288 -0.0328 -0.0956*** 0.187
dummy (0.00685) (0.00228) (0.00485) (0.0183) (0.0609) (0.0313) (0.0294) (0.133)
0.00180 -0.00566*** -0.0140*** 0.0462 -0.0345 0.00391 -0.0197 0.203*
ARM w/reset < 3 yrs dummy
(0.00687) (0.00204) (0.00364) (0.0317) (0.0321) (0.0200) (0.0325) (0.121)
-0.192 0.150 0.322** -0.165 1.235** 0.776** 1.841*** -2.483
Margin rate (0 if FRM)
(0.245) (0.118) (0.147) (0.340) (0.521) (0.363) (0.568) (2.014)
0.00286 0.00374 0.00757 -0.00390 0.00369 0.0109 -0.0189* 0.00180
Prepayment penalty dummy
(0.00572) (0.00325) (0.00467) (0.00476) (0.00894) (0.00753) (0.0111) (0.0159)
0.00274 0.00445** 0.000601 -0.00157 -0.00985 -0.0186** -0.0284** -0.0117
Cash out refinancing dummy
(0.00229) (0.00222) (0.00318) (0.00340) (0.0110) (0.00915) (0.0122) (0.0164)
0.00290* 0.00222* 0.00552** 0.00604** 0.0243*** 0.0415*** 0.0856*** 0.0729***
Purchase loan dummy
(0.00151) (0.00131) (0.00244) (0.00295) (0.00863) (0.00602) (0.00815) (0.0132)
-0.000422 0.00468 0.000339 0.00159 -0.00612 -0.00102 -0.00164 0.0446
Investment property dummy
(0.00305) (0.00304) (0.00378) (0.00486) (0.0216) (0.0143) (0.0163) (0.0301)
-0.00290 -0.00667*** -4.45e-05 0.00166 0.0154 0.0226*** 0.0196* 0.0119
Conforming loan dummy
(0.00190) (0.00184) (0.00288) (0.00344) (0.0120) (0.00864) (0.0107) (0.0184)
-0.0113*** -0.00623*** -0.0190*** -0.00352 0.0255 -0.0312 -0.138*** -0.00455
GSE-securitized loan dummy
(0.00268) (0.00225) (0.00417) (0.00530) (0.0271) (0.0194) (0.0284) (0.0294)
-0.00578** -0.000475 -0.00930** -0.000801 0.00680 -0.0860*** -0.168*** -0.0619**
Private label securitized loan
(0.00269) (0.00207) (0.00424) (0.00635) (0.0183) (0.0148) (0.0228) (0.0241)
Observations 8,887 15,653 13,941 12,932 5,825 19,356 17,359 8,349
R-squared 0.2587 0.2364 0.1997 0.1962 0.1138 0.0934 0.0926 0.0745
SOURCE: FHFA for HPI growth, BLS for unemployment rate and median income, LPS for all other variables.
NOTES: ***, **, and * denote statistical significance at the 1, 5, and 10 percent levels, respectively.
6 Profitwise News and Views September 2010RESEARCH REVIEW
Table 3.A: Average marginal effects from changes in key explanatory variables
Prime
2004 2005 2006 2007
2004 - 2007 (1) (2) (3) (4)
VARIABLES Mean Std. deviation CHANGE (+) default_in12 default_in12 default_in12 default_in12
Baseline predicted
0.0220 0.0217 0.0422 0.0482
default rate
HPI growth 4.8 10.3 10 ppt -0.0003 -0.0010 -0.0178*** -0.0006
-1% -5% -42% -1%
FICO score 710 62 50 points -0.0116*** -0.01*** -0.0172*** -0.0204***
-53% -46% -41% -42%
Loan to value ratio
76.1 16.8 10 ppt 0.0009 0.0031*** 0.0113*** 0.0141***
(LTV)
4% 14% 27% 29%
Debt to income ratio
37.7 14.9 10 ppt -0.0001 -0.0002 0.0009 0.0049***
(if not missing)
0% -1% 2% 10%
Interest rate at
6.25% 0.81% 1% 0.0105*** 0.0062** 0.027*** 0.032***
origination
48% 29% 64% 66%
Margin rate (0 if
2.58% 1.14% 1% -0.0007 0.0008 0.0018** -0.0004
FRM)
-3% 4% 4% -1%
SOURCE: FHFA for HPI growth, BLS for unemployment rate and median income, LPS for all other variables.
