Subprime Lending: Neighborhood Patterns Over Time
Jonathan Hershaff, Board of Governors of the Federal Reserve System
Susan Wachter, University of Pennsylvania
Karl Russo, University of Pennsylvania
For presentation to the conference of:
Promises & Pitfalls
Federal Reserve System’s Fourth Community Affairs Research Conference
on
April 7-9, 2005
Corresponding Author
Susan M. Wachter
Richard B. Worley Professor of Financial Management;
Professor of Real Estate and Finance
Wharton School
University of Pennsylvania
3733 Spruce Street
430 Vance Hall
Philadelphia, Pa 19104
215-898-6355
wachter@wharton.upenn.edu
For comment only: do not cite without permission
1
Abstract
Previous research has shown that subprime lending occurs disproportionately in markets
with higher risk as well as larger shares of minority households. This paper extends the literature
by identifying how subprime lending has changed over time, following subprime lending trends
in seven cities across the US and their neighborhoods, defined by zip codes, and comparing
outcomes in 1997 and 2002.
We find major shifts in subprime lending patterns over this time period. In this period of
overall growth, in many areas, and in two of the cities, subprime lending decreases. We identify
factors associated with subprime lending trends and shifts in lending patterns over time. Growth
in subprime lending is strongly associated with growth in Hispanic households and is also more
likely to occur in neighborhoods with households who have low levels of educational
achievement. Lower median income is less associated with subprime lending in 2002 than in
1997; risk measures and percentage share of African-American households, holding risk constant,
are more strongly associated with subprime market share over time.
2
Subprime Lending: Neighborhood Patterns Over Time
By
Jonathan Hershaff
Susan Wachter
Karl Russo
I. Introduction
In this paper, we track changing patterns of subprime lending across neighborhoods in
seven major cities in the US over a five-year period. Subprime lending increases substantially on
average from 1997 to 2002, however growth is not uniform across neighborhoods. In specific
areas, and in two of the cities, the number of subprime loans decreases over time. We identify
factors associated with shifting neighborhood patterns of subprime lending, by estimating timeseries regressions of subprime growth. We also implement cross-section estimations of the
3
factors associated with subprime lending for 1997 and 2002 to determine whether these
correlates change from 1997 to 2002.
The substantial growth in subprime lending nationwide is shown in Table 1. Mortgage
lending increases steadily from 1996 to 1999, decreases in 2000 due to the downturn in the
economy, and then increases dramatically in the aftermath of the recession and the ensuing flurry
of refinancing activity. Table 2 shows the monetary volume of this trend. After the decline in
2000, the growth trend resumes doubling of the number of subprime loans with an even greater
increase in the dollar volume of lending through 2003.
Despite the robust growth of subprime lending nationwide, there are persistent declines
in subprime lending across and within some cities. Figures 1 and 2 illustrate the changing
patterns of lending for Philadelphia, which along with Baltimore, experiences a decline in the
number of subprime loans from 1997 to 2002. The areas within Philadelphia with the largest
declines appear to be low-income and high-risk neighborhoods, as defined by zip codes in the
bottom quintile of income and in the top quintile of risk for each city, with risk measured by zipcode-level averages of credit scores and default rates.
Tables 3 and 4 present findings on subprime growth by neighborhoods, segmented by
risk and income measures. Absolute and/or relative declines in subprime lending occur in areas
of low income, high credit risk and high default risk.
Much of the literature on the spatial patterns of subprime lending finds a
disproportionately large market share for subprime lending in low-income and heavily
proportioned minority neighborhoods. Subprime, by definition, serves markets with greater risk,
so that an explanation of these observed patterns may be found in the risk characteristics of
neighborhoods. Nonetheless, these will be the markets with greater losses ex post; thus, with
4
“learning” over time, subprime lenders may adjust pricing or may lend less to neighborhoods
with higher ex post measured risk. Similarly, it appears that subprime lenders are expanding
more in relatively higher income areas, perhaps motivated by selecting borrowers with the ability
to pay higher subprime lending rates.
In this paper, we examine whether and how patterns of subprime lending are changing
over this five-year period. Have subprime lenders withdrawn from or expanded less in markets
with lower income and/or greater risk? We analyze the factors associated with differing growth
rates in subprime lending across markets. We also separately estimate cross-section city-level
regressions for 1997 and 2002 to test for whether there are changes in the factors associated with
subprime market share, particularly focusing on whether there is a change in the role of risk and
income measures over time.
The degree to which subprime lending correlates with neighborhood economic and
demographic characteristics is of interest because high subprime default rates are more likely to
have adverse consequences for communities to the extent that subprime loans are concentrated in
neighborhoods that are fundamentally more vulnerable to economic decline. Correlations of
subprime lending with neighborhood demographic characteristics are also of interest because
they may reflect “targeting” through more intensive marketing of subprime products.
In Section II which follows, we review the literature on the spatial patterns of subprime
lending. In Sections III and IV respectively, we review data sources and methodology.
Empirical results are presented and discussed for the subprime change and city-level crosssection estimations in Section V. Section VI briefly concludes.
II. Literature Review
5
The literature on neighborhood patterns of subprime lending examines the frequency of
subprime borrowing relative to prime borrowing in residential mortgage markets in relation to
both borrower and neighborhood characteristics. All such studies rely on data on the individual
characteristics of mortgage loans and borrowers that are collected by the Federal Financial
Institutions Examination Council (FFIEC), in accordance with the Home Mortgage Disclosure
Act (HMDA). HMDA data do not separately identify subprime loans, so these studies rely on
the Department of Housing and Urban Development’s (HUD) list of lenders that specialize in the
subprime market and use loans originated by these institutions as a proxy for subprime loans,
while all other loans are treated as prime.
Previous studies generally find significant concentrations of subprime lending among
minority borrowers or within neighborhoods where minority households predominate. Bunce
(2000) present evidence that on average nationwide, subprime loans are three times more
frequent in low-income neighborhoods than in upper-income neighborhoods and five times more
frequent in predominantly black neighborhoods than in predominantly white neighborhoods.
Canner et al. (1999) find that subprime lending increases the number of loans to low- or
moderate-income and minority households and to low- or moderate-income and predominantly
minority neighborhoods.
Moreover, Canner et al. show that increases in subprime lending go disproportionately to
minority tracts and are responsible for more than one-third of the growth in overall lending to
predominantly minority tracts between 1993 and 1998. Immergluck and Wiles (1999) also show
subprime lending as a share of overall lending has been increasing in neighborhoods with high
concentration of minorities. However, neither study controls for neighborhood characteristics
such as risk.
6
Sheessele (2002) identifies the type of neighborhoods in the nation as a whole where
borrowers are likely to rely on subprime loans for refinancing. This study finds that even after
controlling for several neighborhood characteristics, the percentage of African-Americans is
positively related to the share of subprime refinance. Pennington-Cross et al. (2000) provides
evidence that the subprime market does not primarily originate mortgages to lower income
borrowers; rather such lending primarily serves higher risk borrowers. In addition, this study
finds that black and Asian borrowers have a higher probability of using the subprime market.
Pennington-Cross (2002) indicates that subprime lending is most prevalent in locations with
declining house prices. Calem, Gillen, Wachter (2004) include, together with other neighborhood
variables, neighborhood credit risk measures. They find a correlation associating subprime
lending with minority areas in multivariable regression analyses using 1997 HMDA data; when
including risk variables, coefficients on percent minority drop in general by approximately onehalf. Calem, Hershaff, Wachter (2004) replicate this study for 2002, including additional cities,
and find similar results.
The National Community Reinvestment Coalition (NCRC 2003) partially replicates the
CGW study analyzing subprime lending in ten large metropolitan areas. As in CGW, measures
of the credit-quality composition of neighborhood residents are included along with a number of
neighborhood economic and demographic variables. The NCRC study finds that, in nine of the
ten cities, the proportion of subprime refinance lending increases as the proportion of minorities
in a neighborhood increased, all else equal. Apgar et al. (2004) extend this analysis to more
MSAs, replicating the logistic analysis of CGW, although they do not include measures of the
credit-quality composition of neighborhood residents. They find that the proportion of
7
minorities in a neighborhood is significant and negatively related to the market share of prime
lenders.
