James P. Scanlan, Attorney at Law

Center for American Progress Study of Lending Disparities at TARP Banks

(Dec. 4, 2013; rev. May 22, 2014)

Prefatory note added April 10, 2018: This page discusses data from a Center for American Progress Study with respect to the failure to understand, among other things, lenders that reduce the proportions of loans that are high cost tend to show comparatively large relative racial differences between the proportion of black loans that are high cost and the proportion of white loans that are high cost.  It also discusses the way that the government’s failure to understand that issue may have contributed to the its singling out certain lenders for litigation, especially Bank of America and Wells Fargo from whom the government secured settlements totaling more than half a billion dollars in 2012.  The page has not been materially updated since it was originally created.   

The following are some recent publications (or other materials) pertinent to the issues addressed on the subpage.  My column “What the government gets wrong about fair lending,” American Banker (Apr. 9, 2018), discusses the government’s broader failure to understand that relaxing standards tends to increase, not reduce, relative racial differences in adverse borrower outcomes. It references my April 13, 2017 letter to the Department of Justice that provides some useful illustrations of the point (including credit score data from a putative class action against Wells Fargo).  But the column also discusses case brought by the City of Miami against Bank of America and Wells Fargo, and by the City Philadelphia against Wells Fargo, with regard to the failure to understand that actions lenders and the government take to reduce foreclosures will tend to increase the disproportionate concentration of foreclosures in minority neighborhoods on which the suits are based.  See the Foreclosure Disparities subpage.   

The Addendum to my “EEOC, OMB, and the Collection of Data That Can’t Be Analyzed,” Federalist Society Blog (Sept. 7, 2017) discusses the Miami cases against Bank of America and Wells Fargo with regard to the same failure of understanding discussed in the recent American Banker column.  The Federalist Society Blog post itself principally discusses the partial picture problem, pertinent to claims of both discrimination in compensation and discrimination in loan terms, addressed in the April 13, 2013 American Banker post discussed below.  “Partial Picture Issue Undermines Chadbourne Pay Equity Case,” Law360 (Jan. 25, 2017), discusses that issue with regard to the 2012 Bank of America and Wells Fargo settlements, as well as January 2017 $54 million settlement of claims against JP Morgan Chase Bank.  JP Morgan Chase Bank is another lender ranked at the top of a list based on the aforementioned ratio. 

Other pertinent materials created since this page was originally created include (1) “The Perverse Enforcement of Fair Lending Laws,” Mortgage Banking (May 2014); (2) Amicus curiae brief of James P. Scanlan in Texas Department of Housing and Community Development, et al. v.  The Inclusive Communities Project, Inc., Supreme Court No. 13-1731 (Nov. 17, 2014); and (3) Comments for Commission on Evidence-Based Policymaking (Nov. 14, 2016)

***

In various places, in discussing the pattern by which lenders that responded to federal encouragements or pressures to reduce adverse lending outcomes tended to increase relative racial and ethnic differences in rates at which their loan applicants experienced adverse outcomes (while reducing relative differences in the corresponding favorable outcome rates), I have discussed the matter by stating something to the effect that (a) the most responsive lenders tended to be regarded as the most discriminatory or (b) by responding to federal encouragements or pressures to reduce the frequency of adverse outcome lenders increased the chances that they would be deemed to have discriminated.[i]  While both formulations are correct, the latter is the more careful formulation.  For the comparative size of relative differences in adverse outcomes from lender to lender will be affected by a range of factors apart from the pattern whereby reducing the frequency of adverse outcomes tends to increase relative differences in adverse outcomes. 

Nevertheless, it can still be worthwhile to examine across a range of lenders the extent to which relative differences in adverse outcomes tend to be inversely correlated with (a) the frequency of adverse outcomes and (b) relative differences in favorable outcomes.

I do that to some extent in “When Statistics Lie” (Legal Times, Jan. 1 1996), which concerns denial/approval of mortgage applications. Similar examinations are possible with data made available in a September 2009 study by the Center for American Progress (CAP), “Unequal Opportunity Lenders? Analyzing Racial Disparities in Big Banks’ Higher-Priced Lending,” which addresses differences in proportions of loans to minorities and white that are higher cost loans at banks in the Troubled Asset Relief Program (TARP).  The CAP study has various analytical problems, including a mistaken interpretation of the implications of the large relative differences in adverse outcomes among high-income applicants (the subject of the Disparities – High Income subpage) and the failure to recognize the problematic nature of analyses that examine only persons who experienced an outcome (the subject of the Partial Picture Issues subpage, the Partial-picture problem section of “The Perverse Enforcement of Fair Lending Laws (Mortgage Banking, May 2014), “Fair Lending Studies Paint Incomplete Picture” (American Banker, April 24, 2013), and Section F of the University of Kansas Faculty Workshop paper, “The Mismeasure of Discrimination”). But the study does provide useful information on the rates at which borrowers of various races received high costs loans.