NOTES: ***, **, and * denote statistical significance at the 1, 5, and 10 percent levels, respectively.
rise in LTV generates a stronger origination year. What stands out is the There are also a number of notable
absolute increase in loan defaults. It sheer magnitude of the estimated differences between the prime and
must be noted that the effect of the effects. However, one must be cautious subprime samples. Perhaps the most
leverage on the likelihood of default may in interpreting hypothetical marginal interesting finding is the different
be understated by the LTV measure that effects of the interest rate. While LTV sensitivity of defaults to changes in home
we have. A better measure of how and FICO score cover fairly wide ranges prices. For subprime loans, defaults are
leveraged a borrower is on a given for both prime and subprime loans, much lower when home price growth is
property would be the combined loan- interest rate values are relatively tightly higher for three out of the four sample
to-value ratio (CLTV) that also takes into distributed.11 This means that a years. This relationship is particularly
account second-lien loans on the difference of even 1 percent in loan striking for 2006 loan originations, many
property. This variable is not available in interest rate makes it look quite of which experienced home price
the LPS data, however. If the practice of different from loans with otherwise declines over their first 12 months. For
obtaining such “piggyback loans” is identical characteristics (e.g., FICO prime loans, 2006 is notable as the only
more prevalent in the subprime market, score, LTV, DTI). In such cases, a likely year of origination in which changes in
then the estimated coefficient for LTV explanation is that the lender has home prices are significantly correlated
for subprime loans may be lower than its additional information about the credit with loan defaults. These results suggest
true value. quality of the borrower and is charging a that, relative to subprime defaults, prime
higher interest rate to take into account defaults have a weaker relationship with
At first glance, the interest rate at
additional risk factors – hence, the home prices, once key borrower and loan
origination is similar to LTV and FICO
strong positive association with eventual characteristics (LTV, FICO score, and so
score in having a strong statistical and
default rates. on) are taken into account.
economic effect on both prime and
subprime loan defaults in each
Profitwise News and Views September 2010 7RESEARCH REVIEW
Table 3.B: Average marginal effects from changes in key explanatory variables
Subprime
2004 2005 2006 2007
2004 - 2007 (1) (2) (3) (4)
VARIABLES Mean Std. deviation CHANGE (+) default_in12 default_in12 default_in12 default_in12
baseline predicted
0.1075 0.1571 0.2396 0.2538
default rate
HPI growth 5.4 9.6 10 ppt -0.0189* -0.0163*** -0.0406*** -0.0010
-18% -10% -17% 0%
FICO score 608 55 50 points -0.0351*** -0.0532*** -0.0581*** -0.0519***
-33% -34% -24% -20%
Loan to value ratio
80.4 12.6 10 ppt 0.0059 0.0084*** 0.0189*** 0.0189***
(LTV)
5% 5% 8% 7%
Debt to income ratio
39.4 10.7 10 ppt 0.0069** 0.0108*** 0.0066** 0.0095**
(if not missing)
6% 7% 3% 4%
Interest rate at
7.93% 1.31% 1% 0.0302*** 0.0331*** 0.0471*** 0.0542***
origination
28% 21% 20% 21%
Margin rate (0 if
5.39% 0.72% 1% 0.0112** 0.0059** 0.0125*** -0.0101
FRM)
10% 4% 5% -4%
SOURCE: FHFA for HPI growth, BLS for unemployment rate and median income, LPS for all other variables.
NOTES: ***, **, and * denote statistical significance at the 1, 5, and 10 percent levels, respectively.
