OFCCP, USDOL v. Bank of America (Board decision, April 21, 2016)

Read the official release (dol.gov)

Plain-English summary

The Labor Department's Office of Federal Contract Compliance Programs challenged Bank of America's hiring of African-American applicants for entry-level jobs at its Charlotte facility. An administrative law judge found intentional discrimination in 1993 and 2002–2005 and ordered back pay and other relief. The Board upheld the compliance review, the 1993 discrimination finding, and $964,033 in back pay for unsuccessful 1993 applicants, with additional interest. It reversed the finding and remedies for 2002–2005 because the evidence did not establish the same pattern of discrimination in that later period. Judge Brown concurred separately, and Judge Royce dissented from the reversal of the later-period finding.

Decision snapshot

• Cited authorities: Executive Order 11246; Rehabilitation Act, 29 U.S.C. § 793; veterans' employment law, 38 U.S.C. § 4212.
• Outcome: 1993 finding and $964,033 back pay award affirmed, with additional interest; 2002–2005 findings and remedies reversed.
• Key point: The Board treated the later hiring period separately and found its evidence insufficient to support that part of the ALJ's ruling.

Full text (DOL official public release)

U.S. Department of Labor
Administrative Review Board
200 Constitution Ave., N.W.
Washington, DC 20210

In the Matter of:

OFFICE OF FEDERAL CONTRACT
ARB CASE NO. 13-099
COMPLIANCE PROGRAMS, UNITED
STATES DEPARTMENT OF LABOR,
ALJ CASE NO. 1997-OFC-016

PLAINTIFF,
DATE: April 21, 2016
v.

BANK OF AMERICA,

DEFENDANT.

BEFORE: THE ADMINISTRATIVE REVIEW BOARD

Appearances:

For the Plaintiff:
M. Patricia Smith, Esq; Christopher Wilkinson, Esq.; Beverly I. Dankowitz, Esq.;
Consuela A. Pinto, Esq; and Jeffrey M. Lupardo, Esq.; United States Department of Labor, Washington, District of Columbia

For the Respondents:
Bruce M. Stern, Esq.; McGuireWoods LLP, Charlotte, North Carolina; W. Carter Younger, Esq.; McGuireWoods LLP, Richmond, Virginia; and Elena D. Marcuss, Esq.;
McGuireWoods LLP, Baltimore, Maryland

Before: E. Cooper Brown, Administrative Appeals Judge; Joanne Royce, Administrative Appeals Judge; and Luis A. Corchado, Administrative Appeals Judge

FINAL DECISION AND ORDER

USDOL/OALJ REPORTER
PAGE 1
 This matter arises under the affirmative action and nondiscrimination requirements of Executive Order 11246, as amended, 1 Section 503 of the Rehabilitation Act, 29 U.S.C.A. § 793,

and Section 402 of the Vietnam Era Veterans’ Readjustment Assistance Act, 38 U.S.C.A. § 4212 (Veterans’ Act) (collectively the Equal Opportunity Laws or EO Laws). 2 The Defendant Bank

of America (BOA or bank) appeals three Recommended Decisions and Orders that a Department of Labor Administrative Law Judge (ALJ) issued in the above captioned case. Specifically, BOA appeals the ALJ’s August 11, 2004 Recommended Decision and Order on Cross-Motions for Summary Judgment (Selection R. D. & O.), finding that the Department of Labor’s Office of Federal Contract Compliance Programs (OFCCP) improperly selected BOA’s Charlotte, North Carolina headquarters for a compliance review under the EO Laws. BOA also appeals the ALJ’s January 21, 2010 Recommended Decision and Order (Liability R. D. & O.) finding BOA liable under Executive Order 11246 for intentionally and unlawfully discriminating against African- American 3 (or black) job applicants in 1993 and 2002-2005. Finally, BOA appeals the ALJ’s

September 17, 2013 Recommended Decision and Order Awarding Damages (Remedy R. D. & O.). The ALJ awarded back pay, interest, and equitable relief. After fully considering the arguments of the parties and the record, we affirm the ALJ’s Selection R. D. & O. We affirm in part and reverse in part the ALJ’s Liability R. D. & O. and Remedy R. D. & O.

1
Executive Order 11246 (EO 11246), 30 Fed. Reg. 12,319 (Sept. 24, 1965), was amended by Executive Order 11375; 32 Fed. Reg. 14,303 (Oct. 13, 1967) (adding gender to list of protected characteristics), and Executive Order 12086, 43 Fed. Reg. 46,501 (Oct. 5, 1978) (consolidating enforcement function in the Department of Labor).
2
These provisions authorize the OFCCP to ensure that Federal contractors and subcontractors doing business with the Federal government comply with the laws and regulations requiring nondiscrimination and equal opportunity in employment, as implemented through 41 C.F.R. Part 60- 30 (pertaining to EO 11246), Part 60-741 (pertaining to the Rehabilitation Act), and Part 60-300 (pertaining to the Veterans’ Act) (2015).
3
For the lack of better terms, in discussing the relevant racial groups in this appeal, we feel compelled to use the admittedly inexact terms the Office of Management and Budget (OMB) used in the Standards for Maintaining, Collecting, and Presenting Federal Data on Race and Ethnicity (1997). See OFCCP’s Federal Contract Compliance Manual (FCCM), at 288 and 326 (July 2013)(Key words and phrases). Consequently, we use the following terms: (1) “African Americans” or “black” in discussing persons, not of Hispanic origin, with origins in any of the black racial groups of Africa; and (1) “white” as persons having origins in any of the original peoples of Europe, the Middle East, or North Africa. We understand that, to the extent possible, “Hispanic” applicants were not included in any of the expert analyses.

USDOL/OALJ REPORTER
PAGE 2
 INTRODUCTION

The twenty-year dispute in this matter began as a 1994 desk audit of BOA’s EEO compliance during 1993, wending its way through the courts and the Administrative Review Board (ARB) for years, and evolving into a 2004 compliance matter anchored to 1993. In 1993, the respondent was NationsBank and it had two recruiters for the two job groups (5A2 and 5F2) in question during that time period. By 2004, NationsBank had merged into its current corporate status as Bank of America and the job group in question at that time (5A) involved fifty-eight recruiters and hundreds of hiring managers. 4 After a hearing in 2008 and 2009, the ALJ found

that BOA engaged in unlawful disparate treatment in 1993 and from 2002-2005. The ALJ ordered various types of relief, including approximately $2,000,000 in back pay and requiring BOA to make ten offers of employment to African Americans. As we explain below, we agree with the ALJ’s conclusions that BOA intentionally discriminated against African Americans in 1993. We also find that the ALJ ordered a reasonable remedy for the unsuccessful 1993 African- American applicants. As to 2002-2005, while the record shows BOA poorly performed in hiring African Americans, we cannot say that the OFCCP proved BOA is liable for the damages awarded for the years 2002-2005. As a result, we unanimously affirm the ALJ’s orders pertaining to 1993, and two members of the Board reverse the ALJ’s liability and remedy orders pertaining to 2002-2005 for different reasons.

BACKGROUND 5

The Parties and Summary of 20-year Procedural History 6

BOA is a depository of government funds and an issuing and paying agent for United States savings bonds and, therefore, is a government contractor. 7 The OFCCP periodically

conducts compliance reviews to determine whether covered government contractors are

4
NationsBank merged with BOA after the OFCCP filed the complaint in this case and has been known thereafter as BOA. Therefore, this decision shall hereafter refer to the bank as BOA.
5
The Background is derived from stipulated facts, the ALJ’s findings, undisputed facts, and reasonable inferences arising from those facts.
6
The factual background regarding the OFCCP’s selection of BOA’s Charlotte, North Carolina headquarters for a compliance review is derived from the ALJ’s “Statement of Uncontested Facts” found in the ALJ’s Selection R. D. & O. Selection R. D. & O. at 3-5.
7
Selection R. D. & O. at 3, n.5; see also Oct. 10, 2003 Joint Stipulations of Fact at 1 ¶ 1 (noting that BOA admits that it is a covered government contractor); OFCCP v. Bank of America, ARB No. 00-079, ALJ No. 1997-OFC-016, slip op. at 16, n.12 (ARB Mar. 31, 2003).

USDOL/OALJ REPORTER
PAGE 3
 complying with the affirmative action and nondiscrimination requirements of the EO Laws and their implementing regulations. See 41 C.F.R. Part 60-1 (2015).

On November 24, 1993, the OFCCP’s Regional Director notified BOA by letter that OFCCP had selected BOA’s Charlotte, North Carolina headquarters for a compliance review under the EO Laws and requested that it submit documentation about its affirmative action program and other documentation for review. 8 BOA responded by April 1994 and without

objection (1) provided the requested documents to the OFCCP and (2) permitted the OFCCP to conduct an onsite investigation at BOA’s Charlotte headquarters. 9 The OFCCP conducted an

onsite investigation at BOA’s Charlotte headquarters, where it reviewed hiring records for job positions in job groups located “at the Bank’s headquarters and other [BOA] Charlotte locations and banking facilities.” 10

In October 1994, the OFCCP notified BOA that it found that BOA violated the EO Laws by discriminating against minority applicants for hiring in entry level positions in 1993. 11 The

OFCCP sent a revised notice of violation in June 1995. 12 On July 18, 1997, the OFCCP filed the

complaint in the matter before us, requesting that BOA comply with EO 11246 or be debarred. 13

Extensive judicial and administrative litigation followed the OFCCP’s 1997 complaint.
BOA obtained (1) a federal court injunction 14 against the OFCCP that was reversed and vacated

by the Fourth Circuit Court of Appeals 15 and (2) a summary decision 16 by an ALJ in this matter

that the ARB reversed. 17 The OFCCP subsequently obtained a summary decision that rejected

8
Selection R. D. & O. at 4-5; Joint Exhibit (JX) 2.
9
Selection R. D. & O. at 5.
10
See Liability R. D. & O. at 5; Administrative Law Judge Exhibit (ALJX) 3 (Stipulation of Facts) at 3 ¶ 13.
11
Selection R. D. & O. at 2; Bank of America, ARB No. 00-079, slip op. at 2; JX 4.
12
JX 10.
13
Selection R. D. & O. at 2; Bank of America, ARB No. 00-079, slip op. at 2.
14
NationsBank Corp. v. Herman, No. 3:95CV103-MU (W.D.N.C., Nov. 17, 1997).
15
NationsBank Corp. v. Herman, 174 F.3d 424 (4th Cir. 1999), cert. denied, Bank of America Corp. v. Herman, 528 U.S. 1045 (1999).
16
OFCCP v. Bank of America, ALJ No. 1997-OFC-016 (Aug. 25, 2000).
17
Bank of America, ARB No. 00-079, slip op. at 19-20.

USDOL/OALJ REPORTER
PAGE 4
 BOA’s claim of a Fourth Amendment violation on the evidence BOA presented. 18 BOA also

filed unsuccessful interlocutory appeals with the ARB. This litigation caused an eight-year gap in the review of BOA’s hiring practices, resulting in the background below focused on 1993 and 2002-2005. 19

BOA’s Hiring Process in 1993

For the hiring decisions made in this case in 1993, BOA relied on two recruiters for positions in both the 5A2 and 5F2 job groups: Teresa Simmons and Donna Craddock. Liability D. & O. at 5-6; ALJX 3 (Stipulation of Facts) at 3 ¶ 13. The 5A2 group consisted of bank tellers for prime time, full-time and part-time positions. The 5F2 group consisted of several clerical and administrative positions, including data entry operators, account clerks and remittance processing specialists. ALJX 3 (Stipulation of Facts) at 3 ¶¶ 11, 12.

Upon receiving a manager’s requisition to fill a job vacancy, the two recruiters would advertise the job vacancy and position. When applications were received, assistants for the recruiters required the applicants to complete an “EEO” tear-off form to identify their race and gender. It is unclear whether the applications were always forwarded to the recruiters without the “EEO” forms for their review. After reviewing the applications, the recruiters identified suitable candidates, performed initial interviews, and administered any required assessment tests.
If the recruiters determined that an applicant was qualified and a good fit for the job position, the applicant was scheduled for an interview with the hiring manager, who would decide whether to offer a job to the applicant. If the manager made a job offer, the applicant then was required to undergo a drug test. Liability R. D. & O. at 5; ALJX 3 (Stipulation of Facts) at 3-4 ¶¶ 14-19.

Whenever an applicant was disqualified or rejected for a job position, BOA used many disposition codes to record the reason for the disqualification or rejection. There were two disqualifying codes that fell substantially more harshly on African-American applicants than white applicants: the RC code (disqualified because of credit checks) and the RH code (disqualified for incompatible hours). Liability R. D. & O. at 5; ALJX 3 (Stipulation of Facts) at 4 ¶¶ 20-21.

