In their analysis of the new HMDA data ("New Information Reported Under HMDA and Its Application in Fair-Lending Enforcement," Federal Reserve Bulletin, summer 2005), Federal Reserve Board researchers document substantial racial disparities in the cost of credit.
But after laying out the basic patterns, the paper quickly backtracks in efforts to explain away the uncomfortable findings.
A basic message that comes through is the possibility, if not likelihood, that with sufficient data on the right "objective" measures the racial disparity would disappear. And if this were the case, according to this perspective, the problem of racial discrimination would virtually disappear, except perhaps within a few selected institutions.
The notion that the presence or absence of a statistically significant race coefficient in a multivariate model is the touchstone for determining whether racial discrimination is present, however, is wrongheaded and dangerous. (I'm reminded of the statistical joke that if you torture the data long enough, eventually they will surrender.)
Though the race effect has probably not disappeared (the National Community Reinvestment Coalition found race statistically significant in previous lending studies, even after controlling for credit rating and housing characteristics), its disappearance would not be the end of the discussion. It would simply be the beginning of a much more complex discussion about the continuing role of race in credit markets and American society in general.
But if controlling for income, wealth, property characteristics, credit rating, type of loan, loan-to-value ratio, type of lender, and the rest resulted in a race coefficient that was zero or not statistically significant, this raises many questions and answers few.
If the race effect disappeared in this manner, it would mean that the large unadjusted disparities could be "explained" by the fact that racial minorities earn less money, accumulate less wealth, purchase homes in worse repair, have lower reported (but often inaccurate) credit ratings, need to borrow a higher share of the cost of their housing, and are limited to fewer but higher priced lenders.
All of this raises the question of why.
Part of the answer, no doubt, is that minorities on average attend inferior schools, which reduces their ability to compete in the labor market, undercutting their income and potential wealth accumulation.
Employers still systematically discriminate (as the economists Marianne Bertrand and Sendhil Mullainathan reported in "Are Emily and Greg More Employable than Lakisha and Jamal? A Field Experiment on Labor Market Discrimination," published last year in The American Economic Review). Such discrimination in education and employment no doubt contributes to the fact that minorities generally have higher loan-to-value ratios and lower credit scores.
Discrimination in the nation's housing markets continues to limit choices for racial minorities and the appreciation in the value of the homes they are able to buy.
Though discrimination may have been reduced in the 1990s, the latest HUD/Urban Institute housing-discrimination study found that African-Americans and Latinos encounter disparate treatment in one out of every five initial visits to a rental or sales agent. Historical and ongoing segregation continues to restrict options for racial minorities on a range of fronts, as Douglas S. Massey and Nancy A. Denton documented in their classic book "American Apartheid" and as fair-housing researchers and advocates continue to document.
Segregation continues to lead to fewer housing options and lower wealth accumulation associated with homeownership, inferior neighborhood public schools, higher neighborhood crime rates, fewer job opportunities, and a range of factors that contribute to socioeconomic failure.
One consequence of this panoply of conditions is that property values, and therefore property tax revenues, are suppressed. This of course leads to even fewer resources available to finance public education, which reinforces the process of neighborhood decline and perpetuates this vicious cycle over generations.
Many factors contribute to these patterns. Among them are historical, if not contemporary, practices by financial service providers that at least up until no more than two generations ago overtly and explicitly redlined urban neighborhoods and minority borrowers.
These practices, often nurtured by public policy, created the wide-ranging forms of racial inequality that apparently are now used to justify "unadjusted" racial disparities in mortgage lending.
The point is not to place the entire blame for racial inequality on financial service providers. But it is critical to understand that even if lenders are simply responding objectively to the objective risks that various borrowers present today, this does not mean racial discrimination is irrelevant or that lenders have no culpability.
The Fed's analysis does acknowledge its role in pursuing further investigations if the HMDA data reveal particularly large disparities for selected institutions. Its paper notes a range of steps the Fed might take in these circumstances.
But those steps do not include paired testing - equally qualified white and nonwhite "mystery shoppers" approaching the same service providers to see if they are treated differently - perhaps the most effective and efficient tool for documenting racial discrimination.
Testing has been used by other enforcement agencies and researchers to document discrimination in home rentals and sales, property insurance, automobile sales, other commercial businesses, and mortgage lending. The Fed's resistance to using this tool undercuts its ability to pursue discrimination cases, even in the few instances where it may be inclined to do so.
Racial discrimination, of course, is much more complicated than simply the interaction between two actors in which one intentionally refuses to serve the other because of race. Intentional discrimination rooted in overt prejudice has been reduced in recent years. But the institutional context that has long nurtured racial inequality has not changed as much and still yields unequal and unfair outcomes, as indicated by the disparities in the cost of credit for various racial groups documented by the new HMDA data.
This is part of why CRA and related tools remain essential, and why we cannot allow the presence or absence of a statistically significant race coefficient to be the sole determinant of whether race remains a problem in credit markets.