The contrast between prime and London interbank offered rate, or Libor). introductory periods of less than three
subprime loans is even sharper in the At reset, the ARM rate goes up to the years – the most common mortgage
estimated marginal effects on the debt- sum of Libor and the loan margin. The contract in the subprime market – are not
to-income ratio (DTI) and loan margin margin is set by the lender, and is often significantly different from zero. This
rate. The DTI is widely considered to be thought to capture additional aspects of means that they have the same
one of the main determinants of loan a borrower’s creditworthiness. This is correlation with subprime defaults as
affordability, since it relates household consistent with the fact that the margin fixed-rate mortgages. Put differently,
monthly income to debt service flows. rate is, on average, substantially higher once loan and borrower characteristics
The DTI for prime loans is not for subprime borrowers (see table 1). We are accounted for, the choice of a hybrid
significantly correlated with defaults, find that this variable has no association ARM is not associated with higher
except for loans originated in 2007, but it with defaults among prime loans, with the subprime defaults.
matters consistently for subprime loans. exception of loans originated in 2006. In
The absence of any measurable effects contrast, defaults on subprime loans
Comparisons across years and across loan types
of DTI even on defaults of prime loans originated in every year except 2007 are Since table 2 contains regression
originated in 2006 can be interpreted as significantly higher for loans with higher estimates from multiple non-overlapping
a sign of the resilience of prime margin rates, all else being equal. This samples, the comparison of the relative
borrowers who experienced severe suggests that, for the subprime borrower, importance of the explanatory variables
changes in the prices of their homes. the margin rate contains additional may be tricky. The distribution of loan
information on borrower quality not characteristics varies from year to year
The loan margin rate is one of the key
reflected in FICO scores and other loan and across prime and subprime loans. In
terms in an ARM contract. It defines the
characteristics. It is also interesting that addition, the baseline rates of actual
spread to a reference rate (usually the
the coefficients on ARMs with defaults are quite different across
8 Profitwise News and Views September 2010RESEARCH REVIEW
samples. Because of this, one cannot defaults would have been 1.78 percentage To do this, let’s conduct the following
simply compare two point estimates and points, or 42 percent, lower. The effect of thought experiment. Suppose that it is
conclude that a bigger one indicates a FICO score stands out. A 50-point uniform June 2006 and we are trying to forecast
stronger correspondence with defaults. increase in FICO scores (row 2 of each defaults on prime mortgages originated
panel) is associated with a 41 percent to earlier that year. The most up-to-date
To compare the economic and relative 53 percent decline in predicted default model of defaults available to us at this
importance of the explanatory variables rates relative to the baseline for prime point in time is that of defaults on 2004
across the subsamples, we conduct the loans, and a 20 percent to 34 percent originations. (Recall that to estimate this
following exercise. For each independent relative decline for subprime loans. The model, one needs to observe mortgages
variable, we change its value for each average marginal effects of the LTV are for 12 months since origination.) Further
observation by a specified increment. greater (in a relative sense) for prime loans suppose that as astute analysts, we get a
Then, we compute the predicted than for subprime loans.14 Finally, higher definite sense that house price growth is
subsample default rate using estimated interest rates appear to generate slowing down, even though available data
coefficients for each year of origination incredible increases in defaults for both are not picking this up strongly yet. And
and loan type. We compare the new market segments. For instance, a 1 so in a fit of pessimism, we conclude that
predicted default to the original one. The percentage point increase in interest rates prices may even decline a touch this year
difference between the original translates into a jump in defaults on 2007 after growing at 9 percent, on average, in
prediction and the new one tells us the prime loans of more than 3 percentage 2005. What would our models tell us
marginal contribution of that variable to points—a rise of 66 percent relative to the about the default outlook?
the overall default rate.12 We compare actual default rate. Increasing everyone’s
these figures across years and across The answer is “not much.” In 2004
interest rates by 1 percentage point is
prime loans (table 3, panel A) and (and 2005), the models of prime
equivalent to a substantial deterioration in
subprime loans (table 3, panel B). For mortgage performance detected almost
the quality of the borrower pool, and thus
example, for 2004 prime loans we no relationship between house price
translates into much higher predicted
increase all FICO scores by 50 points, growth and defaults. The coefficients on
defaults. As mentioned earlier, DTI and the
predict a new default rate, and compare it HPI growth were effectively zero, and so
margin rate do not have strong
to the old default rate. The difference is no forecast of HPI, however dire, would
associations with prime mortgage defaults.