This paper extends the literature by focusing on shifting subprime lending patterns over
time and explaining these patterns by including new risk information. Using data on
neighborhood default rates we construct predicted default rates by zip code for the seven cities in
the study to measure and control for property risk. As Calem and Wachter (1999) and Calem,
Gillen and Wachter (2002) show, property and borrower risk are both important as contributors
to the overall risk of mortgage lending. Borrower risk is important in delinquency and property
risk, in default.
III. Data
We use five main sources of data for the analysis. For all sources, data are organized by
zip code, and we construct risk variables by zip code as well. First, for individual characteristics
of mortgage loans and borrowers in each city, we use HMDA data for the years 1997 and 2002.
From these data, we derive several zip-code-level variables. Second, we use HUD’s list of
lenders that specialize in the subprime market to identify each loan as being subprime or not.
Throughout the empirical analyses, we use data for these lenders that are identified as subprime
for refinance mortgage loans only. Third, we use 2000 Census data to construct tract
demographic variables and neighborhood risk measures. Fourth, we use information on the
distribution of credit ratings within tracts available from CRA Wiz®, a product of PCI Services in
Boston that provides comprehensive, geography-based information. Finally, we obtain data on
default activity by zip code from Anonymous Credit Information Company. Table 5 defines
variables that we derive from these data and use in the empirical analysis.
8
From the 2000 Census, we obtain a number of tract-level economic and demographic
variables for use in the analysis. These include the log of tract median family income (LN MED
INCOME), the percent of individuals 25 years of age or older with a bachelor’s degree (PCT
COLLEGE), and the proportion of occupied housing units that are renter-occupied (PCT RENT).
Since economic conditions tend to be better in neighborhoods where residents have higher
incomes or educational attainment, and since borrower financial sophistication tends to be
directly related to educational attainment, we expect subprime borrowing to be inversely related
to these variables. Since home ownership tends to be associated with less risk and higher levels
of household wealth, we expect subprime borrowing to be directly related to percentage share of
renter-occupied housing.
Three measures of percent minority population: the percent of households headed by a
person classified as African American, the percent headed by a person classified as Asian, and
the percent headed by a person classified as Hispanic, (respectively PCT BLACK, PCT ASIAN,
PCT HISPANIC) also are used.
A proxy for the price of risk in real estate investment, the tract’s capitalization rate
(CAP RATE), defined as a ratio of the tract’s annualized median rent divided by the median
house value, also is constructed using 2000 Census data. A larger value for this measure is
consistent with lower expected price appreciation or more uncertain future house prices and,
hence, indicates increased risk. Hence, we expect this variable to be positively associated
with the relative likelihood of a loan being subprime.
Using CRA Wiz®, we calculate two measures of the credit-quality composition of
neighborhood residents by census tract. These are the percent of adult individuals in a tract that
have been classified as very high credit risk (PCT VHIGH RISK), based on their credit score,
9
and the percent with no credit rating (PCT NOINFO RISK). Both are expected to be positively
associated with the relative likelihood of a loan being subprime.
The tract variables are then aggregated to a Census-developed proxy for zip codes: zip
code tabulation areas (ZCTAs). "ZCTAs are generalized area representations of U.S. Postal
Service (USPS) ZIP Code service areas. Simply put, each one is built by aggregating the Census
2000 blocks, whose addresses use a given ZIP Code, into a ZCTA which gets that ZIP Code
assigned as its ZCTA code. They represent the majority USPS five-digit ZIP Code found in a
given area. For those areas where it is difficult to determine the prevailing five-digit ZIP Code,
the higher-level three-digit ZIP Code is used for the ZCTA code
(http://www.census.gov/geo/ZCTA/zcta.html).” Median family income, rent, and house value
were weighted by the number of occupied housing units in the tract. Percent with at least a
bachelor’s degree was weighted by the tract population over 25 years of age.
A number of borrower characteristics from HMDA data are used as independent
variables in the borrower-level logistic regressions. Specifically, we employ dummy variables on
the borrower’s racial and gender characteristics (BLACK, HISPANIC, ASIAN, FEMALE) and
also include the log of borrower income (LN_INCOME).
The analysis below also includes the actual or a predicted probability that a loan is more
than 90 days delinquent, hereafter referred to somewhat loosely as the (predicted) default rate
(DEFAULT and PRED DEF, respectively) as another measure of risk. Because there is some
concern over endogeneity, the predicted variable is constructed using a two-stage estimation
procedure.1 In the first stage, we use the Anonymous Credit Information Company data to
1
Initially, a simple prediction of foreclosure rate was employed which included all of the loan-level and zip-level
variables and additional interactions for collateral class and city dummies. However, we prefer the two-stage
estimator. While there is little or no change in the performance of the regressions on the loan-level variables, we see
some improvement in signs and/or significance of the other risk variables, particularly credit risk..
10
estimate the foreclosure rate as a function of the loan type and age category as well as loan
collateral class. We regress the residual from this first stage on loan-to-value categories, the
demographic and credit-risk variables described above, and city dummy variables. The predicted
default rates from this equation enter the city-level logistic regressions for subprime lending.
These regressions are run without the default rate and with the actual default rate for a check on
robustness.
Table 6 presents sample mean values. The table shows that cap rates are similar across
cities over time except that Philadelphia and Baltimore have high cap rates, consistent with the
fact that they are losing population. Also, the median family income values point to Philadelphia
and Baltimore as relatively low income and to Dallas and Atlanta as high income cities. Minority
percentages vary as expected. The percentage of college graduates ranges from a low in
Philadelphia of 18% (22% in 2002) to a high in Atlanta of 40% (44% in 2002), with the average
of the sample cities at 27% (30% in 2002). Home-ownership rates vary across the cities from a
low of 42% (45% in 2002) in New York to a high of 63% (62% in 2002) in Philadelphia. The
average home-ownership rate for the cities considered is 53% (same in 2002). The increasing
trend in these variables does not reflect changes in the underlying demography of neighborhoods,
since these data all derive from the 2000 census; rather they suggest that the neighborhood
composition of subprime lending is shifting over time towards neighborhoods with higher levels
of social and economic capital.
The borrower-income-level variable shows that the lowest income level is $43,000
($66,000 in 2002) in Philadelphia and that Dallas has the highest income level of $94,000
($98,000 in 2002). The other borrower characteristics vary across cities in expected ways.
11
As can be seen, the percentage of subprime loans varies a great deal across cities. For all
cities, subprime lending as a percent of total lending declines over time from 33% to 14%. This
is in part because of the surge of refinance loans in 2002 during a period of very low interest
rates which spurred refinancing activity. This is evidenced by the more than three-fold aggregate
increase in the number of subprime loans and the nearly four-fold increase in the number of
prime loans. The pattern of the increase in subprime lending varies across cities, with some
areas and cities showing declines. In the following section, we turn to explanations of these
shifts.
IV. Methodology
We begin by looking at simple cross tabulations of subprime lending activity over time,
focusing on the core central cities within each MSA. This represents the unconditional pattern of
subprime lending activity over the period. The next step involves several regressions of the rate
of change in the number of subprime loans from 1997 to 2002 at the zip code level. We first
examine simple pooled regressions of rate of change of subprime lending against city dummies
and dummy variables for whether the zip code is in the highest credit or foreclosure risk or
lowest income quintile for each. This is undertaken to confirm the relationships evident in the
cross tabulations. Other zip-level variables are then added to the regressions to see whether these
relationships are robust to inclusion of other covariates. Finally, the analysis returns to loan level
regressions by city of whether a loan is subprime on all of the zip-code-level and borrower-level
variables listed in Table 5. We include as controls individual characteristics of borrowers, but
because we lack individual credit-score data, we focus on the results for associations of zip-codelevel neighborhood factors with subprime market share.
