All last week the housing finance industry got to witness an “epic Twitter war” over the best measure of mortgage denial rates for black mortgage applicants.

David Stevens, CEO at the Mortgage Bankers Association and prolific Graham, Fisher & Co. analyst Josh Rosner got into a row after HousingWire published remarks made by the MBA's Stevens at the industry group's most recent conference in New York City.

And in effect they were both wrong. And they were both right, as it turns out.

In his remarks, Stevens claimed that "[t]he 2012 HMDA data shows a 56% denial rate on GSE purchase loan applications for African Americans," as part of his argument that minorities are already being crowded out of the U.S. housing market.

Rosner wasn't buying those numbers and took to Twitter to express his views.

Critics of Stevens’ numbers pointed to Zillow's data report, which showed that the denial rate for blacks via conventional loans was 33.5%, which was quite different than the 56% figure cited by Stevens, and so the debate continued on.

In the end, Stevens stuck by the MBA’s interpretation of the numbers.

David Moffat, CEO of Mortgage TrueView, says that neither warrior was wrong, and that the two were looking at the data from different perspectives.

“When the dust settles in all this, David is a trade association president who takes a bigger perspective, and when looking at what more needs to be done for minority lenders, I think he is taking the perspective that any loan application not originated is structurally the same as a denial,” Moffat told HousingWire.

The first chart here shows the overall denial rate for blacks (the blue bars) versus whites (the orange bars) for all property types: 18.83% for whites, 34.13% for blacks -- in line with the Zillow data.

He said he would characterize his reaction to the topic raised in the tweet exchange as follows: each side has a basis for their view but both sides overlook the fact that approximately 10% of the loan applications do not include the applicant’s race.

“In addition, whatever the denial rate, more than 10% of the denied loans do not include a denial reason code or indicate that the loan was denied for reasons classified as ‘other,’” Moffat said. 

Moffat said that the MBA’s decision to include manufactured housing data in with the rest of the data is a main point of the divergence. In this category the denial rate for blacks is above 62% some years, while for whites it is above 42%.

“MBA includes manufactured housing (MH) data in its calculation because MBA members lend on this collateral, and both GSEs have MH loan programs and tout these programs as means to provide affordable homeownership to lower and moderate income homebuyers. MBA effectively eliminates ‘chattel’, a type of MH property that the GSEs do not fund, by ensuring that only loans which are deemed “conventional” in the “loan type” field are counted,” Steven said.

Moffat’s assessment is that Stevens is taking a broader perspective, and using the word “denial” beyond it’s strict definition for the purpose of raising an attention about the problem seems to be accurate by Stevens’ own position.

"MBA includes preapproval denials in our calculation. Preapprovals represent a credit check of a borrower who has not yet identified a property. Being denied at this stage is an important signal of availability of credit. A denial is a denial, regardless of where in the loan process it occurs. The fact remains that of the 128,367 purchase mortgages made to African Americans in 2012, only 35,624 were conventional, non jumbo,” Stevens said. “MBA stands by its analysis of HMDA data, and frankly we are disappointed at the direction of the debate which obscures the alarming trend for African American borrowers.”

That said, it may be arguable that Rosner is right in his role as an analyst in making sure data is presented as cleanly as possible.

“Also, any consideration of denial rates should be in the context of a control group, generally considered to be white applicants, when looking at denials rates.  Finally, there are a number of dimensions – including loan purpose and geographical factors – that bear consideration in reaching a conclusion as to cause and, ultimately, a corrective course of action,” Moffat said.