–0.0116 percentage points, or a 53 have rung the alarm bells regarding
In contrast, higher values of these
percent decrease in the likelihood of prime mortgage defaults.
variables consistently indicate higher
default for loans originated in 2004
default rates for subprime mortgages. What an analyst would have had to
(column 4, row 2 of table 3, panel A). For
However, the economic magnitude of realize was that in 2006 prime borrowers
brevity, we look at just six key explanatory
marginal effects of DTI and the margin will start reacting to HPI in the same way
variables: HPI growth, FICO score, LTV,
rate on defaults (rows 4 and 6 of each as subprime ones. What was needed
DTI, interest rate, and loan margin rate.13
panel) is somewhat muted. then was not a better forecast of housing
The table also reports the means of the
prices, but an understanding that the
relevant variable, its standard deviation, What if market observers foresaw the decline in
statistical relationships from the boom
and the absolute change that we impose. home prices?
years no longer applied. Detecting the
We tried to keep the magnitude of the
We turn our attention now to the role of turning points is never easy, and in this
absolute changes reasonably close to
home prices. We know that home prices instance most observers failed abjectly.
the standard deviations.
were increasing very rapidly in 2004 and
A 10 percentage point increase in 2005 and began to fall quite dramatically
Conclusion
home price appreciation (HPI growth) beginning in 2006. But would it have been
substantially lowers default probabilities possible to quantify the effect of this We have analyzed the default
(first row of each panel in table 3). This reversal on defaults of both prime and experience of prime and subprime loans
effect is more consistent for subprime subprime mortgages in real time? It is also originated over the period 2004–07. Like
loans originated in various years, where it important to be clear about what other studies, we document some decline
translates to decreases of between 10 information would have been available to in underwriting standards during this
percent and 18 percent relative to the analysts at different points in time. This will period for both prime and subprime
baseline default rate in 2004, 2005, and allow us to get a rough sense of the extent loans. We also find that characteristics
2006. For prime loans, the 10 to which market participants were such as the LTV, FICO score, and interest
percentage point increase in the HPI has “surprised” by the performance of prime rate are important predictors of defaults
a big effect only for loans originated in and subprime loans originated in 2006 for both prime and subprime loans.
2006, where the estimates imply that and 2007. However, changes in loan and borrower
Profitwise News and Views September 2010 9RESEARCH REVIEW
characteristics are not enough to have Notes 7 As part of the Housing and Economic
predicted the incredible increase we have Recovery Act of 2008 (HERA), the
seen in prime and subprime mortgage 1 These numbers are based on authors’ Federal Housing Finance Regulatory
defaults. While changes in borrower and calculations using data from LPS Applied Reform Act of 2008 established a single
Analytics, described later in text. regulator, the FHFA, for GSEs involved in
loan characteristics can get us closer to
the home mortgage market, namely,
observed default rates for subprime loans 2 By including prime loans in the analysis,
Fannie Mae, Freddie Mac, and the 12
than they can for prime loans, for both our intention is to complement the very
Federal Home Loan Banks (see www.fhfa.
market segments there were other informative and extensive literature on
gov for additional details).
factors at work. subprime loans that includes, among
others, Bajari, Chu, and Park (2008); 8 Note that we are looking at a relatively
Home prices play a very important role Demyanyk and Van Hemert (2009); short period, and other authors document
in determining mortgage outcomes; this Gerardi, Shapiro, and Willen (2008); and changes in underwriting criteria that
became particularly evident for subprime Mian and Sufi (2009). occurred prior to 2004 (see, for example,
loans by the end of 2005. For prime Gerardi et al., 2008).
