James P. Scanlan, Attorney at Law

United States v. Wells Fargo

(July 17, 2012; rev. March 21, 2013)

Prefatory note added July 3, 2015:  Since the last material updating of this page, I have discussed this case (along with United States v. Countrywide) in a quite a few places, including (1) an amicus curiae brief of in Texas Department of Housing and Community Development, et al. v.  The Inclusive Communities Project, Inc., Supreme Court No. 13-1731 (Nov. 17, 2014); (2) “Race and Mortality Revisited,” Society (July/Aug. 2014); (3) The Perverse Enforcement of Fair Lending Laws (Mortgage Banking, May 2014); and (4) “The Mismeasure of Discrimination,” Faculty Workshop, University of Kansas School of Law (Sept. 20, 2013).  In addition to addressing the principal measurement problem discussed on this page, items 1 (Section I.C at27-30), 3 (at 93), and 4 (at 32 to 35) discuss the fundamental flaws of analyses, regarding claims of either putatively discriminatory assignment to subprime status or putatively discriminatory loan costs, that examine solely persons who accepted an outcome or situation. 

Prefatory note:  This is a subpage to the Lending Disparities page of jpscanlan.com.  Other subpages include Disparities – High Income, Underadjustment Issues, Absolute Differences – Lending,  Lathern v. NationsBank, United States v. Countrywide, File Comparison Issues, Partial Picture Issues, Holder/Perez Letter, and Federal Reserve Letter.  . The Partial Picture Issues sub-page addresses a problem with the claims of both discrimination in assignment to subprime status and discrimination in loan pricing in the Wells Fargo case ( and the Countrywide case) that were not present in cases and studies about racial differences in mortgage rejection rates – i.e., analyses of the subprime and pricing claims fail to examine the entire universe of persons seeking a desired outcome.  The underadjustment issue addressed below is also the subject of my “Statistical Quirks Confound Lending Bias Claims”(American Banker, August 14, 2012).

On July 12, 2012, the Department of Justice announced a $175 million settlement of claims that Well Fargo Bank discriminated against black and Hispanic home loan borrowers.  The settlement was DOJ’s largest in a fair lending case after the $335 million settlement with Bank of America’s Countrywide Financial Corporation announced last December.  Like the complaint in the Countrywide suit (subject of the United States v. Countrywide sub-page), the complaint against Wells Fargo faults the defendant for failing to minimize the proportion of loans that were subprime.  Thus, it implicates the perverse anomaly characterizing fair lending enforcement at least since 1994 whereby lenders are encouraged to do things that make them more likely to be sued.  See my “Misunderstanding of Statistics Leads to Misguided Law Enforcement Policies” (Amstat News, Dec. 2012), “’Disparate Impact’:  Regulators Need a Lesson in Statistics” (American Banker, June 5, 2012), and “The Lending Industry’s Conundrum” (National Law Journal, Apr. 2, 2012).  That is, the Department of Justice and other agencies involved in fair lending enforcement encourage lenders to reduce adverse outcome rates; but, apparently unaware that reducing adverse outcome rates, while tending to decrease relative differences in favorable outcomes, tends to increase relative differences in adverse outcomes, those agencies continue to evaluate lender practices on the basis of relative differences in adverse outcomes.

That a suit/settlement of this nature was in the works is a reason why the Department of Justice would be reluctant, after having the matter brought to its attention by the referenced articles or my letter of April 23, 2012 (the subject of the Holder/Perez Letter sub-page), to recognize and acknowledge that for nearly two decades it had been encouraging lenders to take actions that make them more likely to be sued.  But there are ample other reasons for the Department not to recognize the perverseness of its enforcement policies, assuming that the letter was directed to persons who understood the issue.    

In any event, like the analysis of racial and ethnic differences in the Countrywide complaint, the analysis in the Wells Fargo complaint is subject to the questions about adequacy of adjustment for demographic differences in credit-related characteristics that are the subject of the Underadjustment Issues sub-page.  Further, Paragraph 2 of the Wells Fargo complaint casts additional light on those issues.

The paragraph states (footnotes omitted; bracketed numbers added):

“[1] As a result of Wells Fargo's policies and practices, between 2004 and 2008, approximately 4,000 qualified African-American and Hispanic wholesale borrowers, who received Wells Fargo loans through mortgage brokers, received subprime loans rather than prime loans from Wells Fargo because of their race or national origin, not based on their creditworthiness or other objective criteria related to borrower risk.  [2] These African- American and Hispanic borrowers were placed into subprime loans, with adverse terms and conditions such as high interest rates, excessive fees, pre-payment penalties, and unavoidable future payment hikes, when similarly-qualified non-Hispania white ("white") borrowers received prime loans.  [3] For example, between 2004 and 2008, highly qualified prime retail and wholesale applicants for Wells Fargo residential mortgage loans were more than four times as likely to receive a subprime loan if they were African-American and more than three times as likely if they were Hispanic than if they were white.  [4] Conversely, during the same time period, borrowers with less favorable credit qualifications were more likely to receive prime loans if they were white than borrowers who were African-American or Hispanic.”

A footnote to sentence [3] read:

“For purposes of this paragraph, highly qualified prime applicants for Wells Fargo residential mortgage loans had the following characteristics: FICO scores equal to or greater than 680, debt-to-income (‘DTI’) ratios less than or equal to 45% of the loan amount, loan-to-value (‘LTV’) ratios less than or equal to 80% of the loan amount, and no history of bankruptcy.”