BOA’s credit check process proved to be too subjective to provide useful criteria for valid statistical analyses. There were some factors considered in the credit check process, but there was no credible evidence that BOA applied any consistent minimum standards. 20 It was

18
OFCCP v. Bank of America, ALJ No. 1997-OFC-016 (Aug. 11, 2004).
19
Selection R. D. & O. at 2, 5; Bank of America, ARB No. 00-079, slip op. at 2.
20
BOA listed six independent categories of disqualifying factors used in the credit check process. According to those factors, BOA could disqualify applicants based on a single past due account, one collection effort, a judgment, or even where the account reflected just one lien of any

USDOL/OALJ REPORTER
PAGE 5
 unclear when the credit check was performed. Through the credit check process, BOA inexplicably disqualified many more African-American applicants than white applicants at nearly a two to one ratio. 21 BOA stopped using credit reports in April 1994. See JX 8 (BOA

Mar. 30, 1995 letter) at 4.

Similarly, through the use of the RH code for incompatibility between the work schedule the applicant allegedly requested and the schedule that the job requisition required, BOA disqualified many more African-American applicants than white applicants at a rate of greater than four to one. 22 Like the RC code, this code was not sufficiently objective to meaningfully

incorporate it into standard deviation analyses in this case. 23

For the 5A2 job group in 1993, Dr. Crawford found the relevant applicant pool to consist 24 of 923 applicants (583 white/340 African American) and 92 of these were offered jobs (74

kind. See JX 8. Ultimately, the ALJ did not credit BOA’s testimony that these factors were used consistently (Liability R. D. & O. at 48-51) and substantial evidence supports the ALJ’s finding where testimony suggested unrestricted subjectivity. See Hearing Transcript (HT) at 292 (Simmons testimony that BOA did not score credit checks and applicants could discuss credit check with BOA if they did not pass); Defendant’s Exhibit (DX) 123 (Craddock deposition testimony) at 28 (she did not use credit scores), 29 (problems with credit did not automatically disqualify applicant and applicants could discuss and explain credit check with BOA), 38 (no specific credit score needed to pass), 39 (no guidelines existed that explained how to overcome negative credit score), 42 (no written guidelines or training on how to apply credit check and other disposition codes).
21
See Liability R. D. & O. at 52; Plaintiff’s Exhibit (PX) 15 (Dr. Crawford’s Jan. 2009 report) ¶ 26 (noting that the RC code was assigned to 11.5% of the African-American applicants and 6.6% of the white applicants, a difference of 4.1 standard deviations). We also note that, according to a March 30, 1995 letter, JX 8 - Appendix B, BOA rejected 243 applicants under the RC code when combining the 5A1, 5A2, 5F2, and 5C1 job groups. BOA rejected twice as many African Americans as whites (158 to 77).
22
See Liability R. D. & O. at 52; PX 15 (Dr. Crawford’s Jan. 2009 report) ¶ 27 (noting that the RH code was assigned to 29.3% of the African-American applicants and 6.4% of the white applicants, a difference of 14.7 standard deviations).
23
The OFCCP did not pursue a disparate impact claim despite the evidence it presented of the stark impact that the credit check procedure and the RH code had on African Americans. BOA suggested that studies showed that credit checks in general tended to negatively impact African Americans, apparently suggesting that African Americans as a whole tend to have weaker credit history than white applicants, but we see no concrete data on this point in the record, much less data pertaining to the Charlotte area. BOA also suggested an “hours” requirement might negatively affect African Americans more but the basis for this assertion was utterly vague and quite baffling.
24
BOA’s expert Dr. Haworth’s total (929) differed by only six. See DX 114 at 2, Table 2.

USDOL/OALJ REPORTER
PAGE 6
 whites and 16 African Americans). PX 15, Table 1a. 25 For the 5F2 job group in 1993, Dr.

Crawford found that the relevant applicant pool consisted of 1,387 26 individuals (465 white/922

African American) and 174 of these were offered jobs (92 whites and 82 African Americans).
Id. A total of 655 were disqualified for credit and hours incompatibility. 27 Compare PX 15 (Dr.

Crawford’s Jan. 2009 report), Table 1a and PX 9 (Dr. Haworth’s Aug. 2008 report), Table 1.

From a statistical perspective, the rate at which BOA called African-American applicants for interviews and offered them jobs occurred at a rate significantly lower than the rate for white applicants (exclusive of Hispanic applicants). Using the total number of white applicants and the total number of African-American applicants for the combined 5A2 and 5F2 groups, Dr.
Crawford determined the standard deviation for the African-American applicant hiring rate was 6.9. Id.

BOA’s Hiring Process in 2002-2005

BOA’s hiring process had substantially changed by 2002. BOA accepted applications and resumes at any time and stored them. Whenever BOA hiring managers made job position requisitions, BOA recruiters would search the applications on file and try to match applications to the job position requisitions. See Liability R. D. & O. at 5-6; PX 10. There were now 58 recruiters for the relevant job positions at issue. See HT at 323. Also, BOA no longer conducted credit checks and drug tests. Liability R. D. & O. at 6, 50.

The 1993 5A2 and 5F2 job groups had become part of the 5A job group by 2002. See Liability R. D. & O. at 5; HT at 17-18. The OFCCP ultimately only alleged discrimination during the 2002-2005 period in the 5A job group, which contained jobs similar to those at issue in 1993 plus additional job positions. See HT at 18, 582.

For the 2002-2005 time period, Dr. Crawford found the relevant applicant pool to consist of 731 total individuals who applied for a job position in the 5A job group at issue (465 whites/266 African Americans). PX 15, Table 3. 28 BOA made 474 job offers to this group (327

25
Drs. Crawford and Haworth agreed on the number of offers BOA made to African Americans (16), and only differed by one on the offers made to whites (Crawford 74; Haworth 73). Compare PX 15, Table 1a (Crawford) with DX 114 at 2, Table 2 (Haworth).
26
Dr. Haworth’s total (1,391) differed by only four. See DX 114 at 2, Table 2.
27
The number 655 is inferred from the difference between Dr. Crawford’s total of 2,310 applicants that included applicants disqualified with an RC or RH code and Dr. Haworth’s total of 1,655 applicants that excluded applicants disqualified with an RC or RH code.
28
According to the statistics regarding the relevant BOA job applicants and hires during the 2002-2005 time period contained in BOA’s expert Dr. Haworth’s December 2008 Report, 744 applicants applied for a job position in the 5A job group at issue (505 whites/239 African

USDOL/OALJ REPORTER
PAGE 7
 white/147 African-American applicants). Id. Dr. Crawford determined the standard deviation for the African-American applicant hiring rate for all years (2002 through 2005) combined was 4.0. Id. 29 In terms of raw whole numbers (rounding fractions up or down), Dr. Crawford

determined that BOA should have hired approximately 9 more African Americans in 2002 (43 instead of 34), 4 more in 2003 (48 instead of 44), 10 more in 2004 (54 instead of 44), and 3 more in 2005 (35 instead of 32). Id.

ALJ’s Liability Finding

Based on the record evidence, the ALJ concluded that BOA engaged in a pattern or practice of intentional discrimination against African Americans in 1993 and from 2002-2005.
Specifically, the ALJ was persuaded that there were significant disparities from the job application to job offer, and during two intermediate phases of the selection process: application to interview and from interview to offer. The ALJ was persuaded that BOA did not use the RC and RH disqualifying disposition codes during 1993 in a uniform manner and, in fact, used these codes significantly more against African Americans than white applicants. The ALJ ultimately credited as reasonable the OFCCP’s expert Dr. Crawford’s opinion finding statistically significant shortfalls between the actual and expected number of African Americans hired during both the 1993 and 2002-2005 time periods, resulting in standard deviations for the African- American applicant hiring rates of 6.9 and 4.0, respectively. The ALJ further found that anecdotal evidence from three unsuccessful African-American applicants supported Dr.
Crawford’s statistical analysis. Moreover, the ALJ concluded that BOA hiring records covering the time period between 1994 and 2001, that BOA failed to retain or provide to the OFCCP as the regulations required, may be presumed to be unfavorable to BOA. In contrast, the ALJ found unpersuasive BOA’s expert Dr. Haworth’s opinion finding no discrimination. Consequently, the ALJ found that a preponderance of the evidence established that BOA engaged in a pattern or practice of intentional discrimination against African Americans during 1993 and the 2002-2005 time periods in violation of EO 11246.

The ALJ’s Award of Damages

The ALJ awarded damages for the intentional discrimination in 1993 and from 2002- 2005. After reviewing and weighing the opinions of the OFCCP’s expert Dr. Crawford and BOA’s expert Dr. Johnson, the ALJ credited as reasonable Dr. Crawford’s methods for calculating the damages in this case and rejected Dr. Johnson’s methods as unreasonable. After

Americans). See DX 114 at 14, Table 3. This difference of 27 African-American applicants is significant, given that Dr. Crawford concluded that BOA should have made 24.8 more offers during the 2002-2005 time period to African-American applicants, whereas Dr. Haworth concluded that BOA should have made only 8.6 more offers.
29
In contrast, while Dr. Haworth also combined all the years in his analysis, she separated them into two job groups (tellers and entry level). See DX 114 at 14, Table 3.

USDOL/OALJ REPORTER
PAGE 8
 finding that 50 additional offers should have been made to African Americans in 1993 in the 5A2 and 5F2 job groups, Dr. Crawford considered various factors to determine the average lost earnings and benefits of the unsuccessful applicants for each earnings year from 1993 through 2012. After adding the total lost earnings and benefits, plus Internal Revenue Service (IRS) interest rate, for each earnings year from 1993 through 2012, Dr. Crawford found that the total lost earnings and benefits plus IRS interest for the two 1993 job groups resulted in $964,033 in back pay damages. Similarly, Dr. Crawford calculated a total of $1,217,560 in back pay damages for the 2002-2005 group of unsuccessful African-American applicants. The ALJ also ordered BOA to make offers of employment to ten of the persons in the affected 1993 and 2002- 2005 groups of unsuccessful African-American applicants to be made whole through more than back pay damages alone, which is not challenged on appeal.

JURISDICTION AND STANDARD OF REVIEW

The ARB has jurisdiction to review exceptions to an ALJ’s Recommended Decision and Order (R. D. & O.) and is charged with authority to issue the Department’s final decision in cases arising under EO 11246, the Rehabilitation Act, and the Veterans’ Act. 30 Because no

standard of review exists in EO 11246, the implementing regulations, or Secretary’s delegation of authority, we rely on the Administrative Procedure Act. 31 Under the Administrative

Procedure Act, we have previously determined that our review is de novo and that the standard of proof in administrative adjudications “is the traditional preponderance-of-the-evidence standard.” 32 Even under a de novo review, nothing prohibits us from accepting as our own the

ALJ’s material fact findings that led up to the ALJ’s ultimate finding of fact (i.e., intentional discrimination) if those findings are supported by substantial evidence. 33 In Bobreski v. J. Givoo

Consultants (Bobreski II), we defined substantial evidence as evidence in the record that

30
Secretary’s Order No. 02-2012 (Delegation of Authority and Assignment of Responsibility to the Administrative Review Board), 77 Fed. Reg. 69,378 (Nov. 16, 2012); see 41 C.F.R. §§ 60-30.28, 60-30.30, 60-300.65(b)(1), 60-741.65(b)(1).
31
OFCCP v. Bank of America, ARB No. 07-090, ALJ No. 2006-OFC-003, slip op. at 7 (ARB Sept. 30, 2009).
32
Steadman v. SEC, 450 U.S. 91, 102 (1981) (construing the provision at 5 U.S.C. § 556(d) (1990) that provides, “[e]xcept as otherwise provided by statute, the proponent of a rule or order has the burden of proof”); Director, OWCP v. Greenwich Collieries, 512 U.S. 267, 277 (1994) (reaffirming Steadman).
33
For example, in OFCCP v. Greenwood Mills, Inc., ARB Nos. 00-044, 01-089; ALJ No.
1989-OFC-039, slip op. at 5 (ARB Dec. 20, 2002), while the Secretary explained that review of these cases is not restricted to a substantial evidence review, he did not say we are prohibited from accepting the ALJ’s supported findings of material fact.