3 The servicers included in the data set are
loans, it is only when we analyze data those that participate in the HOPE NOW 9 Such mortgages were known as “hybrid
through the end of 2007 (that is, evaluate Alliance (www.hopenow.com/members. ARMs.” They were also commonly
the performance of loans originated in html#mortgage). This includes some of identified as “2/28” and “3/27” loans,
2006) that we are able to document this the country’s largest home lenders: Bank referring to 30-year ARMs that reset after
of America, Citibank, JPMorgan Chase, two and three years, respectively.
sensitivity. Even very pessimistic
and Wells Fargo. 10 In September of 2007 when the private
assumptions about the future path of
home prices would not have been 4 Alt-A loans are a middle category of securitization market had all but shut
enough to substantially improve loans – more risky than prime and less down, the GSEs were encouraged by
contemporaneous forecasts of prime risky than subprime. They are generally members of Congress to expand their
made to borrowers with good credit portfolios to support the market (see this
mortgage defaults for loans made in
ratings, but the loans have characteristics correspondence between Senator Charles
2006 and 2007. In hindsight, of course, it
that make them ineligible to be sold to the E. Schumer (D-NY) and Dennis Lockhart,
appears self-evident that the GSEs – for example, limited the director of OFHEO, at www.ofheo.gov/
relationships between HPI growth and documentation of the income or assets of newsroom.aspx?ID=383&q1=0&q2=9.
defaults on prime loans might be the borrower or higher loan-to-value ratios 11 Keep in mind that for simplicity, the
different in periods with declining home than those specified by GSE limits. analysis uses the actual interest rate at
prices. Coming up with such revised
5 As Bajari, Chu, and Park (2008) loan origination and not the difference
estimates in real time would not have emphasize, an important feature of the between this rate and some reference risk
been possible using the available data FICO score is that it measures a free rate.
from the recent past. It could, perhaps, borrower’s creditworthiness prior to taking 12 Note that this exercise amounts to
have been done by analyzing data that out the mortgage. FICO scores range computing the average of marginal effects
included earlier episodes of substantial between 300 and 850. Typically, a FICO for individual loans, instead of the marginal
regional price declines. score above 800 is considered very good, effect at the mean, which is obtained by
while a score below 620 is considered multiplying a hypothetical change in an
poor. As reported on the Fair Isaac explanatory variable by its regression
Corporation website (www.myfico.com),
Biographies borrowers with FICO scores above 760
coefficient.
are able to take out 30-year fixed rate 13 In this exercise, the loan margin is
Gene Amromin is a senior financial increased only for ARMs, since fixed rate
mortgages at interest rates that are 160
economist in the financial markets loans by definition have a zero margin
basis points lower, on average, than those
group at the Federal Reserve Bank of under all circumstances. Similarly, we
available for borrowers with scores in the
Chicago. 620–639 range. incremented DTI only for those loans that
had non-missing DTI values.
Anna Paulson is a vice president of 6 If we repeat the analysis using alternative
the financial economics team in the outcome variables and different time 14 As discussed earlier, this may be due to
periods (in default after 18 months, in our inability to accurately account for
economic research department of the
foreclosure, 30 days or more past, and so piggyback loans.
Federal Reserve Bank of Chicago.
on), the results are very similar.
