Gray
and Green: English Bay, Vancouver, November 2005
Fair
Finance Watch resources
March
30, 2006
For sample code
to combine multiple dat files, see this.
best,
elvin
March
5, 2006
I’ve revised the
‘wamu’ code from last year, to produce very quick tabulations of rate-spread
loans by race/ethnicity (I’ve done nothing
yet on
income.) First, you must open up the
“dat” file that the lender sends, and strip out the first one or two lines of
text -- the lines
with
identification information for the lender itself. These lines are not formatted
the same way as all the other data lines, so this step
is necessary in
order to get sas to read in the code.
You can do this in any text editor.
Then save the results. Here’s an example of the
‘stripped’
version for icb. Then you can open SAS and run this code,
making relevant changes to the directory structure I’ve used (in
this case, my
working files are in c:\sasdat\icp). The
result will be two files in your c directory, which can be opened in Excel: this one
presents the
ratespread/non-ratespread tabulations (ratespread=1) for all approved and
originated first-lien conventional single-family
loans; this one presents a breakdown of action taken for all
applications.
best,
elvin
Granger,
Washington, February 2006
January
5, 2006
Matthew,
When you kindly
sent me the *dat files for Wells and the other lenders, I took them through
SPSS, exported to an older version
of SAS, then
read them in with this file. Read this code file just
to see how I defined the variables and such -- don’t run this code,
because you do
not have the intermediate ‘temp’ files referenced in it. The final product of this batch code, however, should
be
useful if you
want to perform analyses on these lenders. It’s about two gigabytes, so please
let me know after you’ve downloaded so
I can clear it
off our server.
best,
elvin
December
13, 2005
Matthew,
I’ve revised the
“newwamu” code again (still called ‘xlswamu’.) If
you run this code using the same directory
structure you’ve
been working with recently, it will automatically produce an excel file,
storing the results in “c:\sastable.xls”.
Note that I’ve
changed the coding of race and ethnicity slightly -- so that Hispanics who do
answer the ethnicity question but
who do not
furnish information on race are classified as Hispanic, not
“missing/unreported.” Is this the kind of table that
you’d prefer?
Let me know, and I can try to revise and reformat as needed.
best,
elvin
San Diego, tourist playground in
the gunbelt: Battleships and bistros; guns and galleries; seafood and seamen.
November
22, 2005
Matthew,
I’ve
revised ratios.sas (the code described below) in three ways. First, the code is now restricted with a set
of fips codes corresponding
to census tracts
in the Bronx; revise these to focus on the areas you want to analyze. Second,
the code drops the 100-minimum lar
restriction.
Third, the code now prints out the total number of white hicost originations,
as well as race/ethnicity missing applicants.
The code is here.
best,
Elvin
November
15, 2005
Matthew,
I
have completed major revisions on the sas command file. It is now designed to
handle lenders that made no ratespread loans to
non-Hispanic
whites, and to prepare a more readable output. You should first download a new,
re-formatted version of the
Transmittal sheet data
for lenders , however; this file replaces the spaces in long lender names
with underscores (“_”). This
allows us to use
the “Dynamic Data Exchange” function to transfer some of the data to Excel
after SAS finishes processing (if the
lender names
have spaces, then Excel mistakenly puts each phrase into a separate column of a
worksheet). So, assuming you’ve
already got the
New York lar downloaded, you should follow four steps: first, download the Transmittal sheet
data for lenders
and replace the
old one you’ve been using; second, download the revised ‘ratios.sas’
code, and open it in the SAS program editor;
third, open a
new, blank Microsoft Excel worksheet, and save it on your computer under the
filename “ddedata”; fourth, toggle
back to SAS
(still keeping Excel running with that ddedata file open) and run the command
file. After processing, you should
get readable
output in sas, and back in Excel, a worksheet with a ranked list of lenders and
their rate-spread lending disparities.
best,
Elvin
November
6, 2005
Matthew,
The
files below are written assuming that you can create a directory,
“c:\sasdat\ffw” (for Fair Finance Watch). Save these files into
this directory
and run the SAS code. You’ll get a ranked list of lenders in the New York
metropolitan area that have reported at
least 100
applications in 2004, along with series of cross-tabulations:
Loan application
register data for New York metro area
as
defined in the 2004 HMDA Cds
Transmittal sheet
data for lenders
exported
from 2004 HMDA Cds and edited to correct some (but not all) text errors
‘nov05.sas,’ a SAS
command file that creates a ranked list and crosstabulation of lenders
The
list is ranked in descending order according to the share of all originations
that are rate-spread; the analyses then
produces
cross-tabulations for those lenders with at least 100 applications in the study
area and where at least 80 percent
of
origination activity involves rate-spread loans. Cross-tabulations display action taken on
loan request, by race/ethnicity,
by
applicant income.
Copy Left 2005, Elvin K. Wyly.