12
V. Empirical Results
Results shown in Table 3 and 4, calculated from the simple cross tabulations, indicate
that subprime lending expanded less or contracted more than lending overall between 1997 and
2002. Both of these effects are robust across all cities and all risk areas with the exception of
overall lending and overall subprime lending in Dallas, where relatively small base numbers in
1997 somewhat distort the percentage changes. The mean rate of growth at the zip-code level is
193% for all cities, consistent with the observed expansion of subprime lending.
Results for Time Series Estimation
Univariate regression results, with subprime lending as the dependent variable, shown in
Table 7, are consistent with the cross tabulations and confirm the pattern of slower growth of
subprime lending in areas of high credit risk (column 1), high default risk (column 3), or lower
area income level (column 5) as defined by zip codes in the top quintile of respective risk or
lowest median family income quintile for each city.
The expanded models: 2, 4, and 6, shown in Table 7, include all other variables, with
high credit risk, high default risk, and low income quintiles entering respectively. Model 7
includes all variables, while model 8 excludes the default measures to check for robustness. In
these expanded models, high-risk areas are no longer significant as detractors to
growth2. Consistently significant variables include percent Hispanic, percent with bachelor
degree, and median family income. Subprime loan growth, is higher in areas with more Hispanic
households, lower educational outcomes, and higher median income.
2
The observed effect of neighborhoods being in the lowest quintile of median family income also dissipates but the
continuous income variable is significant.
13
The observed effect of being high risk, in the cross tabulations and in the single-variate
estimation, is insignificant when controlling for other zip-code level characteristics. Percent
Hispanic, percent college educated, and log of median family income have strong effects on
subprime growth rates. An increase in the Hispanic population from 10% to 20% is associated
with an approximately 70% higher subprime growth rate in a zip code. Subprime lending growth
is also lower in highly educated areas. A one-standard-deviation increase (20-percentage points)
in the percentage of individuals with a bachelor’s degree is associated with a nearly 200percentage-point reduction in the growth rate of subprime lending While seemingly expanding
less in risky markets, it appears that it is low income rather than risk per se that is discouraging
growth. The coefficient on LN MED INCOME implies that a a zip code with a 10% higher
median family income experiences a 50% higher growth rate in subprime lending, all else equal.3
Cross-Section Estimation Results
Tables 8 and 9 show results from loan-level regressions for whether a loan is likely to be
subprime or not for each city for 1997 and 2002, respectively. Using the results shown in Tables
8 and 9, we compare the coefficients for the 1997 and the 2002 regressions to determine for each
variable how many cities have coefficients that are significant and of the correct sign. In 1997,
focusing on neighborhood variables, we find that, all else equal, areas with lower family income
have a higher percentage of subprime loans. This is the only consistently significant
neighborhood variable in 1997.
The results from 2002 show some interesting comparisons. In that year, CAP RATE is
much more consistently positive and significant as is PCT NOINFO RISK. This implies a
3
Median family income for the sample ranges from $11,250 to $171,750 with a mean of $44,800.
14
stronger association of risk variables with subprime lending over time. PCT BLACK also more
consistently has positive and significant coefficients and PCT COLLEGE more consistently has
negative and significant coefficients. In general for 2002, risk variables have the expected
positive signs, indicating riskier areas attract more subprime lending. In addition, percent
college is now inversely related to share of subprime lending. It is noteworthy, however, that
median income is no longer consistently associated negatively with subprime lending as it was in
1997. In 2002, coefficients on median income are negative and significant in three rather than
four markets with size of coefficients in these markets declining by more than one-half. One
market, Dallas, shows a positive and significant coefficient on median income in 2002.
Tables 10 and 11 leave out the predicted default rate while Tables 12 and 13 use the
actual default rate instead. These alternative specifications reveal a broadly consistent story.
The risk variables, however measured, become more strongly associated with subprime lending
over time. This is evidenced by more positive and significant coefficients in 2002 and/or fewer
negative and significant coefficients relative to 1997. PCT COLLEGE also becomes more
strongly inversely associated with subprime lending over time. It has a negative sign in all but
one city in Tables 9, 11, and 13 compared to mixed results in 1997 Tables 8, 10, and 12. Median
income shows the reverse movement of education. It is negative and significant almost
universally in 1997 and has mixed results or insignificance in 2002.
Discussion of Results
When determinants of the rate of growth of subprime lending are analyzed in a singlevariate regression, both default risk and credit risk, defined as zip codes in the top risk quintile,
are detractors to growth. However, in a multi-variant regression with other variables included,
both risk measures become insignificant. Low neighborhood income is a detractor to growth in a
15
single-variant regression and remains a significant detractor (measured as a continuous variable)
in the multi-variate regression.
In the city-level regressions, the coefficients of low area income shift from uniformly
positive and significant in the 1997 regressions to low levels of significance in 2002. This
suggests that low-income areas, all else equal, are less attractive to subprime lending. On the
other hand, neighborhood risk measures become more significant in the 2002 regressions, with
the expected positive coefficients. This result contradicts the apparent negative effect of high
risk on subprime share in the cross tabulations which do not control for any covariates, but is not
inconsistent with the insignificant coefficient of these variables in the multi-variate change
regressions.
In the city-level regressions, coefficients on racial composition of areas increase in their
significance over time. The educational variable also becomes more consistently significant in
2002, with the expected negative coefficients, which again conforms to the apparent role of this
variable in explaining growth in subprime lending over time, with areas with low levels of
educational attainment, all else equal, even more attractive to subprime lending in 2002 than
earlier.
VI. Conclusion
Subprime lending has grown significantly since 1997, particularly in areas of high
Hispanic concentration and in areas with lower levels of household educational achievement.
Subprime lending has grown less in low-income areas than elsewhere. In city-level regressions
for 1997 and 2002, controlling for demographic variables, subprime lenders are more likely to be
active in riskier areas (as defined by credit risk) in 2002 than in 1997. We also find that
minority status -- in particular, percentage of African-American households -- continues to be
16
strongly associated with subprime lending in 2002, holding other variables constant. Consistent
with rate-of-change results, low income is less related to subprime dominance in 2002 than in
1997. Finally, we find that lack of education, holding other variables constant, is a consistently
significant factor in explaining the market share of subprime lending in 2002, as it is in
explaining the growth of subprime lending over time. These results are preliminary; additional
research for an expanded number of cities, incorporating expanded measures of risk is in process.
17
Bibliography
Apgar , William Allegra Calder, and Gary Fauth “Industry Structure Perpetuates Dual Market”
presented at the Federal Reserve System’s Community Affairs Research Conference, April 7,
Washington D.C.
Barakova, Irina, Raphael W. Bostic, Paul S. Calem, Susan M. Wachter. “Does Credit Quality
Matter for Homeownership?” Journal of Housing Economics, 2004.
Bunce. (2000). Curbing Predatory Home Mortgage Lending. Department of Housing and Urban
Development, Working Paper Series.
Calem, Paul S. (1996). “Mortgage Credit Availability in Low- and Moderate-Income Minority
Neighborhoods: Are Information Externalities Critical?” The Journal of Real Estate Finance
and Economics 13(1): 71-89.
Calem, Paul S. (1996). “Patterns of Residential Mortgage Activity in Philadelphia’s Low- And
Moderate-Income Neighborhoods.” Mortgage Lending, Racial Discrimination, and Federal
Policy, editors John Goering and Ron Wienk. Washington D.C., The Urban Institute Press, 671676.
Calem, Paul S., and Susan Wachter. (1999). “Community Reinvestment and Credit Risk:
Evidence from an Affordable Home Loan Program.” Real Estate Economics 27, pp.105-134.
Calem, Paul S., Jonathan E. Hershaff, Susan M. Wachter. “Neighborhood Patterns of Subprime
Lending: Evidence from Disparate Cities.” Housing Policy Debate. Vol. 15, No. 3. 2004.
Calem, Paul S., Kevin Gillen, Susan M. Wachter. “The Neighborhood Distribution of Subprime
Mortgage Lending.” Journal of Real Estate Finance and Economics. Vol. 29, No. 4. 2004.