USDOL/OALJ REPORTER
PAGE 9
 logically supports each of the material findings of fact and the record as a whole does not overwhelm the particular finding or expose the fact finding as genuinely unresolved. 34 Given

the extensive hearing presentation before the ALJ and the ALJ’s firsthand observations, we accept the ALJ’s predicate fact findings supported by substantial evidence. We will review de novo the ALJ’s ultimate finding of discrimination and her legal conclusions.

DISCUSSION

A. Compliance Review Selection

Before addressing the merits of the ALJ’s findings, we summarily resolve BOA’s objections to the OFCCP’s selection of BOA for a compliance review in this matter. In essence, BOA suggests that the notice of violation should be discarded because the OFCCP conducted an illegal search of BOA’s records. We find that the ALJ’s summary judgment ruling combined with BOA’s failure to present additional evidence at the merits hearing settled this issue. The Board rejected BOA’s claim that the OFCCP’s scheduling letter constituted coercion and, therefore, that the OFCCP allegedly never consented to the compliance review performed in this case. Specifically, the ALJ noted that BOA relied strictly on the language of the scheduling letter on remand in support of its contention that it did not voluntarily consent to the OFCCP’s compliance review in accordance with Fourth Amendment requirements, but did not proffer or rely on any other evidence regarding any other factors, such as communications between OFCCP and BOA employees, to support its contention. See Bank of America, ALJ No. 1997-OFC-016, slip op. at 5, n.9, 15, 16, n.21 (Aug. 11, 2004). Consequently, the ALJ held that she need not consider the issue of whether the OFCCP selected BOA’s Charlotte headquarters for a compliance review based on a neutral administrative plan. Bank of America, ALJ No. 1997- OFC-016 (Aug. 11, 2004). We see no reason to revisit that ruling.

Having relied solely on the scheduling letter and legal argument to raise a Fourth amendment challenge, it was incumbent on BOA to present more evidence to the ALJ on remand, a point the ALJ made clear in her summary decision on the asserted Fourth Amendment violation claim. BOA did not present any other evidence before the ALJ at the merits hearing.
Consequently, we find that the previous Board ruling is dispositive on the Fourth Amendment issue and reject BOA’s Fourth Amendment arguments.

B. BOA’s Liability for Intentional Discrimination/Pattern and Practice of Disparate Treatment

Turning to the merits, we begin with the overarching question of whether the ALJ erred in finding that BOA engaged in a pattern or practice of intentional discrimination in 1993 and

34
Bobreski v. J. Givoo Consultants, ARB No. 13-001, ALJ No. 2008-ERA-003, slip op. at 13- 14 (ARB Aug. 29, 2014) (Bobreski II).

USDOL/OALJ REPORTER
PAGE 10
 2002-2005. BOA argues that (1) the OFCCP was required to identify and prove that particular BOA individuals deliberately discriminated against African Americans because BOA only acts through individuals, (2) statistical evidence alone is insufficient to establish intentional discrimination, and (3) the OFCCP allegedly defeated its own disparate treatment claim by “disavowing any intent or ability to prove intent to discriminate . . . .” BOA’s Exceptions Brief at 27. BOA also argues that the ALJ improperly shifted the burden of proof and failed to consider all the evidence, as a whole, before deciding the ultimate question of causation. BOA’s Exceptions Brief at 29.

The OFCCP counters that it established a prima facie case of a pattern or practice of intentional discrimination through its statistical evidence indicating statistically significant standard deviations. It argues that the “burden” shifted to BOA to “demonstrate that the plaintiff’s statistical methodology lacks integrity or that the statistical disparities can be explained by nondiscriminatory factors.” OFCCP’s Response Brief at 46. But, according to the OFCCP, the Bank failed to show that the OFCCP committed any statistical error that would change the “result.” OFCCP’s Response Brief at 46. It argues that the BOA’s evidence was improper rebuttal evidence and, therefore, insufficient to disprove the OFCCP’s prima facie case.

The quality and quantity of evidence supporting the ALJ’s finding of intentional discrimination in 1993 fundamentally differs from the evidence supporting the ALJ’s finding of discrimination for 2002-2005. With respect to 1993, the ALJ relied on a variety of evidence, including (1) multiple statistical analyses showing significant standard deviations occurred at various stages of the hiring process and when employment offers were made, (2) the lack of standards for some decision-making processes (e.g., the RC Code), (3) anecdotal evidence of arbitrary treatment, and (4) troubling and unexplained disparate use of the RH code. As for 2002-2005, the ALJ relied on only the statistical disparity of that four-year period as a whole.
For the following reasons we provide below, we affirm the ALJ’s findings as to 1993 but reverse as to 2002-2005. Given the parties’ fundamental disagreement about the law governing pattern or practice claims of discrimination under EO 11246, we begin with an analysis of the burdens and standards of proof required to establish a pattern or practice of intentional discrimination.

1. The Law

Before discussing the relevant law, we must first point out that the OFCCP unequivocally chose to pursue only a claim of intentional disparate treatment. As confirmed during oral argument, the OFCCP did not pursue a disparate impact claim or a claim that BOA violated its affirmative action obligations under the EO Laws. Accordingly, in addition to relevant provisions of EO 11246, its implementing regulations, and Department precedent, we also look to federal appellate court decisions addressing similar pattern or practice claims of intentional discrimination adjudicated under Title VII of the Civil Rights Act of 1964. 35

35
42 U.S.C.A. § 2000e et seq. (Thomson Reuters 2012).

USDOL/OALJ REPORTER
PAGE 11
 To prove that BOA violated EO 11246 by engaging in a pattern or practice of intentional discrimination, the OFCCP must prove that unlawful discrimination was BOA’s regular procedure or policy. 36 Both parties agree with this point. See OFCCP’s Response Brief at 24;

BOA’s Exceptions Brief at 43. In a pattern or practice claim of intentional race discrimination, the OFCCP must show that there was a sufficient disparity and prove that race was a cause. 37

Often in such cases, the complainant or plaintiff employees point to a substantial disparity in selection rates for a particular job as proof of an unlawful bias against members of a disadvantaged or protected class, an allegation that must be proven by a preponderance of the evidence. Palmer, 815 F.2d at 90. The burden of proving pattern or practice discrimination remains at all times with the plaintiff. Segar v. Smith, 738 F.2d 1249, 1268-69, 1287 (D.C. Cir.
1984); Craik v. Minn. State Univ. Bd., 731 F.2d 465, 486-87 (8th Cir. 1984). Once the OFCCP proves a pattern or practice of discrimination against African-American applicants, each qualified African-American applicant benefits from a rebuttable presumption that he or she suffered from the same discrimination. Teamsters, 431 U.S. at 361-62. This rebuttable presumption shifts the burden of proof to the employer to demonstrate that it rejected the individual applicant for lawful reasons. Id. at 362.

In reviewing the ALJ’s ruling on the merits, we focus on the ultimate question of whether the OFCCP proved that BOA engaged in a pattern or practice of intentionally rejecting African- American applicants and that race was a factor. After a full evidentiary hearing, there is no need to engage in the burden of production analysis to determine whether the OFCCP presented a prima facie case or whether BOA presented legitimate, non-discriminatory reasons for its practices. 38 This burden of production analysis applies to motions for summary judgment and

motions for judgment as a matter of law. To decide the ultimate question of causation the ALJ must consider both the complainant’s and the respondent’s evidence. 39 The complainant’s

evidence may include a wide variety of circumstantial evidence, including motive, bias, work pressures from the employer, past and current relationships of the involved parties, animus,

36
International Bhd. of Teamsters v. United States, 431 U.S. 324, 360 (1977).
37
See, e.g., Palmer v. Schulz, 815 F.2d 84, 97 (D.C. Cir. 1987)(Plaintiff must show by a preponderance of the evidence that a substantial disparity resulted from discrimination.).
38
See, e.g., Reeves v. Sanderson Plumbing Prods., Inc., 530 U.S. 133, 153 (2000)(the “ultimate question in every employment discrimination case involving a claim of disparate treatment is whether the plaintiff was the victim of intentional discrimination.”); Joseph v. Publix Super Mkts., Inc., 151 Fed. Appx. 760, 764 (2005)(“in a disparate treatment case fully tried on the merits, our review . . . is limited to the question whether the plaintiff presented sufficient evidence to prove each element of his or her claim.”).
39
See, e.g., Bobreski II, ARB 13-001. See also Palmer, 815 F.2d at 97 (After both sides have submitted evidence, the question is “whether in light of the totality of the evidence, plaintiffs have shown that, more likely than not, the disparity resulted from discrimination.”).

USDOL/OALJ REPORTER
PAGE 12
 temporal proximity, pretext, shifting explanations, and material changes in employer practices, among other types of evidence. 40 3

We reject the OFCCP’s assertion that BOA failed, as a matter of law, to present proper rebuttal for us to consider on the question of intentional discrimination. The OFCCP argues that, after its presentation of evidence, BOA had the specific burden of showing that the OFCCP’s statistical proof was unsound or to prove that the disparity occurred as a result of legitimate, non- discriminatory reasons. But the burden of proof always remains with the OFCCP. BOA’s task was to present a rebuttal case through admissible evidence. Employment discrimination cases addressing motions for judgment as a matter of law may confuse parties into thinking that defendants’ rebuttal obligations in employment discrimination cases fundamentally differ from other civil tort litigation. But, just like non-employment cases, rebuttal evidence simply means evidence “given to explain, repel, counteract, or disprove facts given in evidence by the adverse party.” 41 In this case, rebuttal evidence includes any evidence that attacks the validity of the

opposing party’s proof of intentional discrimination or supports an alternative causation theory or does both. BOA may attack the OFCCP’s evidence of discrimination in many ways, including but not limited to: (1) cross-examining the OFCCP’s witnesses, (2) offering rebuttal evidence showing an opposing or different view of what occurred, or (3) arguing that the OFCCP’s evidence was legally or factually unpersuasive. 42

Here, BOA attacked the OFCCP’s statistical analysis through cross-examination, testimony from BOA employees, and by presenting its own expert testimony. BOA’s expert testified that the statistical disparities were not as significant as the OFCCP’s expert claimed and also testified that reasons other than race discrimination caused the hiring disparities between white and African-American applicants. This evidence created a genuinely disputed issue of fact for the ALJ, and she was persuaded by a preponderance of the evidence that BOA engaged in a pattern or practice of intentional discrimination against African Americans. See, e.g., Segar, 738 F.2d at 1264. As we explain below, we agree with the ALJ’s conclusions as to 1993 but not as to her finding of intentional pattern or practice discrimination as to 2002-2005.

While some disagreement continues in the courts about the role of statistical evidence in employment discrimination cases, a few principles seem fairly established. For example, statistical evidence may be used to rule out chance as a likely reason for a significant racial disparity. Courts have consistently found significance in disparities exceeding the two standard deviation mark. See Hazelwood School Dist. v. U.S., 433 U.S. 299, 308, n.14 (1977); Adams v.

40
Bobreski v. J. Givoo Consultants, ARB No. 09-057, ALJ No. 2008-ERA-003, slip op. at 13 (ARB June 24, 2011) (Bobreski I).
41
BLACK’S LAW DICTIONARY (Revised 4th ed. 1968).
42
Palmer, 815 F.2d at 99 (point to legitimate, non-discriminatory reasons, challenge the statistics as faulty or based on improper methodologies).

USDOL/OALJ REPORTER
PAGE 13
 Ameritech, 231 F.3d 414, 424 (7th Cir. 2000). Ruling out chance does not automatically mean race discrimination was a motivating factor, but it makes such a reason a viable factor that could be inferred. 43 The more severe the statistical disparity, the less additional evidence is needed to

prove that the reason was race discrimination. Very extreme cases of statistical disparity may permit the trier of fact to conclude intentional race discrimination occurred without needing additional evidence. 44 Ultimately, the OFCCP must present enough evidence to persuade the

ALJ that race discrimination was a motivating factor in BOA’s hiring decisions. What constitutes sufficient evidence must be evaluated on a case-by-case basis.

2. ALJ’s finding that intentional discrimination occurred in 1993

To begin with, we find that substantial evidence supports the ALJ’s material findings of fact that lead up to the ultimate finding of discrimination. For example, among other findings, we accept the ALJ’s findings as to how the hiring process worked, the lack of real standards for the RC and RH Codes, the reliability of the raw data Dr. Crawford used for disparity studies, and the average salaries of BOA workers in the 1993 5A2 and 5F2 job groups. Second, we find the ALJ reasonably relied on the OFCCP’s expert’s conclusions of substantial disparity between the rates at which BOA hired white applicants as opposed to African-American applicants, the statistical evidence, and anecdotal evidence. Dr. Crawford rationally explained why he disregarded certain disqualifying codes, the reason for the controls in his analysis, and he also reworked some of his findings to add more controls and show that the disparities remained significant. Like the ALJ, we find that Dr. Haworth’s conclusions were less credible than Dr.
Crawford’s because, among other reasons, (1) her disaggregation of the data was extreme and (2) she accepted as valid BOA’s RC and RH codes when they had no semblance of consistent objectivity.