10 Profitwise News and Views September 2010CONSUMER ISSUES
Alternative small dollar loans:
Creating sound financial products
through innovation and regulation
by Chris Giangreco, Andrea Kovach, and Matt Unrath
Introduction Review of existing literature Association of America, defend their
product and service to low- and
Low- to moderate-income borrowers There has been a great deal of moderate-income borrowers. Their
need alternatives to payday loans to research on the payday loan industry. argument, published in reports and
meet their short-term credit needs. This Various regional banks of the Federal testimonials, is that payday loan terms
article provides an overview of Reserve System have published papers compare favorably with alternative
consumer demand for smaller loans, and examining the nature of payday loan options, like overdraft or late fees. The
discusses how and why mainstream borrowers and products, and weighed the Center for Responsible Lending has
financial institutions should offer less effectiveness of industry regulation (see encouraged the OCC and other bank
costly alternatives to traditional payday the Federal Reserve Bank of regulators to prohibit banks from
loans. A two-year FDIC pilot, a small- Philadelphia’s “Restricting Consumer offering products similar to payday
dollar loan pool in Baltimore, and Credit Access” and New York’s “Defining loans. Some larger banks have recently
individual case studies suggest that and Detecting Predatory Lending”). The introduced products such as paycheck
such lending can be viable and New America Foundation’s Asset advances, but the charges can amount
profitable. The article concludes with Building program has written extensively to a 120 percent APR or higher. The
recommendations for how financial on the issue, including an often cited National Consumer Law Center
institutions and regulators should paper on how behavioral sciences can authored a recent report warning that
support this effort inform financial services regulation. one should not assume that
Consumer advocacy groups, like the “alternatives” to payday loans are
Many consumers struggling to make
Center for Responsible Lending and inherently less costly. The authors of
ends meet need small, short-term loans.
National Consumer Law Center, have this paper support responsible
Illinois residents received nearly 1.2
authored reports on payday lenders (see alternatives to payday lending with
million payday loans1 from 2006 to 2008;
“Quantifying the Economic Cost of interest rates capped at 36 percent
however, debate about the utility of this
Predatory Payday Lending”). The APR and amortized loan payments
loan product continues. Payday lenders
Woodstock Institute has exposed based on an individual’s ability to repay.
claim that they provide a needed service
regulatory loopholes and published
to those underserved by mainstream
various reports that highlight the need for Current landscape of payday
financial institutions. Consumer
stricter regulation (see “Illinois Payday lending
advocates cite the predatory pricing of
Loan Loophole”). Much of this research
payday loans, and call for stringent The majority of small-dollar loans
has focused on questions about payday
regulation, especially with respect to currently available to consumers across
lending regulations; officials, researchers,
pricing. In an effort to move beyond this the country are payday loans. Payday
and advocates have only recently begun
debate and meet consumer demand (with loans are essentially quick cash
advocating for alternative products, and
a less costly product), advocates, advances, usually of $500 or less,
there is likewise little lending data from
regulators, and financial institutions targeted to low- and moderate-income
which to draw evidence about
should explore the viability of alternative individuals with limited credit history or
effectiveness and profitability.
small-dollar loan products from both a low credit ratings. Payday loan
consumer and business perspective. Payday lenders, represented by the businesses operate outside of the
Consumer Financial Services
Profitwise News and Views September 2010 11CONSUMER ISSUES
mainstream financial sector, often 1. Small-dollar loan pool pilot 600, he or she is required to attend a
relying on a network of retail storefronts, free financial education workshop on
where customers can walk in, provide One model for extending small-dollar understanding credit and meet with a
minimal personal information, and leave loans involves financial institutions financial counselor to prepare a personal
with enough cash to meet their providing capital to a community-based budget. In the past seven years, North
immediate financial needs. The organization, which uses existing Side CFCU has made over 5,000 PALs,
“underwriting process” comprises relationships within a targeted market to disbursing over $2.5 million in PAL loans
documenting information from a pay offer loans. and saving community residents over $3
stub, and the “collateral” is a postdated million in fees and interest from
With support from the FDIC and grants
check or automatic debit authorization, traditional payday loans.
from six financial institutions and one
which covers the principal borrowed and
credit union, Neighborhood Housing In July 2008, North Side CFCU
interest; the loans often require
Services of Baltimore (NHS–Baltimore) launched its newest alternative loan
borrowers to have a checking account.