Canner, Glenn B. and Elizabeth Laderman (1999). “The Role of Specialized Lenders in
Extending Mortgages to Lower-Income and Minority Homebuyers.” Federal Reserve Bulletin,
November.
Engel, Kathleen C., and Patricia A. McCoy. “Predatory Lending: What Does Wall Street Have
to Do with It?” Housing Policy Debate. Vol. 15, No. 3. 2004.
Engel, Kathleen C., and Patricia A. McCoy. “Predatory Lending and Community Development
at Loggerheads” Housing Policy Debate Federal Reserve Bank Conference, December 6, 2004
Immergluck, Daniel and Marti Wiles. Two Steps Back: The Dual Mortgage Market, Predatory
Lending, and the Undoing of Community Development. Chicago, IL. November 1999.
Joint Center for Housing Studies. (2004). Credit, Capital and Communities: The Implications of
the Changing Mortgage Banking Industry for Community Based Organizations.” Cambridge,
MA: Harvard University.
18
Ling, David C. and Susan M. Wachter. “Information Externalities and Home Mortgage
Underwriting,” Journal of Urban Economics 44(3): 317-332.
National Community Reinvestment Coalition. (2003), “The Broken Credit System:
Discrimination and Unequal Access to Affordable Loans by Race and Age: Subprime Lending in
Ten Large Metropolitan Areas,” Washington, D.C.: National Community Reinvestment
Coalition.
Pennington-Cross, Anthony. (2002). “Subprime Lending in the Primary and Secondary
Markets.” Journal of Housing Research 13(1): 31-50.
Pennington-Cross, Anthony and Joseph Nichols. (2000). “Credit History and the FHA
Conventional Choice.” Real Estate Economics 28(2): 307-336.
Pennington-Cross, Anthony, Anthony Yezer and Joseph Nichols. (2000). “Credit Risk and
Mortgage Lending: Who Uses Subprime and Why?” Research Institute for Housing
America, Working Paper No. 00-03.
Scheessele, Randall M. (1998). “1998 HMDA Highlights.” Housing Finance Working Paper
Series, HF-009. U.S. Department of Housing and Urban Development. July.
19
Table 3 - Percent Rate of Change in Subprime Lending Activity
Atlanta
Baltimore
Chicago
Dallas
Los Angeles
New York
Philadelphia
188.7
52.0
172.9
860.6
262.7
296.0
78.2
10.0
-53.8
8.3
900.4
64.9
100.2
-27.0
Number of Loans in High Credit
Risk Areas
78.4
-30.0
16.5
663.8
154.9
158.4
-9.7
Number of Subprime Loans in
High Credit Risk Areas
3.8
-71.5
-21.7
608.8
46.1
52.0
-51.3
126.0
5.3
19.5
684.2
143.1
177.3
-0.3
2.9
-68.6
-25.5
678.3
29.6
58.4
-52.5
145.0
-21.4
215.9
886.2
166.3
129.4
-23.2
2.7
-74.1
14.9
735.4
25.6
26.7
-60.9
Total Number of Loans
Number of Subprime Loans
Number of Loans in High
Default Areas
Number of Subprime Loans in
High Default Areas
Number of Loans in Low Income
Areas
Number of Subprime Loans in
Low Income Areas
Table 4 - Subprime Lending Activity
Total
Number of
Loans
Number of
Subprime
Loans
Number of
Loans in
High Risk
Areas
Number of
Subprime
Loans in
High Risk
Areas
Number of
Loans in
High Default
Areas
Number of
Subprime
Loans in
High Default
Areas
Atlanta
1997
2002
Baltimore
1997
2002
Chicago
1997
2002
Dallas
1997
2002
Los Angeles
1997
2002
3,357
9,693
4,219
6,413
23,289
1,312
1,443
2,374
1,097
551
983
891
373
387
912
551
New York
1997
2002
Philadelphia
1997
2002
63,548
1,634
15,696
7,835
28,418
13,848
54,841
9,377
16,711
8,314
9,006
257
2,571
2,771
4,568
5,016
10,040
4,231
3,089
624
6,528
7,605
160
1,222
2,572
6,557
3,494
9,028
1,972
1,780
659
188
3,595
2,815
91
645
1,459
2,132
2,277
3,462
1,333
649
2,061
852
897
6,311
7,540
203
1,592
2,085
5,069
4,238
11,752
1,962
1,957
567
564
177
3,508
2,614
106
825
1,190
1,542
2,628
4,163
1,228
583
Table 5 – Variable Definitions
Zipcode-level Variable Definitions
Variable
Definition
PCT SUB
Subprime as a Pct. of Loans Originated in the Zipcode
PCT VHIGH RISK
Pct. of Zipcode Population Very High Risk
PCT NOINFO RISK
Pct. of Zipcode Population With No Credit History
PCT BLACK
PCT HISP
PCT ASIAN
PCT COLLEGE
Pct. of Zipcode Homeowners Black
Pct. of Zipcode Homeowners Hispanic
Pct. of Zipcode Homeowners Asian
Pct. of Zipcode Pop. 25+ Years of Age with a Bachelor’s Degree
PRED DEF
Predicted Default Rate
CAP RATE
Median Rent / Median House Value
LN MED INCOME
DEF
Log of Zipcode Median Income
Default Rate
Loan-level Variable Definitions
Variable
Definition
SUB PRIME
Dummy if Loan is Subprime
BLACK
Dummy if Borrower is Black
HISPANIC
ASIAN
LN INCOME
MISSING
Dummy if Borrower is Hispanic
Dummy if Borrower is Asian
Log of Borrower Income
Missing Information
Table 6: Summary Statistics: Sample Mean Values
Atlanta
Variable
1997
Baltimore
2002
1997
2002
Chicago
1997
Dallas
2002
1997
Los Angeles
2002
1997
2002
New York
1997
2002
Philadelphia
Whole Sample
1997
1997
2002
2002