With respect to 1993, we find that the multiple severe disparities, along with other evidence, provided a sufficient basis for the ALJ’s finding of intentional discrimination and similarly convinces us. The evidence demonstrates that there was a significant disparity in the hiring decisions (1) when looking at all applicants that were given a job offer and (2) when you divide the hiring process into two phases (application to interview/interview to job offer).
African Americans fell substantially below the hiring rate for whites even when the experts controlled for job group or timing of applications (from application to interview to hiring decision). There was simply no positive outcome or outcomes that could be dismissed as trivial no matter how one examined the hiring process. More specifically, while criticizing the

43
See, e.g., Watson v. Fort Worth Bank & Trust, 487 U.S. 977, 988, 994 (1988). But see ante, at 994 (proving a causal link requires statistical evidence of a “kind and degree” that connects the disparity to race discrimination).
44
Teamsters, 431 U.S. at 338-340 (involved extreme disparate statistics for African Americans). See also Tagatz v. Marquette Univ., 861 F.2d 1040, 1044 (7th Cir. 1988)(discussing the significance of statistics).

USDOL/OALJ REPORTER
PAGE 14
 OFCCP’s expert Dr. Crawford’s 1993 findings, BOA acknowledged that what it considered were Dr. Crawford’s more accurate results dipped only to 4.3, 4.5, and 2.1 substandard deviations. PX 15 at 5, ¶ 18(d), (e), (f); BOA’s Exceptions Brief at 28. The best that BOA’s expert could say in rebuttal about 1993 is that the results were not “statistically significant,” not that they were good results. BOA’s Exceptions Brief at 29 (admitting to one statistically significant result for tellers in the 2002-2005 time period). In effect, BOA admits that the statistical evidence is not good and, therefore, this case is not a case of a rosy version of the statistics competing with a troubling version; all the versions are troubling. BOA’s Exceptions Brief at 29. Even BOA’s expert’s best numbers rarely dipped below two standard deviations in his findings. While this is not significant, it shows that the best picture is approximately two standard deviations and gets worse from there.

Other evidence provides additional evidence that the 1993 disparities resulted from race discrimination. As further detailed below, there is unrefuted evidence that the RC code and RH codes eliminated a grossly disproportionate number of African-American applicants compared to white applicants. We saw no objective standards that BOA applied that could explain the gross disparity in BOA’s use of these codes. We cannot imagine a rational objective reason for a gross disparity in the use of the RH code. More troubling, there is evidence in the record suggesting the race of the applicants may have been known for some applicants at the time the RC and RH codes were applied. Applications provided substantial information that could have revealed race in many instances, and BOA was not always careful in separating the “race” information from the application as the applicants went through the hiring process. See, e.g., DX 109 at 8 (Dr.
Haworth acknowledges that it was “possible” that BOA recruiters could have seen the EEO tear- off sheet providing racial identification on BOA applications that applicants provided). We find that this evidence not only supports the ALJ’s findings of a pattern or practice of intentional discrimination in 1993 but also establishes the pattern or practice by a preponderance of the evidence under a de novo review.

3. Weight of Expert Conclusions versus. Burden of Proof

Notwithstanding our view of the evidence, we examine and reject BOA’s contention that the ALJ erred by placing the burden of proof on BOA when she examined BOA’s rebuttal evidence. BOA’s argument focuses on some ambiguous language the ALJ used in describing which expert persuaded her. Specifically, BOA focuses on the ALJ’s occasional statement that BOA had to prove that it used the RC and RH disqualifying codes in a “race neutral fashion.” But, at other times in explaining the same point, the ALJ and the OFCCP’s expert also said that the RC and RH disqualifiers were not used “uniformly” or in a “neutral fashion” or according to an objectively consistent standard. Liability R. D. & O. at 48. The ALJ expressly finds that the RH codes were applied “disproportionately” against African Americans and were not applied in a “uniform fashion.” Liability R. D. & O. at 51. She expressly accepted Dr. Crawford’s calculations showing the substantial disparate application of the RH code to African Americans and that there was no credible reason for this disparate result. Liability R. D. & O. at 52.
Consequently, Dr. Crawford concluded that this lack of uniformity or objective consistency eliminated the statistical justification for removing these candidates from the pool of applicants

USDOL/OALJ REPORTER
PAGE 15
 used for standard deviation calculations. The ALJ found that “there was no evidence of any criteria used by the recruiters in application of this disposition code [RC code].” Liability R. D.
& O. at 48.

We find that the context of the entire record clarifies that the ALJ did not switch the ultimate burden of proof but instead exercised her discretion to find the OFCCP expert Dr.
Crawford’s methodology more persuasive because his methodology ignored disqualifiers [or disqualifying codes] that had no defined uniformity. 45 Dr. Crawford, included 651 applicants

who were eliminated with an RC code or RH code, see PX 15 at 8-9, ¶ 25, and, conversely, BOA’s expert, Dr. Haworth, excluded these applicants from the relevant universe of applicants.
Dr. Crawford included these applicants in his calculations, because he found that they were not uniformly applied and could not be used as statistically valid reasons to exclude applicants from the standard deviation calculations. When Dr. Crawford said that the codes were not uniformly applied or race neutral, we understand this to mean that these codes were not statistically valid disqualifiers for standard deviation analyses. More importantly, Dr. Crawford opined that excluding applicants tagged with these inconsistent and vaguely used RC and RH codes would effectively approve of the very discrimination the OFCCP suspected was occurring. In other words, if these vaguely applied RC and RH codes were nothing more than code talk for “African American,” then eliminating these applicants from the statistical study would mask the discrimination. 46 Absent sufficient indication that these codes had some objectively uniform

application apart from race, Dr. Crawford could not be sure that the standard deviation properly reflected the selection process. We understand that the ALJ adopted Dr. Crawford’s rationale that these factors were statistically invalid, and BOA failed to convince the ALJ otherwise, leading the ALJ to reject BOA’s expert Dr. Haworth as unpersuasive. Consequently, rather than being an issue of the ultimate burden of proof, we find that the ALJ explained why Dr.
Crawford’s methodology was more persuasive. Having determined which standard deviation results to accept, the ALJ then resolved the ultimate question of whether the evidence as a whole proved a pattern and practice of intentional discrimination. See Liability R. D. & O. at 62.

Given the ALJ’s reasons for believing the OFCCP’s expert, we also reject BOA’s argument that the ALJ committed reversible error by finding intentional discrimination where the OFCCP allegedly failed to specifically identify which BOA employees engaged in intentional

45
We appreciate that the ALJ made some confusing stray comments, but we infer that such comments only emphasize how convinced the ALJ was that BOA had intentionally treated African Americans differently because of their race. For example, the ALJ said she would have been convinced by Dr. Crawford’s testimony alone and that such evidence would meet the OFCCP’s burden of proof. Liability R. D. & O. at 45, 62 and 64, n.41. We agree that Dr. Crawford’s analysis shows that intentional discrimination occurred at many points but most notably when BOA applied the RH code and secondarily the RC code.
46
Compare McReynolds v. Sodexho Marriott Servs., Inc., 349 F. Supp. 2d 1, 26-27 (D.D.C.
2004) (discussed the concept of “inappropriate,” “tainted,” or “nonsense” variables “masking a discriminatory purpose”).

USDOL/OALJ REPORTER
PAGE 16
 discrimination. In rejecting the use of the RC and RH codes, we infer that the ALJ effectively identified the use of these codes as a point in the hiring process where intentional discrimination occurred. We find this to be especially true of the RH code. There is simply no rational explanation in the record why African Americans’ work schedules should so disproportionately disqualify them from jobs at BOA. A very strong inference arises from the record that race was the motivating factor for applying the RH code so heavily against African Americans and that strong inference supports a finding of race discrimination in 1993, especially where BOA eliminated the RH code in 1994. Similarly, it is undisputed that the RC disqualifying code fell substantially more heavily on African Americans without any record evidence explaining this result. Again, this permits another strong inference of race discrimination in addition to the inference from the RH code. Without anything in the record permitting the ALJ to choose between race as a motivating factor and something else, we find that she had justifiable reasons for concluding that race was a motivating factor. In the end, the record permitted and supported the ALJ’s finding of race discrimination in the use of the RH and the RC codes, which were the major reasons leading to the OFCCP expert Dr. Crawford’s conclusions of significant disparity between the selection of whites as opposed to African Americans. We agree that the disparate use of these codes support a finding of intentional pattern or practice race discrimination in 1993 and it is unnecessary to specifically identify the individual culprits.

4. Period during 2002-2005

As a preliminary matter pertaining to the discrimination claims for the period 2002-2005, we find that we must analyze this time period as a follow-up but separate claim of pattern or practice race discrimination and not as a continuation of the same 1993 pattern or practice claim. 47 First, the OFCCP treated these time periods separately and not as one continuous

pattern or practice of intentional discrimination. The OFCCP expert’s statistical analysis for each of the periods substantially differs. More specifically, the OFCCP expert’s analysis for the 1993 period has many different standard deviation analyses and tests, while his 2002-2005 analyses rests on only two or three bottom line conclusions for the entire period. Second, fundamental changes occurred with BOA and its hiring process between 1993 and 2002. The Bank changed from NationsBank to BOA. The recruiting process dramatically changed, including the number of recruiters (from two to 58) and the discontinued use of credit checks and drug tests. Third, the evidence presented simply does not support the idea that the same pattern or practice of intentional discrimination applies to the hiring practices in 1993 and 2002-2005.
The ten-year gap of data and evidentiary information between these time periods prevents any realistic ability to logically connect those two periods.

47
Board Member Corchado finds that the record evidence fails to establish a separate pattern or practice claim for 2002-2005, but he takes no position as to whether the OFCCP was authorized to conduct a follow-up review for the period of 2002-2005 or the extent to which the OFCCP is authorized to conduct follow-up compliance reviews after finding a violation, especially in a case where BOA refused for many years to submit to a meaningful compliance review. Given Board Member Brown’s separate opinion, a majority of the Board rejects the ALJ’s liability ruling as to the 2002-2005 period.

USDOL/OALJ REPORTER
PAGE 17
 Viewing the record as a whole, the record evidence fails to support the ALJ’s finding of a pattern or practice of intentional discrimination during 2002 through 2005. First, unlike the 1993 period, there is no evidence that BOA used the RC or RH codes or other code to eliminate African-American applicants disproportionately. Second, contrary to the 1993 standard deviation analysis, the OFCCP’s evidence of discrimination in 2002-2005 boils down to one standard deviation of 4.0 (or 4.1) for the four-year period, but no standard deviation conclusions year by year. Also unlike the 1993 analyses, the OFCCP did not argue that there were statistically significant standard deviations for one or more stages of the hiring process. In terms of raw numbers for each year, two of the four years fail to provide powerful statistical evidence of intentional discrimination: in 2003, 44 African Americans were offered a job instead of the expected number of 47.9; in 2005, 32 instead of 34.5. These are small shortfalls. Without more evidence, one bottom line standard deviation of 4.0 for four years with minor shortfalls in two of those years is not enough in this particular case to prove a pattern or practice of intentional racial discrimination.

C. The ALJ’s Award of Damages

The ALJ awarded $964,033 for the intentional discrimination suffered by the 1993 unsuccessful African-American applicants in the 5A2 and 5F2 job groups. The award covers lost earnings/benefits and interest of approximately 50 unsuccessful African-American applicants during the period 1993 through 2012. The award for the 5A2 job group totals $259,387 ($174,500 in lost earnings/benefits plus $84,887 in interest) and the 5F2 job group totals $704,646 ($587,779 in lost earnings/benefits plus $116,867 in interest). To arrive at these awards, the OFCCP’s expert determined how many of the 50 unsuccessful African-American applicants would be working each year (approximately 50 in 1993 to less than 4 remaining applicants by 2012) and determining their expected earnings and benefits with BOA each year, plus interest. The OFCCP’s expert then used SSA earnings statements that the OFCCP provided from approximately 121 48 members of the 1993 unsuccessful African-American applicants and

48
We found no clear finding in the ALJ’s Remedy R. D. & O. as to how many individuals would be entitled to share the award for discriminatory hiring in 1993. The ALJ ultimately credited Dr. Crawford’s calculations of the size of the class of African-American applicants who were not offered any of the relevant entry level positions in 1993. Dr. Crawford determined that 1262 African Americans applied for the relevant entry level positions in 1993 and 98 were offered a position, leaving 1,164 applicants. See PX 15 at 38, Table 1a.