created a small-dollar loan pool. NHS– product, the “Step-Up” loan, a payday
Because payday lenders operate with
Baltimore made 80 loans between August alternative loan of $1,000 available to
very little underwriting, they rationalize
2009 and February 2010, totaling members that have paid off at least five
high interest rates as necessary to
$60,400. The majority of these 12-month PALs. No credit check is required and
ensure profit.
term loans are $1,000, though some as borrowers can pay back the loan in six
The payday loan industry has seen small as $500. All of the loans have a months or one year. Since the launch of
enormous growth. In the three years competitive APR of 7.99 percent. A typical Step-up, North Side has made 527
between 2000 and 2003, national sales consumer in this pilot is an African- loans for a total of $527,000. Ed Jacob,
volumes quadrupled from $10 billion in American woman between 40 and 50 North Side CFCU’s manager (at the
2000 to $40 billion in 2003; researchers years old, earning about $30,000 time) stated, “Our goal isn’t to be just a
put the total costs to consumers for using a annually, who has filed for bankruptcy in cheaper payday lender. We want to give
payday loan at $4.2 billion annually (King, the past but currently holds a bank people a path that will help them reach
Parrish and Tanik 2006). According to the account, and plans to use the loan to pay their financial goals. We want them to
Illinois Department of Financial and bills. Joan Lok, Community Affairs think longer term, and go beyond
Professional Regulation (IDFPR), there are Specialist with the FDIC in Baltimore, needing a $500 loan.”
403 licensed payday lenders operating stated that while financial institutions
under the Payday Loan Reform Act in contributing to the pool were originally 3. FDIC’s small-dollar loan pilot
Illinois. IDFPR found that during the three- skeptical of the program, many have come The Federal Deposit Insurance
year period between February 2006 and to embrace and support it. One goal Corporation (FDIC) began a two-year
December 2008, 1,194,582 payday loans associated with the pilot is the program to pilot program for alternative small-dollar
were taken out by 204,205 consumers in earn enough to be self-sustaining. The loans in February 2008. At the
Illinois—an average of 5.9 loans per FDIC’s Alliance for Economic Inclusion, in program’s conclusion in December
consumer at an average annual percentage conjunction with local financial institutions, 2010, 28 banks with assets from $28
rate (APR) of 341 percent (IDFPR 2009). is developing similar loan pools in Kansas million to $10 billion and offices in 27
City and Seattle. states participated. The program aimed
Alternative small-dollar loan 2. Innovative financial institutions to assess the business practices of the
efforts banks in developing and offering
Banks and credit unions have also profitable small-dollar loan programs
Consumer advocates, financial
begun developing their own alternative alongside other mainstream services.
institutions, and regulators have begun
small-dollar loan programs. In 2002,
working together to promote and The FDIC developed the following
North Side Community Federal Credit
develop responsible, alternative small- guidelines for financial institutions
Union (North Side CFCU), recognizing
dollar loan products that meet participating in the pilot:
the preponderance of predatory payday
consumers’ needs and protect them
from usurious lending practices. The
lenders in its community and the impact • Loan amounts of up to $2,500;
high interest debt had on its members,
following examples highlight some of
decided to develop its Payday Alternative • Amortization loan periods of at least
the innovative current strategies in this
Loan (PAL) program. Loans in the PAL 90 days with minimum payments that
market segment.
program are $500, repaid during a six- reduce the loan principal;
month term, and have an annual
percentage rate of 16.5 percent. If a first-
• Annual percentage rates (APR)
below 36 percent;
time borrower has a credit score below
12 Profitwise News and Views September 2010CONSUMER ISSUES
• No prepayment penalties; damage done to the banking industry borrowers who belonged to a financial
brand” (Benjamin 2009). institution for more than one year
• Origination and/or maintenance fees reported lower delinquency rates
limited to the amount necessary to New customer base – Because of (Williams 2007).
