Chicago,
near 42nd and east of State Street, September, 2005
New Jersey Institute for Social Justice resources
November
23, 2005
Yahonnes,
I’ve
revised the code to produce a new set of lender rankings. Let me know how this
looks.
best,
Elvin
November
16, 2005
Yahonnes,
I’ve
written code that reads in the raw Home Mortgage Disclosure records for the state
of New Jersey, and ranks the individual lenders
according to
their racial disparity in the granting of rate-spread loans. First, we narrow
the focus to only those lenders reporting at least
one hundred
total applications. Then we calculate the percentage of African-American
originations that are made at rates above the
reporting
threshold; we divide this percentage by the percentage of non-Hispanic White
originations made on rate-spread terms. In cases
where no
rate-spread loans are made to non-Hispanic Whites, this type of calculation is
mathematically impossible (any number divided by
zero is
infinity), so I’ve simply written the code to assign these cases “100” values.
But the code also prints out the absolute number of
African American
(and Latino/Latina) rate spread loans, so you can see the many cases in which a
lender made no rate-spread loans to
Whites, but only
a handful of rate-spread loans to minorities. The crucial institutions, of
course, are those with a large number of rate-
spread loans to
African Americans or Hispanics, and high Black/White or Hispanic/White ratios.
You’ll note that the lenders with the
highest ratios
(above ten) are, in most cases, simply making very few rate-spread loans at
all, but doing so only with African Americans.
Further down the
list are somewhat lower ratios (from 5.9, Wells Fargo, to 2.8, Countrywide) but
based on significant numbers of
originations to
African Americans. Farther down the list are “equal opportunity” predators like
Fremont Investment and Loan, where
four-fifths of
all originations are rate-spread, but there is almost no Black/White or
Hispanic/White disparity. All of these statements
apply to
originations regardless of loan purpose, do not distinguish between first- and
second-lien loans, but do exclude loans purchased
by institutions,
applications received in late 2003 but acted on after January 1, 2004, and
FHA/VA-insured loans.
ratios.xls,
an Excel file with New Jersey lenders ranked by Black/White ratio
ratios.lst,
a SAS printout (open this from within Word)
ratios.sas,
the SAS command file, in case you’re interested in how things are defined.
Chicago,
where the Robert Taylor Homes once stood, September, 2005
November
7, 2005
Yahonnes,
I’ve
prepared a set of files for you in case you are able to get access to SAS,
statistical analysis system software. If
you can, create a
directory,
“c:\sasdat\njisj”. Download the files
below into this directory and run the SAS code, which will produce a series of
crosstabulations
of first-lien conventional originations statewide, by race/ethnicity and
income. The code also compares the rate-
spread and HOEPA
status of first-lien conventional originations, and presents the race/ethnicity
and income categories of first-lien
conventional
HOEPAs. If you don’t have access to SAS, then just use the second file below to
follow the logic that I’ve used to define
race/ethnicity
categories, and let me know what additional cross-tabulations would be most
useful.
Loan application
register data for New Jersey
as defined in the 2004 HMDA Cds
‘codes.pdf,’
presenting the codes corresponding the 2004 HMDA Cds
‘nj2004.sas,’ a
SAS command file
nov07.doc, a Word document that presents the SAS output from the code above
Yaletown,
Vancouver, August 2004.
Copy Left 2005, Elvin K. Wyly.