ZIP-code-level variables
CAP RATE
Median Family Income
0.05
0.05
0.08
0.08
0.06
0.05
0.07
0.07
0.04
65,551 69,293 45,359 52,654 54,274 61,792 68,149 66,212 59,002
0.04
0.04
0.04
0.11
0.09
0.06
0.05
57,323 57,166 60,844 41,987 47,416 55,018
59,831
PCT ASIAN
0.02
0.02
0.02
0.02
0.03
0.05
0.04
0.04
0.09
0.10
0.08
0.09
0.04
0.04
0.05
0.07
PCT BLACK
0.52
0.46
0.55
0.40
0.33
0.19
0.13
0.12
0.12
0.10
0.29
0.24
0.45
0.30
0.30
0.18
PCT HISPANIC
0.05
0.05
0.02
0.03
0.21
0.21
0.19
0.21
0.33
0.35
0.18
0.18
0.06
0.05
0.20
0.23
PCT VHIGH RISK
0.18
0.17
0.18
0.18
0.17
0.17
0.18
0.18
0.18
0.18
0.18
0.17
0.18
0.17
0.18
0.17
PCT NOINFO RISK
0.18
0.17
0.18
0.17
0.18
0.17
0.17
0.17
0.17
0.17
0.16
0.16
0.17
0.17
0.17
0.17
PCT COLLEGE
0.40
0.44
0.22
0.29
0.25
0.32
0.37
0.36
0.30
0.29
0.27
0.29
0.18
0.22
0.27
0.30
PCT RENTAL
0.51
0.52
0.43
0.42
0.45
0.43
0.41
0.41
0.50
0.48
0.58
0.55
0.37
0.38
0.47
0.47
87,539 76,789 90,520 42,671 66,052 65,389
85,770
Borrower-level variables
INCOME
68,287 92,481 48,423 77,435 59,902 80,402 93,825 97,944 75,302
ASIAN
0.01
0.01
0.01
0.01
0.03
0.05
0.02
0.04
0.06
0.08
0.05
0.06
0.01
0.02
0.03
0.05
BLACK
0.34
0.21
0.40
0.15
0.24
0.11
0.06
0.05
0.11
0.05
0.15
0.14
0.27
0.11
0.21
0.09
HISPANIC
0.01
0.01
0.01
0.01
0.12
0.15
0.05
0.10
0.13
0.18
0.05
0.07
0.02
0.02
0.09
0.13
MISSING
0.17
0.22
0.22
0.28
0.14
0.13
0.10
0.18
0.16
0.24
0.29
0.31
0.29
0.35
0.19
0.22
SINGLE
0.67
0.73
0.63
0.66
0.49
0.52
0.43
0.51
0.49
0.56
0.56
0.63
0.59
0.64
0.52
0.56
SINGLE FEMALE
0.28
0.28
0.28
0.25
0.21
0.20
0.15
0.16
0.22
0.20
0.18
0.21
0.27
0.22
0.22
0.20
SINGLE MALE
0.39
0.45
0.35
0.41
0.28
0.32
0.28
0.35
0.26
0.35
0.38
0.41
0.32
0.42
0.30
0.36
SUB PRIME
0.39
0.15
0.52
0.15
0.32
0.12
0.13
0.14
0.27
0.13
0.36
0.18
0.45
0.19
0.33
0.14
3,403
9,920
4,816
Number of Loans
8,474 31,536 95,613
3,293 32,419 20,925 110,158 14,076 55,871
9,418 16,732 87,467 329,187
Table 7 Regression Results
Percent Change in Number of Subprime Loans
Central City Zip Codes
1997-2002
Intercept
1
11.45
2
-48.01
3
11.31
4
-47.13
5
11.39
6
-44.23
P-value
0.00
0.05
0.00
0.06
0.00
0.08
HIRISK
-1.42
P-value
0.02
0.08
0.09
-0.53
-0.44
-0.51
0.56
0.65
0.58
HIDefault
P-value
-0.86
-0.16
-0.08
0.15
0.83
0.92
LOWINC
P-value
DEFAULT
7
8
-45.45 -43.07
-1.46
-0.46
-0.36
-0.31
0.01
0.62
0.71
0.74
-2.81
-2.49
-2.90
-2.53
0.50
0.61
0.49
0.60
14.77
8.27
8.34
13.66
12.36
0.66
0.79
0.79
0.69
0.71
-24.54
-27.32
-21.75
P-value
0.41
0.36
0.49
0.53
0.59
PCT BLACK
0.82
0.83
0.66
0.74
0.47
P-value
0.67
0.67
0.74
0.71
0.80
PCT ASIAN
3.58
3.68
3.54
3.44
4.04
P-value
0.30
0.29
0.31
0.32
0.24
PCT HISPANIC
7.18
7.31
7.24
7.15
6.99
P-value
0.00
0.00
0.00
0.00
0.00
-9.37
-9.57
-9.02
-8.91
-8.25
P-value
0.03
0.03
0.04
0.05
0.06
PCT RENTALS
2.30
2.35
2.28
2.23
1.66
P-value
0.35
0.34
0.36
0.37
0.50
-0.16
-0.22
-0.09
-0.03
0.08
0.97
0.96
0.98
1.00
0.99
5.62
0.01
Yes
5.69
0.01
Yes
5.33
0.02
Yes
5.33
0.02
Yes
5.06
0.02
Yes
0.39
387
0.39
387
0.38
392
P-value
PCT VHIGH RISK
P-value
PCT NOINFO RISK
PCT COLLEGE
CAP RATE
P-value
LN MED INCOME
P-value
City Dummies?
R-squared
N
Yes
0.32
401
0.39
387
Yes
0.32
401
0.39
387
Yes
0.32
401
-20.34 -16.76
Table 8 - Loan-Level Regression Results for Refinance Loans 1997
Variable
INTERCEPT
INTERCEPT
ASIAN
ASIAN
BLACK
BLACK
HISPANIC
HISPANIC
LN INCOME
LN INCOME
MISSING
MISSING
CAP RATE
CAP RATE
LN MED INCOME
LN MED INCOME
PCT RENT
PCT RENT
PCT COLLEGE
PCT COLLEGE
PCT ASIAN
PCT ASIAN
PCT BLACK
PCT BLACK
PCT HISPANIC
PCT HISPANIC
Estimate
ProbChiSq
Estimate
ProbChiSq
Estimate
ProbChiSq
Estimate
ProbChiSq
Estimate
ProbChiSq
Estimate
ProbChiSq
Estimate
ProbChiSq
Estimate
ProbChiSq
Estimate
ProbChiSq
Estimate
ProbChiSq
Estimate
ProbChiSq
Estimate
ProbChiSq
Estimate
ProbChiSq
Atlanta
1997
Baltimore
1997
Chicago
1997
Dallas
1997
Los Angeles
1997
New York City
1997
Philadelphia
1997
Whole Sample
1997
12.58
0.13
-1.74
0.10
0.45
0.00
-1.98
0.06
-0.84
0.00
1.45
0.00
0.66
0.93
-0.26
0.57
-0.90
0.46
-1.07
0.67
13.15
0.00
0.38
0.88
2.36
0.42
31.45
0.00
-0.24
0.61
1.00
0.00
1.07
0.01
-0.73
0.00
1.57
0.00
-4.27
0.31
-2.87
0.00
-0.73
0.41
1.52
0.18
5.63
0.14
0.49
0.66
-8.15
0.01
7.77
0.00
-0.13
0.21
1.02
0.00
-0.11
0.04
-0.52
0.00
1.22
0.00
0.15
0.83
-0.74
0.00
-0.30
0.35
0.15
0.75
0.28
0.55
1.92
0.00
0.36
0.46
-5.55
0.58
-0.57
0.29
0.46
0.04
-0.28
0.30
-0.65
0.00
1.04
0.00
3.12
0.41
0.58
0.38
0.38
0.72
-2.10
0.15
-4.78
0.06
1.16
0.57
0.36
0.82
7.55
0.01
-0.39
0.00
0.74
0.00
0.06
0.29
-0.26
0.00
1.48
0.00
-4.28
0.12
-1.17
0.00
-0.70
0.04
1.34
0.03
-0.54
0.11
1.52
0.01
-0.31
0.54
10.79
0.00
-1.12
0.00
0.48
0.00
0.00
0.98
-0.97
0.00
1.65
0.00
-4.89
0.02
-0.72
0.01
-1.24
0.00
-1.48