The Joint Stipulation Regarding Certain Remedy Phase Hearing Exhibits indicates that Drs.
Crawford and Johnson were provided “copies of 121 Social Security Statements containing Earnings Records” to calculate interim earnings. Dr. Johnson stated in his reports that the OFCCP provided him with SSA earnings statements from 121 of the total class of unsuccessful applicants to review, which represented ten percent of the overall class of unsuccessful applicants, see DX 126 at 2; DX 127 at 10 ¶ 19; 13 - Exh. 2; 19 ¶ 24; HT at 107, 112. Of the 121 SSA earnings statements from the OFCCP, Dr. Johnson states that 117 were from the 1993 job groups (34 were from the 1993 5A2 job

USDOL/OALJ REPORTER
PAGE 18
 subtracted their actual average earnings during the period 1993 through 2012, representing the actual interim or mitigating earnings they were paid in other employment. Dr. Crawford then added the totals of those years to reach the $964,033 amount. Because it was not possible to identify which 50 applicants would have been hired, the ALJ required that the 1993 award be distributed evenly to all of the unsuccessful African-American applicants.

1. Legal Standard for the Award of Remedies

BOA raises many exceptions to the ALJ’s damages award, and we address them below.
Logically, we must begin with BOA’s objection to the ALJ’s class-wide method of calculating the back pay damages rather than calculating based on a person-by-person, individual assessment. This objection obviously affects the bulk of the damages award.

The legal standards developed under Title VII of the Civil Rights Act of 1964 in employment discrimination cases apply to employment discrimination cases brought under EO 11246. See Greenwood Mills, Inc., ARB Nos. 00-044, 01-089, slip op. at 5; OFCCP v.
Cleveland Clinic Found., 1991-OFC-020, slip op. at 3 (ARB July 17, 1996); U.S. Dep’t of Labor v. Honeywell, Inc., 1977-OFC-003, 1993 WL 1506966, slip op. at 10 (Sec’y June 2, 1993)(Honeywell I). In this case, the ALJ found that BOA intentionally and unlawfully discriminated against African-American candidates in hiring for entry level positions in 1993 and in 2002-2005. Liability R. D. & O. at 64-65. After a finding of discrimination, a remedy is appropriate to restore the injured discriminatee, that is, “to make persons whole for injuries suffered on account of unlawful employment discrimination.” Albemarle Paper Co. v. Moody,

group and 83 from the 5F2 job group), while only 4 were from the 2002-2005 5A job group (including 3 from 2002 unsuccessful applicants, 4 from 2003 and 2004 unsuccessful applicants, and none from 2005 unsuccessful applicants). See DX 127 at 17 ¶ 23; at 19 ¶ 24 (noting that “results based on such small samples [for the 2002-2005 job group] could potentially be skewed” and “not adequately represent the entire group.”).

Dr. Crawford testified that the OFCCP provided him with SSA earnings statements of “roughly” 125 of the total class of unsuccessful applicants to review, which represented ten percent of the overall class of unsuccessful applicants, see JX 56 at 66 (“roughly” 125); HT at 40 (125 “sounds about right”). See also PX 21 at 7-8 (where Dr. Crawford notes “[t]he Social Security data consists of earnings records provided by the Social Security Administration (marked BOA-REM- 0000593 to BOA-REM-0000838), which show the years that the individual worked up to 2009 or 2010, the individual’s taxed Social Security earnings per year, and the individual’s taxed medicare earnings per year.”); JX 56 at 21, 66-67 (Dr. Crawford’s June 2012 Deposition where he states he reviewed “roughly” 125 SSA earnings statements indicating “the earnings of unsuccessful African American applicants as reflected in the Social Security data,” which he found to be “reliable.”). Dr.
Crawford states of the 121 SSA earnings statements from the OFCCP, only 5 were from the 2002- 2005 5A job group (including 4 from 2002 unsuccessful applicants, 5 from 2003 unsuccessful applicants, 4 from 2004 unsuccessful applicants, and none from 2005 unsuccessful applicants). PX 27 at 10 ¶ 24.

USDOL/OALJ REPORTER
PAGE 19
 422 U.S. 405, 418 (1975). Back pay is one element of the “make whole” relief that may be provided to a victim. 41 C.F.R. § 60-1.26(a)(2).

Rather than calculating damages based on individual assessments of the loss of each victim, a class-wide formula, such as the ALJ used in this case, may be used in calculating a back-pay award. See McClain v. Lufkin Indus., Inc., 519 F.3d 264, 280-281 (5th Cir. 2008);
Segar, 738 F. 2d at 1289-1291; Greenwood Mills, ARB Nos. 00-044, 01-089, slip op. at 5-6. If the case is complex, the class is large, or the illegal practices continued over an extended period of time, a class-wide approach to measure back pay may be necessary. McClain, 519 F.3d at 280-81; Pettway v. Am. Cast Iron Pipe Co., 494 F.2d 211, 261 (5th Cir.1974).

The ALJ “found that all three of these factors were present in this case.” Remedy R. D.
& O. at 3; see also Id. at 13, 15. Specifically, BOA submitted a request to the ALJ for the issuance of 1,147 subpoenas to each member of the class of African-American applicants who were not offered any of the relevant entry level positions in 1993 and in 2002-2005, for the production of their employment, earnings and tax records to assess the loss of each victim. Id. at 3. As there were at least 1,147 potential victims in this case, the ALJ determined that calculating damages based “on an individualized hearings approach” “was inappropriate, impractical, and infeasible for this case.” Id.

We agree with the ALJ that determining damages on a class basis was a reasonable course to follow in this case for the reasons she explained. There was no way to determine which 50 individuals of the 1,147 applicants would have been hired in the absence of discrimination. The inability to identify the actual victims occurred for many reasons, including the fact that there were missing records, ambiguous and highly subjective use of disqualifying codes, and because the liability hearing occurred more than a decade after the violations due to BOA’s extensive litigation efforts to end the OFCCP’s compliance review. Therefore, we hold that the ALJ did not abuse her discretion, nor clearly err, in adopting the class-wide formula- driven approach in this case. See McClain, 519 F.3d at 281. Simply, class based determination of damages was the right choice.

In fashioning a remedy for employment discrimination, “the court must, as nearly as possible, ‘recreate the conditions and relationships that would have been had there been no’ unlawful discrimination.” Teamsters, 431 U.S. at 372 (quoting Franks v. Bowman Transp. Co., 424 U.S. 747, 769 (1976)). Back pay remedies should serve the “make whole” purposes of Title VII without constituting a “windfall” for the victim of discrimination at the expense of the employer. Ingram v. Madison Square Garden Ctr., Inc., 709 F.2d 807, 812 (2d Cir. 1983).

But the “process of recreating the past will necessarily involve a degree of approximation and imprecision.” Teamsters, 431 U.S. at 372. The courts have provided three general rules to guide us in the determination of the appropriate back pay figure:

(1) [U]nrealistic exactitude is not required;

USDOL/OALJ REPORTER
PAGE 20
 (2) [A]mbiguities in what an employee or group of employees would have earned but for discrimination should be resolved against the discriminating employer; and

(3) [T]he [trier of fact] . . . must be granted wide discretion in resolving ambiguities.

Stewart v. General Motors Corp., 542 F.2d 445, 452 (7th Cir. 1976) (citations omitted).

While formula or class-wide relief may generate a windfall for some employees who would have never been hired even if the jobs had been filled on a nondiscriminatory basis and may undercompensate the genuine victims of discrimination by forcing them to share the award with some undeserving recipients, it is the best that can be done under the circumstances. Id.
“Any method [of calculating damages] is simply a process of conjectures.” Pettway, 494 F.2d at 260-261. Given a choice between no compensation for those who have been illegally denied jobs and an approximate measure of damages, courts have chosen the latter. See Stewart, 542 F.2d at 452.

2. Standard of Review of ALJ’s Award of Damages

Again, as the legal standards developed under Title VII apply to intentional race discrimination claims brought under EO 11246, see Greenwood Mills, ARB Nos. 00-044, 01- 089, slip op. at 5; Cleveland Clinic Found., 1991-OFC-020, slip op. at 3; Honeywell I, 1977- OFC-003, 1993 WL 1506966, slip op. at 10, we look to ARB precedent as well as federal appellate courts that have reviewed Title VII remedial orders. Like the federal appellate courts, we will adopt the ALJ’s methodology for awarding damages if she exercised reasonable discretion. 49 Thus, relying on the ALJ’s discretion makes sense given the complexity of

determining back pay compensation. 50

3. Experts’ Opinions and ALJ’s Findings

Dr. Crawford, on behalf of the OFCCP, and Dr. Johnson, on behalf of BOA, used the adjusted shortfalls Dr. Crawford calculated between the expected number of African-American

49
See U.S. v. Brennan, 650 F.3d 65, 138 (2d Cir. 2011)(quoting Ass’n Against Discrimination in Emp’t, Inc. v. City of Bridgeport, 647 F.2d 256, 279 (2d Cir.1981)(“Our function is not to exercise our own discretion, but to determine . . . whether the [] judge has abused his”)). See also U.S. v. Lee Way Motor Freight, Inc., 625 F.2d 918, 932 (10th Cir. 1979) (noting that the courts of appeal are in general agreement that the formula of computation . . . [is] a matter[] within the discretion of the [lower court judge] (citing Kaplan v. Int’l All. of Theatrical and State Emps., 525 F.2d 1354, 1362-63 (9th Cir. 1975) and Thornton v. E. Tex. Motor Freight, 497 F.2d 416, 421 (6th Cir. 1974)).
50
See Equal Emp’t Opportunity Comm’n v. Enter. Ass’n, Steamfitters, Local No. 638, 542 F.2d 579 (2d Cir. 1976), Hairston v. McLean Trucking Co., 520 F.2d 226 (4th Cir. 1975).

USDOL/OALJ REPORTER
PAGE 21
 applicant hires and the actual number of African-American applicant hires that the ALJ credited in her decision on liability for both the 1993 and 2002-2005 group of applicants to calculate the amounts of lost earnings and benefits by using the average earnings for hires in the relevant job group and year (the 1993 and 2002-2005 group of hires or “cohorts”). Remedy R. D. & O. at 4, 7. For the 1993 cohort, Dr. Crawford calculated an adjusted shortfall of 49.9 of the expected number of African-American applicant hires and the actual number of African-American applicant hires. See PX 22 at 7 ¶ 20; PX 27 at 32, 34 (Corrected Tables 3-5A2-M1 and 3-5F2- M1); DX 126 at 2 ¶ 1; DX 127 at 6 ¶ 14.

As noted, the ALJ adopted the class-wide formula, using the shortfall method, to determine the damages in this case. Remedy R. D. & O. at 4. The ALJ stated that the class-wide formula includes “the shortfall between the discriminated and non-discriminated job applicants; the average earnings for hires in the relevant job group and year (identified . . . as “cohorts”); the average employment period for those hired applicants in the relevant cohort (identified . . . as “tenure” [or survival rate]; interim and/or mitigating earnings of the discriminated class members; and the prejudgment interest rate to apply on any back pay awards.” Remedy R. D. & O. at 4-5 (emphasis added).

4. The ALJ credited Dr. Crawford’s calculations of damages as reasonable

After reviewing and weighing the opinions of Dr. Crawford and Dr. Johnson, the ALJ credited Dr. Crawford’s methods for calculating the damages in this case. Thus, we review the ALJ’s determination to credit Dr. Crawford’s calculations based on his choices to A) control for job group; B) supplement missing W-2 earnings information from the hired applicants with additional data and estimates; C) use the survival rate of the hired applicants when it leveled off and became constant; D) use SSA earnings statements that the OFCCP provided of unsuccessful applicants to estimate their interim and mitigation earnings; and E) use the rate that the IRS has established for the underpayment of taxes to calculate prejudgment interest on the back pay awarded.

A. Controlling for job group

The ALJ directed that calculations of damages control for job group rather than job title, as she determined that controlling for job title did not properly account for bias in BOA’s practice of considering and rejecting applicants for job titles for which they had not applied or expressed interest in. See Remedy D. & O. at 4, 14; Liability D. & O. at 54; PX 21 at 3. Thus, Dr. Crawford provided calculations controlling for job group, in addition to controlling for job title, consistent with the ALJ’s directions. Remedy D. & O. at 14. We agree with the ALJ for the reasons she cited that controlling for job group was a reasonable method for the calculations of damages. As we discuss below, the ALJ found it “reasonable” to rely on Dr. Crawford’s calculations over those that BOA proffered from Dr. Johnson. See Remedy R. D. & O. at 15-17, 19.

