A good example of data-based investigative journalism from the Dayton Daily News..big front page article:

 

Well -off blacks denied loans more than low income whites

 

Subheader:

 

A Dayton Daily News investigation: In all income levels, blacks are more likely than whites to be rejected for home loans and are more likely to be sold a high-cost, subprime loan.

 

DAYTON — More than four decades after racial discrimination in home lending was outlawed, lenders in the Dayton area deny a much higher percentage of loans to blacks than whites, even when income levels are comparable.

 

Blacks are also more likely than whites to be sold a high-cost, subprime loan.

 

A Dayton Daily News examination of 2008 Home Mortgage Disclosure Act data found significant racial disparities exist at every income level.

 

 

The article has this graphic of denial rates (there is another one mapping high-cost mortages):

 

 

The DDN has related articles.  One is particularly interesting from a geographical POV:

 

Neighborhoods in danger when home loans dry up

 

Fred Steed and his wife Florence can afford to live anywhere in the Dayton area.

 

But they prefer to live in the West Dayton home and neighborhood where his family has lived since 1952.

 

>snip<

 

Steed’s Westwood neighborhood has its share of crime, drugs and the ill effects of poverty normally associated with inner city communities. But it also has its share of residents like Steed: people who want to do what they can to make their community a better place.

 

And if the Fred Steeds of Westwood can’t get loans, the neighborhood doesn’t stand a chance.

 

The article continues in-depth, providing bankers and appraisers POVs, as well as those of fair housing advocates.

 

 

Wait is that where they want to get a mortgage at or where the person lives at the time of the loan?

Hard to say.  It  seems across the board from lenders.  The lenders say they dont use racial profiling (and it could be that they dont;), but they are not providing all the criterea they use to deny a loan. It seems to be the combination of criterea add up to this discrepancy by race that the paper uncovered. 

 

The cumulative effect hits areas hardest where there is a majority of blacks, with the same effect as redlining of the old days...urban decline  due to the unavailabilty of home finance.

 

 

This is definitely troubling, to put it mildly.  It has always been a little annoying to me that banks treat their lending formulas as such great secrets.  I would honestly defend them even if there were a little bit of wiggle room for arbitrariness in the formulas (discretion of the loan officer, etc.)--I think that that is defensible.  But prospective applicants should know that a loan officer has a degree of discretion in the approval process, if he/she does, before they walk into the bank.

 

I consider this irritation of mine a faint echo of the seething hatred I harbor for TransUnion, Equifax, and Experian, who have the official power to ruin lives and not tell anyone why.  An individual bank is much easier to avoid than the three big credit rating agencies.

 

With respect to the DDN article: I'd be particularly interested in seeing the data broken down by bank.  The only one specifically named in the article was Cincy's Fifth Third.  Was the disparity roughly equal among all the banks operating in the area?  Were national banks more likely to turn down a minority applicant than regional banks, which in turn were more likely to turn the person down than local ones?  Or was it just the opposite?

 

Most importantly, did they subdivide the 8-county area of the study, or did the aggregate it?  The reason I ask this is one factor that is definitely important to any bank (or any secured lender, for that matter) is whether the collateral is likely to maintain its value.  The article mentioned that they turned down an African-American with a credit score of 820 (whatever the heck that means ... see previous reference to seething hatred of credit bureaus) for a refinance in a neighborhood that was 99 percent minority.  Would the exact same loan to the exact same person (including status as a minority) been approved had the collateral been a physically identical house located in Beavercreek or Centerville?  Or, another way of asking it: Are even upper-income blacks often applying for mortgages on houses that are in at-risk neighborhoods in which the bank is likely to lose more money on the house in the event of foreclosure?

 

I would be very interested in seeing the DDN follow up on this.  I respect corporations' rights to trade secrets, but I also think they deserve the PR consequences of choosing to keep information that is extremely relevant to their customer base (or potential customer base) hidden from them.

^As of 2008, a vast majority of mortgages being originated were FHA or GSE loans, so the individual bank underwriting formulas were probably not very different from one another, effectively.  But application results might differ a great deal do to their branch locations, etc. which would draw different applicant pools.

 

Wait is that where they want to get a mortgage at or where the person lives at the time of the loan?

 

It's where the property that would be mortgaged is located.

 

The biggest problems with HMDA data are the lack of borrower credit info and the lack of LTV/CLTV data.  It's really hard to look at the raw racial disparities ad know what to  make of them, because the mostly black census tracts likely have much lower property values (of course you'll get rejected for a $35k refi loan if the bank thinks your home is worth only $16k; the example in the article) and because blacks, on average, have lower credit scores even when controlling for income.  This obviously has sinister ripples as credit ratings are now used to calculate car insurance rates and even as a screening mechanism for employment.

 

The reason I ask this is one factor that is definitely important to any bank (or any secured lender, for that matter) is whether the collateral is likely to maintain its value.

 

Gramareye, interestingly, i don't think the GSE or FHA underwriting formulas do break down expected collateral performance at this fine a geography.  For a time (which I think included 2008), Freddie and Fannie had more stringent underwriting in "declining markets," but this caused a political firestorm, so I think it was lifted.  And even then, I think the "declining market" label was assigned at a metro-wide level.  I'd love to hear anyone's first hand experience if it was different, particularly from anyone here who was a mortgage broker or loan officer.

Where did you get the figure that the example in the article was of a home that was only worth $16k?  I didn't see that stat.  Or is it just that you know this area and this is a street of $16k homes?

^The photo caption (though I got the number wrong- $14k):

 

Fred Steed grew up in his house at 166 Lorenz Ave., where he still lives with his wife, Florence. Steed, the director of community health for Public Health – Dayton and Montgomery County, applied for a $35,000 refinance loan from Fifth Third Bank in August. Despite having a high income and a credit score of 820, he was denied the loan after the home was appraised at $14,000.

 

Even if the underwriting is indifferent to geography in the narrowest sense, the collapse of values in a lot of inner city neighborhoods will obviously have some pretty devastating consequences.  And is also why, IMHO, there is still a roll to be played by carefully crafted lending programs, including the CRA.

Someone was asking about the numbers by bank. Here is a link to set of tables:

 

By bank: Racial discrepancies in loan denial rates (2008)

 

...and...

By bank: Racial discrepancies in high-cost home loan rates (2008)