0.03
0.74
0.04
0.14
0.80
1.02
0.09
8.24
0.09
-0.91
0.00
0.87
0.00
0.03
0.85
-0.82
0.00
1.46
0.00
0.76
0.69
-0.87
0.08
-0.58
0.27
2.29
0.04
1.74
0.05
1.12
0.15
-0.02
0.99
10.17
0.00
-0.47
0.00
0.83
0.00
-0.05
0.16
-0.60
0.00
1.44
0.00
-1.36
0.01
-0.75
0.00
-0.92
0.00
-0.07
0.75
0.16
0.37
0.68
0.01
1.05
0.00
Variable
PRED PCT DEFAULT
PRED PCT DEFAULT
PCT VHIGH RISK
PCT VHIGH RISK
PCT NOINFO RISK
PCT NOINFO RISK
SINGLE FEMALE
SINGLE FEMALE
SINGLE MALE
SINGLE MALE
Estimate
ProbChiSq
Estimate
ProbChiSq
Estimate
ProbChiSq
Estimate
ProbChiSq
Estimate
ProbChiSq
Atlanta
1997
Baltimore
1997
Chicago
1997
Dallas
1997
Los Angeles
1997
New York City
1997
Philadelphia
1997
Whole Sample
1997
0.35
0.50
-36.72
0.05
-7.06
0.81
-0.12
0.28
-0.03
0.81
-0.17
0.40
8.48
0.51
8.27
0.56
-0.30
0.00
-0.32
0.00
-0.14
0.22
-2.29
0.54
6.38
0.29
0.11
0.00
0.16
0.00
0.13
0.70
4.06
0.80
-5.12
0.78
0.34
0.05
0.61
0.00
-0.16
0.13
7.13
0.22
19.91
0.00
0.11
0.01
0.03
0.43
0.11
0.31
5.05
0.39
-0.59
0.92
-0.21
0.00
0.02
0.72
0.00
1.00
4.54
0.55
7.54
0.43
-0.18
0.01
0.29
0.00
0.07
0.12
-3.28
0.11
-2.37
0.36
0.00
0.86
0.10
0.00
Table 9 - Loan-Level Regression Results for Refinance Loans 2002
Variable
INTERCEPT
INTERCEPT
ASIAN
ASIAN
BLACK
BLACK
HISPANIC
HISPANIC
LN INCOME
LN INCOME
MISSING
MISSING
CAP RATE
CAP RATE
LN MED INCOME
LN MED INCOME
PCT RENT
PCT RENT
PCT COLLEGE
PCT COLLEGE
PCT ASIAN
PCT ASIAN
PCT BLACK
PCT BLACK
PCT HISPANIC
PCT HISPANIC
Estimate
ProbChiSq
Estimate
ProbChiSq
Estimate
ProbChiSq
Estimate
ProbChiSq
Estimate
ProbChiSq
Estimate
ProbChiSq
Estimate
ProbChiSq
Estimate
ProbChiSq
Estimate
ProbChiSq
Estimate
ProbChiSq
Estimate
ProbChiSq
Estimate
ProbChiSq
Estimate
ProbChiSq
Atlanta
2002
Baltimore
2002
Chicago
2002
Dallas
2002
Los Angeles
2002
New York City
2002
Philadelphia
2002
Whole Sample
2002
-14.08
0.03
-0.39
0.41
1.07
0.00
0.46
0.16
-0.65
0.00
1.23
0.00
10.04
0.05
0.18
0.58
0.53
0.55
1.03
0.57
6.00
0.05
3.94
0.04
-1.23
0.56
-0.40
0.95
0.21
0.64
0.66
0.00
0.80
0.02
-0.78
0.00
1.29
0.00
5.15
0.26
-0.12
0.85
-0.88
0.32
-0.77
0.48
4.32
0.27
1.02
0.33
-2.09
0.43
-0.73
0.68
0.03
0.72
1.35
0.00
0.74
0.00
-0.44
0.00
1.12
0.00
1.12
0.02
-0.29
0.04
-0.04
0.87
-0.93
0.01
-0.42
0.23
2.28
0.00
0.05
0.89
-1.89
0.53
-0.60
0.00
1.09
0.00
-0.29
0.00
-0.65
0.00
0.98
0.00
4.50
0.00
0.78
0.00
-0.71
0.03
-4.21
0.00
1.45
0.07
-0.19
0.77
2.09
0.00
-1.27
0.43
0.12
0.00
0.88
0.00
0.45
0.00
-0.05
0.01
0.19
0.00
8.39
0.00
-0.45
0.00
-0.06
0.75
-0.75
0.03
0.04
0.83
1.28
0.00
-0.35
0.20
1.67
0.36
-0.48
0.00
0.60
0.00
0.53
0.00
-0.30
0.00
0.88
0.00
-4.39
0.00
-0.35
0.02
-0.54
0.01
-3.47
0.00
-0.43
0.03
0.95
0.00
-0.23
0.52
-2.01
0.63
-0.33
0.07
0.38
0.00
-0.23
0.15
-0.57
0.00
0.47
0.00
3.41
0.06
0.19
0.64
-0.54
0.24
-0.45
0.60
0.14
0.87
0.69
0.26
0.24
0.76
-1.79
0.01
0.02
0.58
0.88
0.00
0.48
0.00
-0.35
0.00
0.70
0.00
0.32
0.25
-0.03
0.57
-0.22
0.02
-2.24
0.00
-0.08
0.45
0.95
0.00
0.23
0.10
Variable
PRED PCT DEFAULT
PRED PCT DEFAULT
PCT VHIGH RISK
PCT VHIGH RISK
PCT NOINFO RISK
PCT NOINFO RISK
SINGLE FEMALE
SINGLE FEMALE
SINGLE MALE
SINGLE MALE
Estimate
ProbChiSq
Estimate
ProbChiSq
Estimate
ProbChiSq
Estimate
ProbChiSq
Estimate
ProbChiSq
Atlanta
2002
Baltimore
2002
Chicago
2002
Dallas
2002
Los Angeles
2002
New York City
2002
Philadelphia
2002
Whole Sample
2002
-0.52
0.16
13.57
0.34
41.83
0.05
-0.06
0.55
0.12
0.17
-0.11
0.55
22.94
0.09
-6.72
0.63
0.00
0.97
0.20
0.03
-0.20
0.02
3.96
0.17
13.99
0.00
0.36
0.00
0.40
0.00
0.32
0.00
-17.92
0.00
-9.78
0.09
0.20
0.00
0.14
0.00
-0.12
0.05
12.15
0.00
9.31
0.00
0.37
0.00
0.18
0.00
-0.08
0.21
10.94
0.00
4.58
0.19
0.19
0.00
-0.32
0.00
0.03
0.78
8.98
0.16
-10.26
0.17
0.25
0.00
0.47
0.00
-0.03
0.38
5.93
0.00
3.78
0.02
0.29
0.00
0.17
0.00
Table 10 - Loan-Level Regression Results for Refinance Loans 1997
Variable
INTERCEPT
INTERCEPT
ASIAN
ASIAN
BLACK
BLACK
HISPANIC
HISPANIC
LN INCOME
LN INCOME
MISSING
MISSING
CAP RATE
CAP RATE
LN MED INCOME
LN MED INCOME
PCT RENT
PCT RENT
PCT COLLEGE
PCT COLLEGE
PCT ASIAN
PCT ASIAN
PCT BLACK
PCT BLACK
PCT HISPANIC
PCT HISPANIC
Estimate
ProbChiSq
Estimate
ProbChiSq
Estimate
ProbChiSq
Estimate
ProbChiSq
Estimate
ProbChiSq
Estimate
ProbChiSq
Estimate
ProbChiSq
Estimate
ProbChiSq
Estimate
ProbChiSq
Estimate
ProbChiSq
Estimate
ProbChiSq
Estimate
ProbChiSq
Estimate
ProbChiSq
Atlanta
1997
Baltimore
1997
Chicago
1997
Dallas
1997
Los Angeles
1997
New York City
1997
Philadelphia
1997
Whole Sample
1997
8.05
0.11
-1.74
0.10
0.45
0.00
-1.99
0.06
-0.84
0.00
1.45
0.00
4.31
0.38
-0.46
0.20
-0.23
0.75
0.39
0.77
12.93
0.00
2.07
0.00
0.62
0.66
32.21
0.00
-0.24
0.60
1.00
0.00
1.07
0.01
-0.73
0.00
1.57
0.00
-5.25