A footnote to sentence [4] read:

 “For purposes of this paragraph, Wells Fargo borrowers with less favorable credit qualifications had the following characteristics: FICO scores between 620 and 680, DTI between 45% and 55% of the loan amount, and LTV between 80 and 90% of the loan amount.”

Sentence [3] is quite provocative, given that in context it suggests that the identified differences in assignment to subprime are the result of discrimination, and it was picked up by the media in several accounts of the settlement. Certainly it would suggest to many that it is highly unlikely that, absent discrimination, one would find such large disparities even among highly qualified applicants – where there is no apparent reason why any loan should be assigned to subprime status much less why it should happen to minorities much more often than whites.  Few observers, however, would appreciate that racial and ethnic differences in assignment to subprime status tend to be large among highly qualified applicants – and probably larger than among less qualified applicants – simply because rates of assignment to subprime status tend to be very low among highly qualified applicants.  Differences in rates of assignment to prime status, however, tend to be small among such applicants.  In any case, seemingly large relative differences in some outcome are much less probative of discrimination when the outcome is rare than when it is common.

The emphasis on the large disparity in adverse outcomes among highly qualified borrowers is reminiscent of the manner in which, in the 1990s, the fact that rejection rate disparities were larger among high income borrowers than low income borrowers was proffered as a rebuttal to contentions that unaccounted for income differences might explain racial differences in mortgage rejection rates.  Even if it were soundly shown that racial differences in outcomes were greater among high-income groups, the argument made no sense.  Such a pattern would merely indicate that whatever the forces driving the differences (including unaccounted for income differences), such forces were greater among high income groups.  But entirely overlooked was the fact that rejection rates disparities tend to be large (and approval rate disparities tend to be small) among high income groups simply because rejection rates are low among high income groups.  See the Disparities – High Income sub-page.  See also “Can We Actually Measure Health Disparities” (Chance, Spring 2006), “Race and Mortality” (Society, Jan/Feb 2000), and Section E.1 (Whitehall Studies) of the Measuring Health Disparities page regarding the varied misinterpretations of the implications of large relative differences in adverse outcomes within comparatively advantaged subpopulations.

Sentence [4] references relative differences in the favorable outcome (receipt of prime loans), though the sentence does not quantify the disparity.  The disparity in assignment to prime loan status commonly would be larger in the less favorable credit group than the highly qualified group because receipt of prime rather than subprime loans is less common in the less favorable credit group (while the disparity in assignment to subprime would be smaller in the less favorable credit group because that outcome is more common there).  Probably that is why the drafters mentioned differences in the favorable outcome rather than differences in the adverse outcome for this group.  Were the agencies to start evaluating lender performance on the basis of relative differences in favorable outcomes it would essentially revolutionize fair lending enforcement, often causing views as to which were the least and most discriminatory lenders to be reversed.  See the Lathern v. NationsBank sub-page and the January 1, 1996 Legal Times article “When Statistics Lie” concerning the study that led to a putative class action against NationsBank and which, while showing that NationsBank showed one of the smallest relative differences in approval rates.  In any case, while reducing the frequency of subprime assignment as the complaint suggests the defendant should have done will tend to increase the relative differences in outcome rates identified in sentence [3], it will tend to reduce the relative difference in outcome rates identified in sentence [4].

The crucial issue in an appraisal of the potential merit of the claim that either the disparity in adverse outcomes referenced in sentence [3] among highly qualified applicants (HQA) or the disparity reference in sentence [4] among less qualified applicants (LQA) is whether white and minorities within each of the two categories can in fact be deemed comparable with respect to factors that affect assignment to subprime status. 

As it happens, data from another case against Wells Fargo was already being used in the Underadjustment Issues sub-page to illustrate problems with efforts to adjust for differences in characteristics (as well as in the Credit Score Illustrations sub-page of the Scanlan’s Rule page to generally illustrate the patterns by which differences in outcome rates tend to be affected by the overall prevalence of an outcome).  Table 1 below, which is based on that data, shows the proportion black borrowers comprise of the combined black and white borrowers and the proportion Hispanic borrowers comprise of the combined Hispanic and white borrowers.[i]  The second column indicates the categories used in Paragraph 2 of the complaint as well as applicants with even less favorable qualifications than the LQA group.

Table 1:  Proportions Black and Hispanic Borrowers Comprise and Combined Black and White or Combined Black and Hispanic Borrower (data from expert report in In re Wells Fargo litigation[ii])

Table 1 shows that within each of the DOJ’s two categories blacks and Hispanics have lower credit scores than whites. That would also be so with regard to any more refined categorization within the two ranges identified in the complaint.

The pattern of credit scores strongly suggests that one would observe a similar pattern with regard to debt-to-income ratios within the categories identified in the complaint.  I am inclined to think there would be a similar pattern with regard to loan-to-value ratios.  But, inasmuch as such ratios turn on things not directly related to general creditworthiness factors, that would not necessarily be the case.

While Paragraph 2 of the Wells Fargo complaint discusses disparities in terms of relative differences, other discussion of differences in assignment to subprime status refer to differences in odds.  Thus, the points about odds ratios made in Section 6 of the main Lending Disparities and Section 3 of the United States v. Countrywide page regarding odds ratios issues generally apply to the Well Fargo complaint as well.[iii]  I note as well that the use of odds ratios suggests that the DOJ carried out regression analyses, which probably used more refined subcategories than used in Paragraph 2 (or so one would hope).  Such efforts at adjustment for characteristics, however, would remain subject to the general underadjustment issues raised on the Underadjustment Issues sub-page.