Fed Governor Sees Harm In Relying on Statistics To Set Fair-Lending

ANNAPOLIS, Md. - Federal Reserve Board Governor Lawrence B. Lindsey says he's worried about the growing use of statistics in fair-lending analysis.

Regulators are placing the loan process at risk by using numbers to back up fair-lending claims, he said in an address last week.

"We may ultimately be trivializing an important issue," Mr. Lindsey said at the Mid-Atlantic bank compliance conference held here. "Statistics may be obscuring the facts."

Taking statistics too seriously could stifle the way bankers decide to make loans, Mr. Lindsey said.

By the end of the decade, he predicted, the banking industry will be dominated by computer analysis rather than by people.

Ironically, that shift may prevent rather than promote lending to the people whom fair-lending laws are designed to protect, he said.

"We humans have a sense of justice that transcends . . . what a computer can provide," he said.

Mr. Lindsey singled out two methods of fair-lending testing for reproach: matched-pair testing and regression analysis.

In both types of study, he said, the results can lead to wrong conclusions.

"Statistics can be used to confuse as well as enlighten," Mr. Lindsey warned.

On its face, matched-pair testing seems straightforward, Mr. Lindsey said. The technique calls for studying perfectly matched pairs of loan applicants - similar to each other in every way except for what is being tested for, such as race or sex. But such pairs are extremely difficult to find, Mr. Lindsey said.

In addition, these tests only look at a few applicants, he pointed out. "The statistics you see cannot be viewed in isolation," Mr. Lindsey said.

Regression analysis has different problems, he said. Rather than matching specific people, it assesses a bank's overall decision-making process.

"A probability-based model cannot be used to prove anything about a single individual," said Mr. Lindsey.

Even more troubling, he said, is the possibility that juries and judges unfamiliar with the banking industry may be asked to look at regression analysis as part of a fair-lending trial.

Statistical studies, such as a widely publicized one conducted by the Federal Reserve Bank of Boston, should not be considered flawless, Mr. Lindsey said.

"Research always pursues truth, it never finds it," he quipped.

Mr. Lindsey also criticized the idea of disparate treatment, under which a class of people are intentionally or unintentionally discriminated against.

The law should allow bankers to know if an action they are taking is illegal, Mr. Lindsey said.

In the current fair-lending arena, he told the bankers, "you've got a pretty narrow path to tread."