0.19
-2.75
0.00
-1.21
0.08
0.96
0.30
5.83
0.12
-0.38
0.34
-7.88
0.01
8.73
0.00
-0.13
0.21
1.02
0.00
-0.11
0.04
-0.52
0.00
1.22
0.00
-0.40
0.46
-0.67
0.00
-0.61
0.00
-0.30
0.31
0.44
0.33
1.16
0.00
0.93
0.00
-8.81
0.27
-0.58
0.28
0.47
0.04
-0.27
0.31
-0.63
0.00
1.04
0.00
3.42
0.36
0.64
0.32
0.85
0.18
-1.82
0.09
-5.11
0.04
1.90
0.00
-0.24
0.65
8.92
0.00
-0.40
0.00
0.74
0.00
0.05
0.34
-0.26
0.00
1.47
0.00
-4.68
0.09
-1.05
0.00
-1.05
0.00
0.68
0.08
-0.27
0.35
0.66
0.00
0.41
0.03
9.81
0.00
-1.12
0.00
0.48
0.00
0.00
0.99
-0.97
0.00
1.65
0.00
-4.38
0.03
-0.82
0.00
-0.98
0.00
-1.03
0.04
0.62
0.07
0.65
0.00
0.45
0.07
8.24
0.09
-0.91
0.00
0.87
0.00
0.03
0.85
-0.82
0.00
1.46
0.00
0.75
0.67
-0.87
0.04
-0.58
0.25
2.28
0.00
1.74
0.04
1.11
0.00
-0.01
0.97
9.43
0.00
-0.47
0.00
0.84
0.00
-0.05
0.16
-0.60
0.00
1.44
0.00
-1.01
0.03
-0.79
0.00
-0.75
0.00
0.18
0.27
0.07
0.68
1.06
0.00
0.72
0.00
Variable
PCT VHIGH RISK
PCT VHIGH RISK
PCT NOINFO RISK
PCT NOINFO RISK
SINGLE FEMALE
SINGLE FEMALE
SINGLE MALE
SINGLE MALE
Estimate
ProbChiSq
Estimate
ProbChiSq
Estimate
ProbChiSq
Estimate
ProbChiSq
Atlanta
1997
Baltimore
1997
Chicago
1997
Dallas
1997
Los Angeles
1997
New York City
1997
Philadelphia
1997
Whole Sample
1997
-25.00
0.00
11.90
0.15
-0.13
0.28
-0.03
0.81
4.18
0.72
-2.28
0.73
-0.30
0.00
-0.32
0.00
-5.64
0.02
-0.81
0.65
0.11
0.00
0.16
0.00
9.20
0.30
1.38
0.85
0.36
0.04
0.62
0.00
0.63
0.88
14.02
0.00
0.11
0.02
0.03
0.47
9.92
0.01
4.29
0.21
-0.21
0.00
0.02
0.71
4.53
0.40
7.52
0.14
-0.18
0.01
0.29
0.00
-0.96
0.47
1.34
0.25
0.00
0.84
0.10
0.00
Table 11 - Loan-Level Regression Results for Refinance Loans 2002
Variable
INTERCEPT
INTERCEPT
ASIAN
ASIAN
BLACK
BLACK
HISPANIC
HISPANIC
LN INCOME
LN INCOME
MISSING
MISSING
CAP RATE
CAP RATE
LN MED INCOME
LN MED INCOME
PCT RENT
PCT RENT
PCT COLLEGE
PCT COLLEGE
PCT ASIAN
PCT ASIAN
PCT BLACK
PCT BLACK
PCT HISPANIC
PCT HISPANIC
Estimate
ProbChiSq
Estimate
ProbChiSq
Estimate
ProbChiSq
Estimate
ProbChiSq
Estimate
ProbChiSq
Estimate
ProbChiSq
Estimate
ProbChiSq
Estimate
ProbChiSq
Estimate
ProbChiSq
Estimate
ProbChiSq
Estimate
ProbChiSq
Estimate
ProbChiSq
Estimate
ProbChiSq
Atlanta
2002
Baltimore
2002
Chicago
2002
Dallas
2002
Los Angeles
2002
New York City
2002
Philadelphia
2002
Whole Sample
2002
-6.90
0.07
-0.39
0.41
1.07
0.00
0.47
0.15
-0.65
0.00
1.23
0.00
5.06
0.16
0.45
0.10
-0.49
0.33
-1.14
0.24
6.49
0.03
1.42
0.00
1.23
0.30
0.51
0.94
0.21
0.64
0.66
0.00
0.80
0.02
-0.78
0.00
1.29
0.00
4.33
0.32
-0.07
0.90
-1.21
0.08
-1.11
0.24
4.37
0.27
0.44
0.28
-1.91
0.48
0.97
0.55
0.03
0.72
1.35
0.00
0.74
0.00
-0.44
0.00
1.12
0.00
0.29
0.38
-0.21
0.13
-0.55
0.00
-1.53
0.00
-0.12
0.71
1.15
0.00
0.93
0.00
-7.69
0.00
-0.60
0.00
1.09
0.00
-0.28
0.00
-0.64
0.00
0.98
0.00
4.80
0.00
0.78
0.00
0.09
0.61
-3.37
0.00
0.91
0.24
1.60
0.00
0.78
0.00
-0.62
0.69
0.12
0.00
0.88
0.00
0.45
0.00
-0.05
0.01
0.19
0.00
8.42
0.00
-0.34
0.00
-0.29
0.05
-1.27
0.00
0.22
0.15
0.61
0.00
0.16
0.08
2.50
0.15
-0.47
0.00
0.60
0.00
0.53
0.00
-0.30
0.00
0.88
0.00
-4.73
0.00
-0.29
0.05
-0.74
0.00
-3.77
0.00
-0.35
0.07
0.59
0.00
0.19
0.15
-2.12
0.61
-0.33
0.07
0.38
0.00
-0.23
0.15
-0.57
0.00
0.47
0.00
3.57
0.04
0.15
0.69
-0.50
0.25
-0.28
0.64
0.11
0.89
0.86
0.00
0.05
0.90
-1.60
0.01
0.02
0.56
0.88
0.00
0.48
0.00
-0.35
0.00
0.70
0.00
0.22
0.38
-0.01
0.80
-0.29
0.00
-2.33
0.00
-0.04
0.67
0.82
0.00
0.34
0.00
PCT VHIGH RISK
PCT VHIGH RISK
PCT NOINFO RISK
PCT NOINFO RISK
SINGLE FEMALE
SINGLE FEMALE
SINGLE MALE
SINGLE MALE
Estimate
ProbChiSq
Estimate
ProbChiSq
Estimate
ProbChiSq
Estimate
ProbChiSq
-4.32
0.45
13.16
0.03
-0.06
0.53
0.12
0.17
20.21
0.11
-13.98
0.05
0.00
0.97
0.20
0.03
-1.21
0.49
3.56
0.01
0.36
0.00
0.40
0.00
-6.18
0.04
5.80
0.01
0.21
0.00
0.15
0.00
8.24
0.00
4.56
0.01
0.37
0.00
0.18
0.00
7.39
0.00
1.11
0.55
0.19
0.00
-0.32
0.00
10.18
0.03
-8.52
0.04
0.25
0.00
0.47
0.00
5.13
0.00
2.57
0.00
0.29
0.00
0.17
0.00
Table 12 - Loan-Level Regression Results for Refinance Loans 1997
Variable
INTERCEPT
INTERCEPT
ASIAN
ASIAN
BLACK
BLACK
HISPANIC
HISPANIC
LN INCOME
LN INCOME
MISSING
MISSING
CAP RATE
CAP RATE
LN MED INCOME
LN MED INCOME
PCT RENT
PCT RENT
PCT COLLEGE
PCT COLLEGE
PCT ASIAN
PCT ASIAN
PCT BLACK
PCT BLACK
PCT HISPANIC
PCT HISPANIC
DEF
DEF
Estimate
ProbChiSq
Estimate
ProbChiSq
Estimate
ProbChiSq
Estimate
ProbChiSq
Estimate
ProbChiSq
Estimate
ProbChiSq
Estimate
ProbChiSq
Estimate
ProbChiSq
Estimate
ProbChiSq
Estimate
ProbChiSq
Estimate
ProbChiSq
Estimate
ProbChiSq
Estimate
ProbChiSq
Estimate
ProbChiSq
Atlanta
1997
Baltimore
1997
Chicago
1997
Dallas
1997
Los Angeles
1997
New York City
1997
Philadelphia
1997
Whole Sample
1997
11.53
0.04
-1.74
0.10
