Calculated Risk Vs. Austin Goolsbee, Round Two

In Round 1 of this installment of the great subprime debate, I interpreted Calculated Risk as objecting to certain economists' excessive affection for market outcomes, certain economists including (in CR's view) Professor Goolsbee.  There is a fair amount of religion in these types of arguments. As one of those excessively affectionate economists I prefer to think of a core belief in market efficiency as a reasoned faith, but the truth is we are all operating on assumptions that are difficult to adjudicate with the meager evidence we have at hand.  But there is one part of CR's criticism that, I think unintentionally, omits some pertinent facts, and as a consequence flirts with unfairness:

... I don’t think theory can really explain the revolting disingenuousness at the end of [the Goolsbee] op-ed:

The Center for Responsible Lending estimated that in 2005, a majority of home loans to African-Americans and 40 percent of home loans to Hispanics were subprime loans. The existence and spread of subprime lending helps explain the drastic growth of homeownership for these same groups.

“Drastic”?

This is actually what CRL has to say on this topic:

According to the Fed report, even after adjusting for differences in the borrower characteristics contained in the HMDA data, African-American and Latino borrowers were more likely to receive higher-rate loans. Furthermore, a recent study released by CRL shows that disparities tend to persist even after additional adjustments were made for differences in credit scores, equity, and other risk factors not available in HMDA data. The Fed authors also adjust for originating lender. Though this adjustment reduces the disparities substantially, significant differences remain. . . .

The CRL study found that, even after controlling for legitimate risk factors, African-American and Latino borrowers were still more likely to receive higher-rate subprime loans than similarly-situated non-Latino white borrowers. With raw disparities in higher-rate loans between groups basically unchanged from 2004 to 2005, there is little reason to believe that legitimate risk factors would account for all of the disparity evident in the 2005 data.

In other words, CRL is suggesting that a pattern of finding subprime loans given to minority borrowers with similar credit, income, and equity profiles to non-Latino whites who get prime loans may imply a certain “inefficiency” in the mortgage market somewhere. For Goolsbee to use this data to buttress an unregulated free-for-all by claiming that it helps out the traditionally disadvantaged is, well, dishonest.

The "Fed report" in question is "New Information under HMDA and Its Application in Fair Lending Enforcement," by Robert Avery, Glenn Canner, and Robert Cook.  For the uninitiated, here is what "HMDA" means:

Most lending institutions with offices in metropolitan statistical areas are required by the Home Mortgage Disclosure Act of 1975 (HMDA) to disclose information to the public about applications for home loans and the home loans that they originate or purchase during each calendar year. The law’s requirements arose from concerns that, in some cases, lenders were contributing to the decline of certain neighborhoods by failing to provide adequate home financing to qualified applicants on reasonable terms and conditions. The disclosure of lending activity is intended to help determine whether lenders are adequately serving their communities’ housing finance needs, to facilitate enforcement of the nation’s fair lending laws, and to guide investment activities in both the public and the private sectors. HMDA is implemented by the Federal Reserve Board’s Regulation C.

In a nutshell, this is what Avery, Canner, and Cook find with respect to the pricing of loans to minority individuals:

The foregoing analysis indicates that the information in the HMDA data—that is, adjusting the HMDA data for borrower-related factors plus lender—is insufficient to account fully for racial or ethnic differences in the incidence of higher-priced lending; significant differences remain unexplained.

Does that settle it?  Not even close:

Clearly the HMDA data do not include all the factors that are involved in credit underwriting and pricing. However, by controlling for variations so as to make borrowers as similar as possible on the dimensions of the data that are available, one can account for some of the factors that may explain differences in the outcomes of the lending process among groups...

Explaining the remaining differences is likely to require more details about such factors as the specific credit circumstances of each borrower, the specific loan products they seek, and the business practices of the institutions they approach for credit.

Absent information about specific credit circumstances, specific loan factors, the institutions that particular types of lenders seek to obtain loans, and the like, you are pretty much free to see things as you wish:   

The disproportionate borrowing by non-Asian minorities from higher-priced lenders could occur because of often benign factors such as a ‘‘segmented’’ marketplace in which different lenders offer different products and borrower groups self-select the product-lender combination that best matches their credit or other circumstances. Such a marketplace does not necessarily raise public-policy concerns regarding fair lending: For example, compared with non-Hispanic whites, minority groups on average perhaps because on average they may have less savings to meet down-payment and closing cost requirements), which are typically higher priced and which are the specialty of certain lenders. This explanation could account for differences in lender choice, but demonstrating it requires loan-specific information— such as loan-to-value ratios—as well as other information that is not in the HMDA data.

There is, of course, a non-benign interpretation of the Avery et al results:

However, a situation that might suggest an inadequately functioning marketplace—and that could trigger fair lending concerns—would occur if minority borrowers are incurring prices on their loans that are higher than is warranted by their credit characteristics. Such a problem could arise in one or both of the following circumstances: (1) neighborhoods with high proportions of minority residents may be less well served by lenders offering prime products, a circumstance that would make obtaining lower-priced loans more difficult for well-qualified minorities, or (2) some minority borrowers may be steered to lenders who typically charge higher prices than the credit characteristics of these borrowers warrant.

Calculated Risk's link to the Center for Responsible Lending study is apparently broken, but I think it refers to "Unfair Lending: The Effect of Race and Ethnicity on the Price Subprime Mortgages," by Debbie Gruenstein Bocian, Keith Ernst, and Wei Li.  The study purports to rectify some of the ambiguity in the Federal Reserve analysis:

This study extends previous analyses of home loan pricing disparities by supplementing HMDA data with additional loan-level information from a large, proprietary subprime database. By merging the datasets, we were able to evaluate whether race and ethnicity affect subprime loan pricing after controlling for key risk factors, including credit scores and loan-to-value ratios. The results show that African-American and Latino borrowers are more likely to receive higher-rate subprime home loans than white borrowers, even when we control for legitimate risk factors.

So that settles it?  Not really:

This analysis does not allow us to estimate precisely how much race and ethnicity increase the prices charged to borrowers. It is also beyond the scope of this paper to determine definitively why these disparities exist. However, we do posit several possible causes, including the considerable leeway mortgage originators have to impose charges beyond those justified by risk-based pricing.

The CRL -- a self-proclaimed "Resource For Predatory Lending Opponents" -- definitely choose the non-benign spin on the Bocian et al results.  But having admitted that their study cannot definitively determine why disparities in loan prices exist, it would appear that we have returned to the realm of religion.

So how about a little inter-faith charity?  Amen.