What the government gets wrong about fair lending
There’s been an ongoing debate about Home Mortgage Disclosure Act reporting requirements as a result of a measure included in the Senate’s regulatory relief legislation.
It would be useful if these discussions were informed by an understanding of some of the anomalies involved in fair-lending enforcement policies, not just under the last administration, but ever since the government began giving serious attention to racial differences in mortgage rejection rates in the early 1990s.
As I’ve written before, there is a perverse feature of fair-lending enforcement arising from a fundamental misunderstanding of statistics. Enforcement was premised on the belief that relaxing standards would reduce the several-fold differences between the rates at which racial minorities and whites commonly had their home mortgage loan applications rejected or experienced other adverse outcomes. But, in fact, relaxing standards tends to increase those differences.
It is true that relaxing lending standards tends to reduce relative (percentage) racial/ethnic differences in rates of meeting those standards. Thus, lowering lending standards tends to reduce relative racial/ethnic differences in mortgage approvals and other favorable borrower outcomes.
Yet relaxing standards also tends to increase relative differences in failing to meet the standards.
As I explained in a letter to Attorney General Jeff Sessions last April, income and credit score data show — in a way that is hardly open to debate — that lowering income or credit score requirements for receiving some favorable borrower outcome, while tending to reduce relative differences in meeting the requirements, will tend to increase relative differences in failing to meet the requirements. For example, as shown in Table 2 of the letter, white and black rates of meeting an income requirement of $85,000 are 34.6% and 17.3%. Thus, the white rate is twice the black rate. Lowering the requirement to $50,000 would increase the white and black rates of meeting it to 61.0% and the 39.2%. The white rate of meeting the lower requirement would be only 56% greater than the black rate.
But white and black rates of failing to meet the $85,000 requirement are 65.4% and 82.7%. Thus, the black rate of failing to meet the requirement is only 26% greater than the white rate. Lowering the requirement to $50,000 would reduce the white and black rates of failing to meet it to 39.0% and 60.8%. The black rate of failing to meet the lower requirement would thus be 56% greater than the white rate.
Unaware that relaxing standards tends to increase relative differences in failure to meet the standards and associated adverse borrow outcomes — in fact, believing the opposite — federal agencies have monitored lender compliance with fair-lending laws on the basis of relative difference in adverse borrower outcomes at the same time that the agencies have encouraged lenders to relax standard in order to reduce the disparate impact of those standards on minority borrowers.
Significantly, this created the anomaly whereby lenders that followed government encouragements to relax standards increased the chances that the government and others would sue them for discrimination.
A variation on this anomaly may be found in cases like those the city of Miami brought against Bank of America and Wells Fargo. Miami alleged that discriminatory lender practices caused financial harm to the city as a result of the disproportionate concentration of foreclosures in minority neighborhoods. The Supreme Court upheld the legal basis for the cases in a decision last May, despite raising the bar for plaintiffs to prove their claims as the case proceeds in an appeals court. But those problems did not deter Philadelphia from immediately bringing a similar lawsuit against Wells.
In these cases, the problem is that the actions lenders — or the government — take to generally reduce foreclosures tend to increase the concentration of foreclosure in minority neighborhoods and thus the perceived strength of the cities’ claims.
So far, however, these issues are unknown to federal enforcement agencies, Congress or the courts. Even when they become known, however, there will remain a question as to whether the disparate impact doctrine is unconstitutionally vague in many contexts. For it requires that covered entities adopt less discriminatory alternatives to practices having a disparate impact. But knowing what practices have the lesser impact can be problematic when modifying standards tends to reduce the disparate impact as to one outcome, even as it increases the disparate impact as to the opposite outcome.