0.46
0.00
-1.98
0.06
-0.84
0.00
1.46
0.00
-1.92
0.78
-0.65
0.10
-0.42
0.57
0.18
0.89
8.14
0.14
1.79
0.01
1.32
0.39
0.03
0.19
34.81
0.00
-0.23
0.62
1.00
0.00
1.06
0.01
-0.73
0.00
1.57
0.00
-5.22
0.19
-3.01
0.00
-1.35
0.05
1.18
0.22
7.85
0.07
-0.35
0.38
-8.68
0.01
-0.01
0.34
7.66
0.00
-0.13
0.21
1.01
0.00
-0.11
0.03
-0.52
0.00
1.22
0.00
-0.40
0.45
-0.57
0.00
-0.56
0.00
-0.35
0.23
0.75
0.12
1.10
0.00
0.97
0.00
0.01
0.03
-9.22
0.26
-0.56
0.29
0.45
0.04
-0.28
0.29
-0.64
0.00
1.03
0.00
2.77
0.47
0.65
0.33
0.80
0.21
-1.62
0.14
-5.24
0.04
1.78
0.00
-0.14
0.80
0.02
0.32
7.04
0.01
-0.39
0.00
0.72
0.00
0.05
0.39
-0.26
0.00
1.47
0.00
-4.00
0.15
-0.84
0.00
-0.84
0.00
0.44
0.27
-0.08
0.78
0.58
0.01
0.37
0.04
0.02
0.00
9.54
0.00
-1.12
0.00
0.48
0.00
0.00
0.99
-0.96
0.00
1.65
0.00
-3.88
0.05
-0.79
0.00
-0.95
0.00
-1.00
0.04
0.73
0.03
0.55
0.01
0.40
0.10
0.01
0.09
9.56
0.06
-0.91
0.00
0.87
0.00
0.03
0.86
-0.82
0.00
1.46
0.00
0.49
0.79
-0.99
0.03
-0.69
0.19
2.39
0.00
1.82
0.03
1.14
0.00
-0.09
0.83
0.00
0.47
8.80
0.00
-0.47
0.00
0.83
0.00
-0.05
0.15
-0.60
0.00
1.44
0.00
-1.10
0.02
-0.73
0.00
-0.72
0.00
0.16
0.33
0.14
0.42
1.01
0.00
0.75
0.00
0.01
0.00
Variable
PCT VHIGH RISK
PCT VHIGH RISK
PCT NOINFO RISK
PCT NOINFO RISK
SINGLE FEMALE
SINGLE FEMALE
SINGLE MALE
SINGLE MALE
Estimate
ProbChiSq
Estimate
ProbChiSq
Estimate
ProbChiSq
Estimate
ProbChiSq
Atlanta
1997
Baltimore
1997
Chicago
1997
Dallas
1997
Los Angeles
1997
New York City
1997
Philadelphia
1997
Whole Sample
1997
-28.54
0.00
9.77
0.25
-0.13
0.27
-0.02
0.84
3.70
0.76
-0.20
0.98
-0.30
0.00
-0.32
0.00
-6.75
0.00
-0.14
0.94
0.11
0.00
0.16
0.00
10.55
0.24
1.82
0.81
0.35
0.04
0.62
0.00
-0.82
0.84
11.96
0.00
0.11
0.01
0.03
0.44
10.10
0.01
3.48
0.32
-0.21
0.00
0.02
0.73
4.73
0.38
7.62
0.13
-0.18
0.01
0.29
0.00
-1.48
0.28
1.17
0.32
0.00
0.86
0.10
0.00
Table 13 - Loan-Level Regression Results for Refinance Loans 2002
Variable
INTERCEPT
INTERCEPT
ASIAN
ASIAN
BLACK
BLACK
HISPANIC
HISPANIC
LN INCOME
LN INCOME
MISSING
MISSING
CAP RATE
CAP RATE
LN MED INCOME
LN MED INCOME
PCT RENT
PCT RENT
PCT COLLEGE
PCT COLLEGE
PCT ASIAN
PCT ASIAN
PCT BLACK
PCT BLACK
PCT HISPANIC
PCT HISPANIC
DEF
DEF
Estimate
ProbChiSq
Estimate
ProbChiSq
Estimate
ProbChiSq
Estimate
ProbChiSq
Estimate
ProbChiSq
Estimate
ProbChiSq
Estimate
ProbChiSq
Estimate
ProbChiSq
Estimate
ProbChiSq
Estimate
ProbChiSq
Estimate
ProbChiSq
Estimate
ProbChiSq
Estimate
ProbChiSq
Estimate
ProbChiSq
Atlanta
2002
Baltimore
2002
Chicago
2002
Dallas
2002
Los Angeles
2002
New York City
2002
Philadelphia
2002
Whole Sample
2002
-6.86
0.08
-0.39
0.41
1.07
0.00
0.47
0.15
-0.65
0.00
1.23
0.00
4.92
0.36
0.45
0.11
-0.49
0.33
-1.15
0.25
6.39
0.12
1.42
0.00
1.25
0.33
0.00
0.97
6.70
0.35
0.21
0.64
0.65
0.00
0.79
0.02
-0.78
0.00
1.28
0.00
4.33
0.32
-0.68
0.30
-1.49
0.03
-0.76
0.42
9.27
0.04
0.55
0.18
-4.04
0.15
-0.02
0.02
0.73
0.66
0.03
0.73
1.35
0.00
0.74
0.00
-0.44
0.00
1.12
0.00
0.28
0.39
-0.19
0.20
-0.54
0.00
-1.55
0.00
-0.07
0.84
1.14
0.00
0.94
0.00
0.00
0.62
-7.40
0.00
-0.59
0.00
1.10
0.00
-0.28
0.00
-0.65
0.00
0.98
0.00
4.87
0.00
0.76
0.00
0.10
0.60
-3.32
0.00
0.86
0.27
1.58
0.00
0.77
0.00
0.00
0.81
-0.78
0.62
0.12
0.00
0.88
0.00
0.45
0.00
-0.05
0.01
0.19
0.00
8.58
0.00
-0.32
0.01
-0.26
0.09
-1.29
0.00
0.24
0.13
0.60
0.00
0.16
0.10
0.00
0.37
1.82
0.30
-0.47
0.00
0.60
0.00
0.53
0.00
-0.30
0.00
0.88
0.00
-4.15
0.00
-0.23
0.12
-0.64
0.00
-3.81
0.00
-0.22
0.27
0.44
0.00
0.12
0.37
0.01
0.00
-0.37
0.93
-0.33
0.07
0.38
0.00
-0.23
0.15
-0.57
0.00
0.47
0.00
3.25
0.06
-0.01
0.98
-0.63
0.15
-0.17
0.79
0.27
0.76
0.89
0.00
-0.05
0.90
-0.01
0.20
-1.74
0.01
0.02
0.57
0.88
0.00
0.48
0.00
-0.35
0.00
0.70
0.00
0.21
0.42
0.00
0.99
-0.27
0.00
-2.33
0.00
-0.03
0.74
0.80
0.00
0.35
0.00
0.00
0.04
Variable
PCT VHIGH RISK
PCT VHIGH RISK
PCT NOINFO RISK
PCT NOINFO RISK
SINGLE FEMALE
SINGLE FEMALE
SINGLE MALE
SINGLE MALE
Estimate
ProbChiSq
Estimate
ProbChiSq
Estimate
ProbChiSq
Estimate
ProbChiSq
Atlanta
2002
Baltimore
2002
Chicago
2002
Dallas
2002
Los Angeles
2002
New York City
2002
Philadelphia
2002
Whole Sample
2002
-4.34
0.45
13.12
0.03
-0.06
0.53
0.12
0.17
17.23
0.17
-8.06
0.28
0.00
0.97
0.20
0.03
-1.39
0.43
3.64
0.01
0.36
0.00
0.40
0.00
-6.17
0.04
5.52
0.01
0.20
0.00
0.15
0.00
7.66
0.00
4.45
0.02
0.37
0.00
0.18
0.00
8.10
0.00
-0.65
0.73
0.19
0.00
-0.33
0.00
10.72
0.02
-8.86
0.04
0.25
0.00
0.47
0.00
4.93
0.00
2.46
0.00
0.29
0.00
0.17
0.00
Figure 1 - Decline in Sub-prime Lending
Sub-prime Lending
Share, 1997
Change in Sub-prime Lending
Share, 1997-2002
Figure 2 - Philadelphia Subprime Lending Share of Total Loan Volume
1997
(1997 Quintile Breaks)
2002
(1997 Quintile Breaks)