USDOL/OALJ REPORTER
PAGE 22
 B. Use of W-2 Data and Supplemental Data

Both Dr. Crawford and Dr. Johnson used the W-2 data that BOA provided from the hired applicants to calculate the amount of lost earnings of the unsuccessful applicants. Remedy R. D.
& O. at 7; HT at 24. Dr. Crawford states that BOA provided and he reviewed 249 W-2 forms of applicants who were hired in 1993. PX 22 at 3 ¶ 15. BOA also provided data recorded in the Bank’s eWorkplace database, which was first created in 1999, including the termination dates of its hired applicants. Remedy R. D. & O. at 7 n.7, 8. Using the W-2 data that BOA provided from the hired applicants, the experts were able to determine the average annual earnings of the 1993 hired applicants in each relevant job group (5A2 and 5F2) for each year from 1993 through 2012. Remedy R. D. & O. at 7. In addition, the experts separately used the W-2 data from the 1993 hired applicants to determine the “survival rate” at which they continued to work for BOA over their careers to calculate the “survival rate” or the percentages at which the unsuccessful African-American applicants would have been expected to continue working if they had been hired in 1993 or the 2002-2005 time period. Remedy R. D. & O. at 9; HT at 35. 51

But Dr. Crawford found that the W-2 data that BOA provided was incomplete, as some hired applicants’ W-2s were missing, or either the period covered by some hired applicants’ W- 2s ended before their date of termination as recorded in BOA’s eWorkplace database or extended beyond their termination date, or some hired applicants had more than one recorded termination date while others had no recorded termination date. Remedy R. D. & O. at 7-8. Thus, for calculating the average annual earnings for the 1993 hired applicants, some of the hired applicants had no W-2s showing annual earnings for a particular year to include in the calculation.

51
The parties submitted a Joint Stipulation Regarding Certain Remedy Phase Hearing Exhibits, withdrawing joint exhibits from the record that contained personally identifiable information (PII) regarding BOA employees and unsuccessful applicants. The exhibits include “Excel data workbooks pertaining to Hires and Rejected Applicants from the 1993 and 2002-2005” periods including “Information for Applicants Turned Down in 1993 [and] 2002-2005,” “comprising [of] Excel spreadsheets with information on applicants to Bank of America who were not hired for positions in the 5A2 and 5F2 job groups in 1993 including PII” and “Excel spreadsheets with information on applicants to Bank of America who were not hired for positions in the 5A job group in 2002-2005 including PII.” In addition, the exhibits include “Earnings reports from SSA (BOA-REM-0000593 to BOA-REM-0000838), comprising copies of 121 Social Security Statements containing Earnings Records ” that both Drs. Crawford and Johnson reviewed to calculate the damages including interim earnings and that Dr. Crawford used for applying the IRS interest rate to his back pay calculations.
Finally, the exhibits included “[p]rogramming materials and electronic files supporting” Drs.
Crawford’s and Johnson’s reports.

As the ALJ notes, “[t]hese exhibits are not part of the record in this proceeding.” Remedy R.
D. & O. at 5, n. 5. Thus, we cannot review Drs. Crawford’s and Johnson’s specific calculations, because the parties jointly decided not to rely on these exhibits in their briefing and proceedings before the ALJ and the Board.

USDOL/OALJ REPORTER
PAGE 23
 Whereas Dr. Johnson assumed that any missing earnings information or gaps in an individual’s W-2s reflected periods when the individual was not working, Dr. Crawford utilized two methods to calculate lost earnings for unsuccessful applicants in a particular year when there was missing earnings information due to missing W-2s. With his “Method 1,” Dr. Crawford “filled in” any missing information or gaps due to missing W-2s with estimates from other available data, such as the termination date listed in the eWorkplace database. Remedy R. D. & O. at 8, 15. If W-2s continued beyond the date of termination as recorded in the eWorkplace database, Dr. Crawford assumed that the recorded date of termination was incorrect and the individual continued working up to the year of the last W-2. Also, Dr. Crawford would fill in missing W-2 information for a year with the average of the W-2 incomes on each side of the missing year, or would assume earnings equal to the last reported earnings. Remedy R. D. & O. at 8; see HT at 25. Alternatively, with his “Method 2,” Dr. Crawford assumed like Dr. Johnson that there were zero earnings in years for which there were no W-2s and that the recorded date of termination was correct. Id. Consequently, Method 1 would overestimate lost earnings to some extent, while Method 2 would underestimate lost earnings to some extent. Id.; see also PX 21 at 5. But with the overwhelming majority of the individuals he analyzed, Dr. Crawford found no missing information and, therefore, consistent results under either method. Id.

BOA contends on appeal that Dr. Crawford’s Method 1 assumptions when there was missing W-2 information are illogical and use the eWorkplace database that was not created until 1999.
The ALJ noted that while Dr. Johnson also recognized that the W-2 data that BOA provided was incomplete, he made no attempt like Dr. Crawford to address or supplement the missing information but merely ignored the gaps in making his damages calculations. Remedy R. D. & O. at 15-16. In addition, while Dr. Johnson criticized Dr. Crawford’s Method 1 estimates as being “factually incorrect,” the ALJ rejected Dr. Johnson’s criticism as misplaced and inconsistent with the ALJ’s determination that damages in this case can only be calculated through a class-wide formula. Specifically, the ALJ found that Dr. Johnson’s criticism inaccurately assumes that the record contains information that could allow for a factually accurate calculation of damages on an individual basis, which is inconsistent with the ALJ’s determination that calculating damages based on an individualized basis is not feasible in this case. Remedy D. & O. at 15. We agree with the ALJ that Dr. Crawford’s assumptions about the W-2 were reasonable in determining a class-wide loss of earnings where BOA was responsible for the insufficient or incomplete W-2 earnings information provided and only offered an explanation for missing W-2 information with regard to one individual. Remedy R. D. & O. at 15-16. The ALJ properly determined that factual “exactitude” is not required in calculating damages and that any ambiguities arising from BOA’s providing insufficient or incomplete W-2 earnings information are to be resolved against BOA as the discriminating employer. Remedy R.
D. & O. at 16-17; see Stewart, 542 F.2d at 452. Consequently, we agree with the ALJ’s determination that Dr. Crawford’s Method 1 was reasonable, especially as the result was only a four percentage point difference from Dr. Johnson’s own calculations. Remedy R. D. & O. at 15-17.

USDOL/OALJ REPORTER
PAGE 24
 C. Determining Survival Rate

To determine the “survival rates” or percentages of African Americans who would have continued working for BOA if they had been hired in 1993 or in the 2002-2005 time period, Drs.
Crawford and Johnson again used the annual W-2 forms from hired applicants for the relevant positions at issue, which revealed the rate that they continued working for BOA. Remedy R. D.
& O. at 9. They then assumed that the unsuccessful applicants would have continued to work for BOA or “survived” at the same rate as the successful hired applicants. Id.

But W-2 forms and, therefore, actual survival rate data for the 1993 group of hired applicants was not provided for periods after 2009. Remedy R. D. & O. at 9, 17; see PX 22 at 6 ¶ 18(a)-(b). Dr. Johnson used the survival rate as determined from all the data dating from 1993 onwards, including the early years of employment between 1993 and 1999 when the turnover rate was high. Remedy R. D. & O. at 12, 17. In contrast, the survival rate of the 1993 hired applicants was lower in the early years of their employment between 1993 and 1999, when their turnover rate was high, than it was in the later years of employment between1999-2009 when the survival rate leveled off and became constant. Thus, Dr. Crawford did not consider the data prior to 1999, when the survival rate became constant, and used the later period’s constant survival rate to make estimated damages calculations for periods after 2009 for both the 1993 and 2002-2005 cohorts. Id.; PX 27 at 18-19. Consequently, the ALJ found that Dr. Crawford’s use of the later period’s constant survival rate was reasonable for calculating damages after 2009 in this case. Remedy R. D. & O. at 17. Again, Dr. Crawford’s choices fall within a range of reasonableness in trying to hypothetically determine the loss of earnings caused by a discriminating employer.

D. Establishing Interim and Mitigation Earnings

Interim earnings are the amounts earned from employment that is a substitution for the employment the victim would have had with BOA. See Greenwood Mills, ARB Nos. 00-044, 01-089, slip op. at 11 (quoting OFCCP Federal Contract Compliance Manual, § 7F07(c)(1) (1998)). 52 On appeal, BOA contends that the ALJ erred in holding that BOA has the burden of

proving interim earnings. Instead, BOA reiterates the argument it made before the ALJ that the OFCCP’s burden to prove damages includes the earnings or back pay that the victims of discrimination would be owed absent, or but for, the discrimination minus the actual interim earnings the victims were paid. Only then, BOA asserts, does the burden shift to the defendant to “further” establish that the damages were less than what the OFCCP claimed because the amount of interim earnings the victims earned were actually more than the OFCCP claimed or due to the victims’ lack of diligence to mitigate damages by seeking or accepting other employment. BOA’s Exceptions Brief at 82-85.

52
The OFCCP Federal Contract Compliance Manual is a matter of public record, see http://www.dol.gov/ofccp/regs/compliance/fccm/FCCM_FINAL_508c.pdf; see generally 29 C.F.R. § 18.201, and Dr. Johnson quoted section 7F07 of the Manual, as “publically available information,” in his report, see DX 127 at 9 ¶ 17 n. 18-19, and at 45.

USDOL/OALJ REPORTER
PAGE 25
 The ALJ noted that courts have held that once a plaintiff in a Title VII case has established the amount of damages resulting from the discriminatory acts, the burden of producing further evidence to establish the amount of interim earnings or lack of diligence properly falls on the employer, in this case BOA, citing the holding of the United States Court of Appeals for the Fifth Circuit in Marks v. Prattco, Inc., 633 F.2d 1122 (5th Cir. 1981). R. D. & O. at 13, 17-18.

As the ALJ noted, the Circuit Courts have addressed the issue of which party bears the burden of proof to establish interim earnings in a back pay proceeding in Title VII and National Labor Relations Board (NLRB) cases and held that the plaintiff has the initial burden to show the “gross” amount of back pay due and then the burden of producing sufficient evidence to establish the amount of interim earnings, or lack of diligence to mitigate damages by seeking or accepting other employment, shifts to the defendant.

For instance, in a Title VII pattern or practice class action discrimination case, in which the employer was held to have had a discriminatory maternity leave policy over a multi-year period, the appellate court held that the lower court had erred in requiring the employees to prove their efforts to mitigate their damages. Wooldridge v. Marlene Indus. Corp., 875 F.2d 540, 548 (6th Cir. 1989). Although the Sixth Circuit noted that its “leading” case on the subject in Rasimas v. Michigan Dept. of Mental Health, 714 F.2d 614 (6th Cir. 1983), “involved an individual action, it applies the duty to mitigate damages and the attendant burdens on the parties in the same context and phase of the proceedings as [in a pattern or practice class action discrimination case] before us here.” Wooldridge, 875 F.2d at 548. Because “the issue of mitigation of damages is always considered after discrimination has been established and plaintiff has presented some evidence on damages,” see Rasimas, 714 F.2d at 623, the Sixth Circuit found that “there is no practical distinction between an individual action and a class action for the purpose of determining the issue of mitigation.” Wooldridge, 875 F.2d at 548.

In accordance with the guidelines set forth in Rasimas, 714 F.2d at 623, “[o]nce a claimant establishes a prima facie case and presents evidence on the issue of damages, the burden of producing sufficient evidence to establish the amount of interim earnings or lack of diligence shifts to the defendant.” Thus, the Sixth Circuit Court held in Wooldridge that the defendant has “the burden of producing sufficient evidence to establish the amount of interim earnings.” Wooldridge, 875 F.2d at 548; see also Hance v. Norfolk S. Ry. Co., 571 F.3d 511, 521 (6th Cir. 2009)(citing Rasimas). 53

53
Other Circuit Courts have made similar holdings. As the ALJ noted, the Fifth Circuit Court of Appeals has held that “[o]nce a plaintiff in a Title VII case has established a prima facie case and established what he or she contends to be the damages resulting from the discriminatory acts of the employer, the burden of producing further evidence on the question of damages in order to establish the amount of interim earnings or lack of diligence properly falls to the defendant.” Jurgens v.
E.E.O.C., 903 F.2d 386, 390-391 (5th Cir. 1990)(citing Marks, 633 F.2d at 1125). See also Nord v.
U.S. Steel Corp., 758 F.2d 1462, 1470-1471 (11th Cir. 1985)(citing Marks, 633 F.2d at 1125).