cover actual costs; and high demand, financial institutions can
attract new customers by offering an CRA credit – The Federal Financial
• An automatic savings component. alternative loan product at competitive Institutions Examination Council recognizes
The FDIC released final results of the prices. Through such loans, banks and that small-dollar loans meet an important
program in late June 2010 on the credit unions can build the financial credit need of underserved communities
impact and effectiveness on banks’ skills and knowledge of their customers, and low- and moderate-income borrowers.
profitability 2 and long-term customer graduating them to more sophisticated By offering these types of loans or
relations. In the first year of the program, financial products. More than half of the supporting the development of a small-
banks made over 16,000 loans, for an banks in the FDIC’s small-dollar loan dollar loan pool, banks can earn
aggregate principal balance of $18.5 pilot reported that customers moved to Community Reinvestment Act (CRA) credit.
million. The total amount of loans other bank services after using a small- Such loan products allow borrowers to
charged off in the first year was dollar loan. Most pilot banks opened avoid high cost credit, and ostensibly serve
$187,378, or 3.4 percent of loans of all deposit accounts for customers who the purpose and mission of the CRA.
loans originated (Burhouse, Miller, and successfully used a small-dollar loan
Sampson 2009). Banks reported that product, and some banks transitioned Recommendations
job losses and other economic problems customers into more sophisticated
products. One participating bank found The case for offering alternative
in their market areas led to increased
that auto loans were a “next step” in small-dollar loans through mainstream
delinquencies across loan categories
building the lending relations with small- financial institutions does not negate
and to a reduction in the pool of
dollar loan customers who successfully the need for regulation of the industry.
acceptable borrowers. Common factors
paid off their loan (Burhouse, Miller, and Borrowers and lenders do not enter into
cited for operating successful loan
Sampson 2008). lending contracts on equal footing, in
programs included strong senior
either financial understanding or
management and board support; an Leverage advantage in the market – bargaining power (Saunders and Cohen
engaged and empowered “champion” in Banks and credit unions have two 2004). Regulation must aim to narrow
charge of the program; proximity to inherent advantages over the payday this divide and protect consumers from
large consumer populations with loan industry in successfully offering predatory practices and their own
demand for small-dollar loans; and, in small-dollar loan programs – behavior tendencies (Barr, Mullainathan,
rural markets, limited competition. infrastructure and relationships. While and Sharif 2008).
payday loan stores must spend capital
Benefits of small-dollar loans for on space, staff, advertising, and more, Regulators at the state and federal
financial institutions banks and credit unions already have level play an important role in
qualified staff, a large network of developing alternative small-dollar loans
Mainstream financial institutions can
physical facilities, and functioning and assuring proper consumer
benefit from alternative small-dollar
collection processes. Their ability to protections. The following are
lending by serving as the pacesetter of
advertise through bank statements and recommendations that state and federal
sound and competitive financial practices
existing marketing materials helps bring government offices should follow to
for low- and moderate-income clients.
attention to the product and quickly support this effort:
Image improvement – Participation in
initiatives to develop alternative small-
draw a market. Banks and credit unions • Support efforts to develop small-
can build on their relationship with dollar lending pool pilots and study
dollar loans can help re-position banks clients to help determine the type of their effectiveness;
and other mainstream financial loan best suited for a borrower, as well
institutions amid the current economic as streamline the underwriting • Support efforts to develop informed
situation. Luis Ubiñas, president of the process—a necessary step if banks and policy on unregulated payday and
Ford Foundation, said at a meeting in credit unions wish to compete with the consumer installment loans and
Brooklyn, New York, in June 2009, “The present payday loan industry (Burhouse, provide guidance on features for
economic downturn has tarnished bank Miller, and Sampson 2008). This small-dollar loan products;
brands; offering innovations and underwriting process will help mitigate
providing new opportunities to non- delinquency risks. Research from the
• Encourage responsible alternative
traditional customers can help repair the small-dollar loan products;
Woodstock Institute found that
Profitwise News and Views September 2010 13You can also read
