Banking on racism

National Review, Nov 21, 1994 by Ed Rubenstein

Are Minorities held to higher credit standards than whites? Federal regulators certainly think so. Their "evidence" comes from a widely publicized - and fatally flawed - study of three thousand mortgage applications conducted by the Boston Fed in 1992.

The Fed found that rejection rates for minority applicants were much higher than for whites: 28 per cent versus 10 per cent. No surprise there, since white applicants also had higher incomes, better credit histories, and lower appraisal/loan ratios. A close look at the raw data shows, if anything, signs of reverse discrimination:

Old-fashioned discrimination, where clearly qualified black applicants are rejected - and unqualified whites accepted - is nowhere to be seen. Even subtle discrimination seems unlikely, given that blacks and whites receiving mortgages defaulted on them at the same rates. If bankers had rejected a larger percentage of equally qualified blacks, the black default rate would be lower than that of whites.

Yet according to the statistical equation developed by the Boston Fed, marginal black applicants were 60 per cent more likely to be rejected for mortgages. than whites with comparable financial characteristics. To the "discrimination police" this was a veritable smoking gun. Broad new federal requirements for lending to minorities were initiated shortly afterward.

But earlier this year David Home, an economist with the FDIC, took another look at the Boston study. You don't have to be a statistician to appreciate the absurdities Mr. Home uncovered:

- Five rejected minority applicants had sought special low-income loans for which they were clearly not qualified. Several had incomes of over $7,000 per month. They were rejected precisely because they could have easily procured a conventional mortgage.

- The Boston Fed based its conclusions on data in preliminary mortgage applications rather than the final, verified version lenders require for approval. Creditworthiness of rejected applicants is often overstated in the initial application.

- Eight allegedly rejected applications were, in fact, withdrawn when buyers decided not to purchase the property. Six other applicants refused the bank's counteroffer.

- Neither gifts nor home equity was included among the financial variables used to predict whether an applicant would be approved for a mortgage. Since whites are far more likely than minorities to finance homes with these items, their omission created an unwarranted appearance of discrimination by lending institutions.

More generally, FDIC examiners found data in 57 per cent of the instances where an application had been "unfairly denied" according to the Boston Fed. Some were wildly off, like the $3,115,000 loan supposedly approved for a $445,000 house (the actual loan was for $311,500), and the $55,000 loan allegedly procured by a person with net worth of minus $7.9 million. These both involved whites, further biasing the Fed's conclusions.

Do government regulators care about flawed statistics? Probably not. Many banks have already imposed minority lending quotas on themselves, hoping to avoid further regulation or litigation. The big losers, ultimately, will be creditworthy borrowers of all races.

                  Where's the Bias? Financial
            Characteristics of Mortgage Applicants

                                 Approved           Denied
                              White    Black    White    Black

Debt Payment/Income (%)        33.0     34.0     37.0     38.0
Net Worth ($)                93,000   39,000   75,000   33,000
Monthly Income ($)            4,666    3,333    4,471    3,600
% with Poor Credit History     14.6     23.4     38.9     51.5
% Self-Employed                12.0      7.5     22.4      7.4

Source: Federal Reserve Bank of Boston