USDOL/OALJ REPORTER
PAGE 26
 Likewise, the Eighth Circuit Court of Appeals has held “[o]nce the gross amount of back pay owed [the plaintiff] has been determined, the burden shifts to [the defendant] to prove what should be deducted from that award as ‘(i)nterim earnings or amounts earnable with reasonable diligence.’” Paxton v. Union Nat. Bank, 688 F.2d 552, 574 (8th Cir. 1982). Finally, the Fourth Circuit Court of Appeals, within whose jurisdiction this case arises, has held that “[o]nce the amount of back pay due a worker has been established, the burden shifts to the employer to produce evidence to mitigate its liability” and “[d]eductions are made from a worker’s gross back pay for the worker’s interim earnings and for losses that were wilfully incurred by an unjustified refusal to take substantially equivalent employment,” but “[t]he defense of a wilful loss of earnings . . . is an affirmative defense and it is the employer’s responsibility to carry that burden.” Lundy Packing Co. v. N.L.R.B., 856 F.2d 627, 629 (4th Cir. 1988) (citations omitted).

In regard to the BOA’s specific argument that it is the plaintiff OFCCP’s burden to prove that damages include back pay minus interim earnings, we note that both the Sixth Circuit and Seventh Circuit Courts of Appeals have rejected this same argument. In N.L.R.B. v. Overseas Motors, Inc., 818 F.2d 517, 521 (6th Cir. 1987)(emphasis added), the Sixth Circuit held:

We also find no merit in [the defendant’s] argument that the Compliance Officer had a duty to investigate the sources and amounts of [the plaintiff’s] possible interim earnings, and that his failure to do so was error. [The defendant] clearly did not comprehend its burden of proof in the back pay proceeding. The General Counsel has a duty only to show the gross amount of back pay due, i.e., the amount of money that the employee would have earned had the employer not violated the National Labor Relations Act. . . . Once he has done so, “the burden is upon the employer to establish facts which would negative the existence of liability to a given employee or which would mitigate that liability.”

Similarly, the Seventh Circuit rejected this argument in N.L.R.B. v. Alwin Mfg. Co., Inc., 78 F.3d 1159, 1163 (7th Cir. 1996), holding:

As for the Board’s alleged interference with [the defendant’s] efforts to comply with the 1994 [NLRB] Order [to comply with the National Labor Relations Act], through its failure to furnish [the defendant] with information about the affected employees’ interim earnings, [the defendant] assumes a duty on the part of the Board that does not exist.
It may be that the Board is not being as helpful to [the defendant] as it might. But [the defendant] was the wrongdoer when it violated the Act by instituting the unilateral changes in minimum performance standards. It is therefore [the defendant’s] burden to collect the information about the aggrieved employees’ interim earnings that is necessary for compliance with the Order.

USDOL/OALJ REPORTER
PAGE 27
 Consequently, we reject BOA’s contention that the OFCCP’s burden to prove damages also includes proving interim earnings.

BOA argued before the ALJ, and reiterates its argument on appeal, that the only information regarding interim earnings that it was permitted to use was a small number of Social Security Administration (SSA) earnings statements that the OFCCP provided, as the ALJ denied its request for the issuance of individual subpoenas or to obtain questionnaire responses from the unsuccessful applicants. BOA’s Exceptions Brief at 79. As previously noted, the Joint Stipulation Regarding Certain Remedy Phase Hearing Exhibits indicates that Drs. Crawford and Johnson were provided “copies of 121 Social Security Statements containing Earnings Records” to calculate interim earnings. Dr. Johnson notes in his reports that the OFCCP provided him SSA earnings statements from 121 of the total class of unsuccessful applicants to review, which represented ten percent of the overall class of unsuccessful applicants, see DX 126 at 2; DX 127 at 10 ¶ 19; 13 - Exh. 2; 19 ¶ 24; HT at 107, 112. Dr. Crawford testified that the OFCCP provided him SSA earnings statements from “roughly” 125 of the total class of unsuccessful applicants to review, see JX 56 at 21, 66-67; PX 21 at 7-8; HT at 40. However, the ALJ found that BOA’s argument again inaccurately assumes that the record contains information that could allow for a factually accurate calculation of damages on an individual basis, which is inconsistent with the ALJ’s determination that calculating damages based on an individualized basis is not feasible in this case. Remedy D. & O. at 19. We also note that in a NLRB News Release issued in 2007, the NLRB noted that “employees’ interim earnings from the time of their discharge” is “usually derived from social security data.” See NLRB Modifies Rule in Backpay Cases Concerning Evidence of Employees’ Job Search Efforts, N.L.R.B. 07-2644, 2007 WL 2956442 (Oct. 11, 2007). Thus, we also discount BOA’s contention regarding the use of SSA earnings statements to determine interim earnings in this case.

Dr. Crawford and Dr. Johnson agreed that it is appropriate to use Social Security (SSA) data to determine interim and mitigation earnings. Remedy R. D. & O. at 18. To estimate the interim or mitigating earnings of the discriminated class of unsuccessful applicants, Drs.
Crawford and Johnson used the SSA earnings statements that the OFCCP provided of unsuccessful applicants. Damages for the unsuccessful applicants includes the earnings or back pay that the victims of discrimination would be owed absent, or but for, the discrimination minus the victims’ actual interim earnings.

At the hearing, Dr. Crawford explained, “I took the Social Security data that OFCCP had obtained from the Social Security Administration and looked at the average, the averages across all the individuals in the cohort. So, for example, all of the ’93 unsuccessful applicants for the 582 jobs [in the 5A2 job category] and a separate calculation for the 5F2 jobs and averaged earnings for [ ] people each year.” HT at 62-63. For years in which unsuccessful applicants’ SSA earnings statements indicate that they had zero earnings, Dr. Crawford treated them as years in which the unsuccessful applicants actually had zero earnings. Remedy R. D. & O. at 19; PX

USDOL/OALJ REPORTER
PAGE 28
 21 at 8; HT at 63. 54 In contrast, Dr. Johnson excluded years in which unsuccessful applicants

had zero earnings from his calculations, which resulted in an increase in the amounts of interim earnings and decreased the amount of damages that he calculated. Remedy R. D. & O. at 19.

The ALJ found that Dr. Johnson’s calculations were based on an unreasonable assumption that those unsuccessful applicants whose SSA earnings records showed years of zero earnings were not making reasonable efforts to secure employment and that any ambiguities or incomplete information are to be resolved against BOA as the discriminating employer. Remedy R. D. & O. at 19, n.14; see Stewart, 542 F.2d at 452. Given the facts of this case and the litigation history, we find that the OFCCP offered a reasonable estimate of interim earnings from the records of more than 100 employees to offset the estimated lost earnings of approximately 50 employees.

E. Prejudgment Interest Rate on the Back Pay Award

Interest is paid on back pay awards to compensate the discriminatees for the loss of the use of their money. OFCCP Compliance Manual, §7F07(e) (1998); see also Greenwood Mills, ARB Nos. 00-044, 01-089, slip op. at 12. In determining the amount of prejudgment interest on the back pay that the ALJ awarded, the ALJ used the rate that the IRS has established for the underpayment of taxes, the rate required by the regulations. 41 C.F.R. § 60-1.26(a)(2). Remedy R. D. & O. at 19. But BOA argues that as its expert Dr. Johnson advocated, see DX 127-128, the three-month U.S. Treasury bill rate should have been used as the appropriate interest rate for determining the amount of prejudgment interest.

But “an agency is bound by its own regulations, and an agency head acting in an adjudicatory capacity has no authority to review the validity of those regulations.” OFCCP v.
Goya De Puerto Rico, Inc., ARB No. 99-104, ALJ No. 1998-OFC-008, slip op. at 6 (ARB Mar.
21, 2002) (citing United States v. Nixon, 418 U.S. 683, 695-696 (1974)) (“So long as the regulation is extant it has the force of law.”). 55 Consequently, we affirm the ALJ’s reliance on

the IRS interest rate in determining the amount of prejudgment interest on the back pay that the

54
Dr. Crawford’s “annual average annual mitigating earnings” and “annual average annual mitigating benefits” calculations for the 1993 5A2 and 5F2 job groups, as credited by the ALJ, for the years 1993 through 2012 are found in his “Corrected Table 3-5A2-M1” and “Corrected Table 3- 5F2-M1” at PX 27 Exhibit 2 at 32, 34.
55
Accord Secretary’s Order No. 02-2012 (Delegation of Authority and Assignment of Responsibility to the Administrative Review Board), 77 Fed. Reg. 69,378, § 5(c)(66) (Nov. 16, 2012)(“The Board shall not have jurisdiction to pass on the validity of any portion of the Code of Federal Regulations that has been duly promulgated by the Department of Labor and shall observe the provisions thereof, where pertinent, in its decisions.”).

USDOL/OALJ REPORTER
PAGE 29
 ALJ awarded as it is “in accordance with the Executive Order implementing regulations regarding such payments set out in 41 C.F.R. § 60-1.26(a)(2).” Id. 56

5. The ALJ’s Back Pay Award is Affirmed

The ALJ credited Dr. Crawford’s Method 1 calculation of the back pay damages for the 1993 5A2 and 5F2 job groups of unsuccessful applicants.

Adding the total lost earnings and benefits, plus IRS interest, for each earnings year from 1993 through 2012, Dr. Crawford found that the total lost earnings and benefits plus IRS interest for job group 5A2 was $259,387, see PX 27 at 32 (Corrected Table 3-5A2 ) and was $704,646 for job group and 5F2, see PX 27 at 34 (Corrected Table 3-5F2-M1), which resulted in an overall total of $964,033 in back pay damages for the 5A2 and 5F2 job groups in the 1993 cohort of unsuccessful applicants. See also PX 27 at 68 (Corrected Table 11-Alt, indicating for the 5A2 and 5F2 job groups in the 1993 cohort of unsuccessful applicants during the earnings year from 1993 through 2012 a total lost earnings and benefits of $762,279, plus a total IRS interest of $201,754, resulting in an overall total of $964,033 in back pay damages).

Thus, the ALJ credited Dr. Crawford’s calculation of an overall total of $964,033 in back pay damages for the 1993 5A2 and 5F2 job groups of unsuccessful applicants during the earnings years from 1993 through 2012. The ALJ found that Dr. Crawford’s assumptions in calculating the amount of back wage damages were reasonable to resolve the ambiguities BOA created in failing to provide complete earnings data. Remedy R. D. & O. at 20. Thus, resolving ambiguities against BOA, the ALJ credited as reasonable Dr. Crawford’s higher Method 1 calculation of the back pay damages for the 1993 5A2 and 5F2 job groups (and the 2002-2005 unsuccessful applicants).

After reviewing the ALJ’s findings and BOA’s contentions, we conclude that the ALJ, within her broad discretion, permissibly resolved ambiguities in the record in what the unsuccessful applicants would have earned but for BOA’s discrimination. Stewart, 542 F.2d at 452. Moreover, as exactitude is not required in calculating the amount of back wage damages, and the ALJ is granted wide discretion in making such a determination, we hold that the ALJ neither abused her discretion nor clearly erred in crediting Dr. Crawford’s calculation of back pay damages for the 5A2 and 5F2 job groups in the 1993 cohort of unsuccessful applicants. Id.
Consequently, we affirm the ALJ’s award of $964,033 in back pay damages for the 1993 5A2 and 5F2 job groups of unsuccessful applicants during the earnings year from 1993 through 2012.

56
The actual effective annual IRS interest rates from 1993 through 2015, assuming quarterly compounding, that Dr. Crawford used in making his damages estimates are located in PX 27 at 74, Exhibit 2 - Appendix Table C; see also PX 27 at 13 ¶ 31 n. 14. Dr. Crawford’s annual “IRS Interest” calculations for the 1993 5A2 and 5F2 job groups, as credited by the ALJ, for the years 1993 through 2012 are found in his “Corrected Table 3-5A2-M1” and “Corrected Table 3-5F2-M1” at PX 27 Exhibit 2 at 32, 34.

USDOL/OALJ REPORTER
PAGE 30
 As for the damage award for the 2002-2005 applicants, we must reverse the ALJ’s award having rejected the ALJ’s finding that BOA intentionally and unlawfully discriminated against African-American candidates in hiring for entry level positions in 2002-2005. Remedy R. D. & O. at 22.

6. Other Remedies

Because the BOA has not raised any exceptions concerning the other remedies that the ALJ ordered, we find the ALJ’s award is final as to those remedies.