Such information suffers somewhat from the issues addressed on the Partial Pictures Issues subpage page,[ii] but not in a way that I think would materially affect the subject of this subpage.  The study generally breaks down the data into that pertaining to all applicants and that pertaining to high income applicants.  The latter is the more useful information, among other reasons, because the figures for all applicant are functions of the proportion high income applicants comprise of all applicants.[iii] 

Table 1 below is based on the information in the table at page 2 of the CAP study on the proportion of total loans to white and black borrowers that were high cost loans.  In Table 1 the lenders (which are limited to those that showed some number of high cost loans to both blacks and whites) are ordered according to the size, from largest to smallest, of relative differences between those proportions, which rankings are also reflected in the “(1)Rel Df HC Rank” column.  This is the way the federal government would appraise the comparative size of the differences.  The “(2) Rel Df NHC Rank” column shows the rankings according to the relative difference in rates of not receiving a high cost loan.  The “(3) White HC Rate Rank” shows the rankings according to the size of the proportion of white loans that were high cost.  The rankings according to the proportion of white loans that are high cost is a better indicator (than the proportion of total loans that are high cost) of the general tendency of the lenders to provide higher cost rather than regular loans, since the figure is uninfluenced by the differing proportions minorities comprise of all loan applicants at the various lenders.[iv]   But for completeness I include in the “(4) All HC Rate Rank” column rankings according to proportion of total loans that were high cost. 

Table 1.  Rankings of Lender According to (1) Relative Differences Between Proportions of Loans to Blacks and Whites That Were High Cost, (2) Relative Differences Between Proportions of Loans to Blacks and Whites That Were not High Cost, (3) Proportions of White Loans That Were High Cost, (4) Proportion of All Loans That Were High Cost – High Income Borrowers Only (ref B4801].

 

Lender	(1) Rel Df HC Rank	(2) Rel Df NHC Rank	(3) White HC Rate Rank	(4) All HC Rate Rank
JP Morgan Chase	1	4	5	4
Wells Fargo	2	6	9	7
SunTrust	3	10	10	10
Bank of America	4	5	8	9
Citigroup	5	1	3	2
GMAC/Ally Bank	6	7	7	6
PNC Financial Services	7	2	2	3
U.S. Bancorp	8	9	6	8
Capital One Financial Corp.	9	8	4	5
Regions Financial Corp.	10	3	1	1

  

Notably, the lenders ranking 2 through 4 were subjects of suits brought by the Department of Justice that ended in substantial settlements (Wells Fargo – $175 million, SunTrust – $21 million, and Bank of America ­–  $335 million).  Possibly rankings akin to those in the first data column of Table 1 underlay the decision of federal regulators to target those banks.  It should be recognized, however, that, as shown in the table on page 6 of the CAP study, Bank of America and Wells Fargo were numbers one and two in numbers of loans, which is also a factor that might affect their targeting.  (SunTrust, however, ranked number 7 in total loans.)  Further, it is impossible to know whether regulators would be more focused on the overall figures, such as set out in Table 2 below, than on the more refined (though, again, not very refined) figures for high income borrowers. 

The rankings according to relative differences in rates of avoiding a high cost loan show that Wells Fargo and Bank of America rank towards the middle of the group (though Bank of American’s ranking of the two measures are very close).  SunTrust, however, shows the smallest relative difference between rates of avoiding subprime status.

As suggested above, the rankings according to the proportion of white loans best indicated the comparative extent to which the lenders minimized the frequency of high cost loans, with the lower rankings’ reflecting the greatest extent to which the lender did so.  On this indicator, SunTrust, Wells Fargo, and Bank of America have the lowest rankings, suggesting that they made the greatest efforts to reduce the frequency of subprime status, with corresponding increase in relative differences in adverse lending outcomes that on which the suits against them would be based.  One gets a slightly different result on that ranking measure based on the proportion of total loans that were high cost. 

Table 2 shows the same information for loans among all income borrowers that Table 1 shows for high income borrowers.  Because of the lesser utility of overall figures, I limit my comments at this time to noting that with SunTrust, Wells Fargo, and Bank of America again showed the lowest proportion of white loans that were subprime (again with slightly different rankings for proportion of total loans that were high cost). 

Table 2.  Rankings of Lender According to (1) Relative Differences Between Proportions of Loans to Blacks and Whites That Were High Cost, (2) Relative Differences Between Proportions of Loans to Blacks and Whites That Were not High Cost, (3) Proportions of White Loans That Were High Cost, (4) Proportion of All Loans That Were High Cost – All Income Borrowers.

 

Lender	(1) Rel Df HC Rank	(2) Rel Df NHC Rank	(3) White HC Rate Rank	(4) All HC Rank
JP Morgan Chase	1	4	5	4
Bank of America	2	7	9	9
Wells Fargo	3	5	8	7
GMAC/Ally Bank	4	6	7	6
SunTrust	5	10	10	10
PNC Financial Services	6	3	3	3
Citigroup	7	2	2	2
U.S. Bancorp	8	9	6	8
Capital One Financial Corp.	9	8	4	5
Regions Financial Corp.	10	1	1	1