CONCLUSION

Because BOA did not present any other relevant evidence to the ALJ at the merits hearing, the previous Board ruling is dispositive on BOA’s objections to the OFCCP’s selection of BOA for a compliance review as an illegal search in violation of the Fourth Amendment in this case and BOA’s Fourth Amendment arguments are REJECTED.

Furthermore, because the evidence supports the ALJ’s findings and establishes a pattern or practice by BOA of intentional racial discrimination against African-American applicants in 1993 for jobs in the 5A2 and 5F2 job groups, the ALJ’s conclusions that BOA intentionally discriminated against African Americans in 1993 is AFFIRMED. In addition, because the ALJ ordered a reasonable remedy for the unsuccessful 1993 African-American applicants, the ALJ’s award of $964,033 in back pay damages and other remedies for the 1993 5A2 and 5F2 job groups of unsuccessful applicants is AFFIRMED. Further, it is also ORDERED that BOA pay an amount of additional interest on the back pay award for the period from September 18, 2013, to the date on which BOA has paid the monetary judgement in full.
This additional interest payment should be made in accordance with the regulations regarding such payments set out in 41 C.F.R. § 60-1.26(a)(2).

As to the discrimination claims for the period 2002-2005, the OFCCP improperly found BOA liable. Consequently, the ALJ’s finding of a pattern or practice of intentional discrimination during the 2002-2005 period and remedy orders pertaining to the 2002-2005 period are REVERSED.

SO ORDERED.

LUIS A. CORCHADO
Administrative Appeals Judge

USDOL/OALJ REPORTER
PAGE 31
 Judge Brown, concurring:

I concur in the majority’s resolution of this appeal. I write separately because I am of the opinion that BOA’s due process rights were violated by the way in which the OFCCP incorporated the job data for 2002-2005 into the enforcement proceeding before the ALJ without having first afforded BOA the procedural protections mandated under EO 11246 and 41 C.F.R.
Part 60-1.

On appeal, BOA argues that “the ALJ erred in permitting OFCCP to expand the allegations of discrimination beyond 1993.” 57 I agree. “Where the rights of individuals are

affected, it is incumbent upon agencies to follow their own procedures.” Morton v. Ruiz, 415 U.S. 199, 235 (1974). While not necessarily rising to the level of a constitutional due process violation, see United States v. Caceres, 440 U.S. 741, 751-55, 751 n.14, 754 nn.18, 19 (1979), federal courts have repeatedly overturned actions of administrative agencies, under the arbitrary and capricious doctrine, for failure to follow their own procedures, especially where individual rights were involved. See Mabey v. Reagan, 537 F.2d 1036, 1042 (9th Cir. 1976); D’Iorio v.
Cty. of Del., 447 F. Supp. 229, 240-41 (E.D. Pa.) (vacated on other grounds, 592 F.2d 681 (3d Cir. 1978). Required is compliance with procedural regularity before governmental action is taken, i.e., that the government follow the procedures it has specified by its laws, regulations, and customs. Vitarelli v. Seaton, 359 U.S. 535, 539-40 (1959); United States ex rel. Accardi v.
Shaughnessy, 347 U.S. 260, 266-69 (1954).

The regulations implementing EO 11246 authorize the OFCCP to conduct compliance evaluations of a covered contractor “to determine if the contractor maintains nondiscriminatory hiring and employment practices.” 41 C.F.R. § 60-1.20(a). These evaluations take the form of a “compliance review,” which can involve a “desk audit” in which the OFCCP analyzes contractor-provided data at its own office, an on-site review conducted at the contractor’s establishment, id. at § 60-1.20(a)(1)(i)-(ii), “an off-site analysis of information supplied by the contractor or otherwise gathered during or pursuant to the on-site review,” id. at § 60- 1.20(a)(1)(iii), or all three. Id. at § 60-1.20(a)(1). Where deficiencies are found as a result of the review and evaluation, provision is made for securing compliance through a conciliation process that allows the offending contractor to voluntarily commit to correcting the identified violations.
Upon approval, the contractor will be considered to be in compliance. Id. at § 60-1.20(b).

Where identified violations are not corrected pursuant to the conciliation process, administrative enforcement proceedings are authorized. 41 CFR § 60-1.26(b). An enforcement proceeding, which is commenced through the filing of an administrative complaint that is heard before a Department of Labor ALJ, id. at §§ 60-1.26(b)(2), 60-30.5(a), must be preceded by the service of a notice giving the contractor 30 days to show cause why the enforcement proceeding should not be instituted. Id. at §§ 60-1.28, 60-2.2(c)(1). During the 30-day period, the regulations provide that “every effort” is to be made “through conciliation, mediation, and 57
BOA’s Exceptions Brief at 28.

USDOL/OALJ REPORTER
PAGE 32
 persuasion to resolve the deficiencies.” Id. If the contractor fails within the 30-day period to show good cause for its failure to remedy the identified violations, the case is processed for enforcement through the filing of the administrative complaint. Id. at § 60-2.2(c)(2). Such proceedings can result in civil sanctions, including the cancellation or termination of existing contracts, as well as debarment from further contracts. E.O. 11246, § 209(a).

As BOA notes on appeal, it is undisputed that in the instant case the foregoing procedures were followed with regard to the 1993 violations. Specific violations were identified pertaining to two job group classifications, 5A2 and 5F2, pursuant to the OFCCP’s November 1993 selection notice. The ensuing compliance review resulted in a Notice of Violation being issued on October 19, 1994, as amended by a subsequent Notice of Violation issued June 29, 1995, accompanied by a proposed Conciliation Agreement. Upon rejection by BOA of the conciliation process, the required Notice to Show Cause was issued on August 17, 1995, in which the violations charged against BOA were again identified. When the Notice to Show Cause failed to result in compliance, the Administrative Complaint commencing enforcement was filed with the Office of Administrative Law Judges on July 18, 1997. At all times throughout this process, the charges of violations were expressly limited to the 1993 period.

It is also undisputed that the foregoing procedures were not followed with regard to the violations that the OFCCP subsequently identified for the 2002-2005 period after enforcement proceedings had commenced with respect to the 1993 violations. The OFCCP did not commence a compliance review for the subsequently identified violations. Nor did the agency issue a notice of violation or notice to show cause for, nor engage in conciliation efforts related to, any period other than 1993. Instead, the charges of discrimination against BOA in the hiring of individuals for the 2002-05 period arose out of discovery conducted pursuant to the enforcement proceeding the OFCCP initiated against BOA for the 1993 violations.

Not only were the violations for the 2002-05 period incorporated into an enforcement proceeding without the benefit of the procedural protections afforded by EO 11246 and 41 CFR Part 60-1, the violations charged against BOA for the 2002-05 period did not involve the same job categories in all respects as those identified for the 1993 period, but identified new job categories for which BOA had not previously been charged with violations. The 1993 “5A2” job group consisted of prime time tellers, part-time tellers, and full-time tellers. The “5F2” job group for the 1993 period consisted of several clerical and administrative positions, including data entry operators, account clerks, and remittance processing specialists (or “lockbox” specialists). In comparison, both Dr. Haworth and Dr. Crawford, the parties’ respective experts, testified that the job makeup of the 2002-05 “5A” group for which violations were charged differed in a number of significant respects. 58

58
Regarding the job make-up of 2002-2005 “5A” job group, Dr. Haworth testified:

A There is a distinction between the jobs in what you're calling the job group 5A. Some of the jobs are Teller jobs like we

USDOL/OALJ REPORTER
PAGE 33
 Based on the foregoing, I am of the opinion that the failure of the OFCCP to afford BOA the procedural safeguards under EO 11246 to which BOA was entitled prior to commencement of an enforcement action dictates rejection of the ALJ’s finding of violations by BOA for the 2002-05 period. I do not mean to suggest that the OFCCP was not entitled to pursue discovery beyond the 1993 period as part of the enforcement action filed charging violations for the identified period. Given the overall purposes of EO 11246, such post-violation discovery would be warranted in order to determine, for example, if the charged violations are continuing or have been remedied. However, as the comments accompanying the regulations in effect at all times relevant to this action attest, up until the present case it has apparently been the OFCCP’s practice to limit its assessment of liability for discriminatory practices in a contractor’s employment practices to the two-year period prior to the initiation of a compliance review. See 61 Fed. Reg. 25,518 (May 21, 1996); 57 Fed. Reg. 48,105 (Oct. 21, 1992). Absent a change or amendment to the OFCCP’s regulations permitting the agency to institute an enforcement action for violations identified for periods subsequent to those initially charged, I find nothing that persuades me from my conclusion that the different course the OFCCP instituted in this case was both arbitrary and capricious. Without the benefit of the non-adversarial pre-enforcement procedures for reconciling charges of discriminatory hiring practices, the charges of violations for the 2002-05 period, pertaining to job categories that appear significantly different from those pertaining to the 1993 period, were thrust upon BOA as part of the ongoing enforcement proceeding, bringing with it the potential for criminal and civil sanctions under EO 11246. For this reason, I reject the ALJ’s finding of liability for the 2002-05 period, and thus concur in dismissing the charges of violation against BOA pertaining to this period.

E. COOPER BROWN
Administrative Appeals Judge

were analyzing in 1993, but some of the other jobs are very difficult to compare to anything that was going on in 1993.

HT at 582. Dr. Crawford corroborated Dr. Haworth to the same effect:

Q In the 2002 through 2005 time period the testimony is, is that the jobs in that time period are different than the jobs that you were looking at in ‘93, is that right? For example, was the Remittance Processing Specialist job still in existence?
A I don’t believe so, no.
Q Were the Support Specialist jobs in existence?
A I don’t believe so, no.
Q How about the Proof Operator?
A I don’t recall that one specifically, but there were lots of jobs that disappeared.
HT at 511.

USDOL/OALJ REPORTER
PAGE 34
 Judge Royce, concurring in part and dissenting in part:

I agree with the plurality opinion rejecting BOA’s Fourth Amendment arguments and finding BOA engaged in a pattern or practice of disparate treatment against African-American applicants in 1993. I also agree that the ALJ ordered a reasonable remedy for the unsuccessful 1993 African-American applicants. But I do not agree that BOA may not be held liable for disparate treatment against African Americans during the 2002-2005 time period.
The plurality opinion analyzes the 2002-2005 time period as a separate claim of pattern or practice race discrimination rather than as a continuation of the 1993 claim for several reasons. First, the opinion notes that the ALJ relied on substantially less evidence to conclude that a pattern or practice of disparate treatment occurred during the 2002-2005 time period.
Also, as the opinion points out, the hiring process changed significantly between 1993 and the 2002-2005 period. Finally, the opinion states that “the evidence presented simply does not support the idea that the same pattern or practice of intentional discrimination applies to the hiring practices in 1993 and 2002-2005.” The ALJ found otherwise holding that a class-wide approach to damages was warranted given the size and complexity of the case and that “the illegal practices continued over an extended period of time.” 59 I would affirm the ALJ’s holding

that BOA unlawfully discriminated against African-American applicants for certain entry-level teller, clerical, and administrative positions in 1993 and 2002-2005.
As the ALJ stated in her March 2, 2005 Status Order Regarding Outstanding Discovery Motions, the OFCCP’s 1997 Administrative Complaint alleged “that the Defendant violated Executive Order 11246 since at least January 1, 1993 by failing to hire minority applicants for certain entry level clerical and administrative positions based on their race. . . .” 60 The OFCCP’s

discovery requests in 2004 requested documents relevant to those allegations which the ALJ found reasonable because they were “narrowly tailored to the allegations in the Administrative Complaint.” The ALJ stated:
In the Administrative Complaint, the Plaintiff [OFCCP] alleges that the Defendant [BOA] discriminated against minorities in filling certain positions in three job groups at the Defendant’s Charlotte facility, from at least January 1, 1993. The Plaintiff’s discovery requests are specifically directed to the Defendant’s hiring practices in these three job groups, seeking information such as the identity of persons interviewed for the positions, the persons hired, the reasons for non-selection, and the selection criteria. The Plaintiff’s document requests seek information and documents containing personnel information. I find that these requests are 59
Remedy R. D. & O. at 3.
60
March 2, 2005 Status Order Regarding Outstanding Discovery Motions at 2.

USDOL/OALJ REPORTER
PAGE 35
 narrowly tailored to the allegations in the Administrative Complaint, and are relevant, or may lead to relevant information.[ 61]

The ALJ reasoned that because the Administrative Complaint alleged violations “since at least January 1993,” the Complaint contained “an allegation of ongoing violations.” 62 As such, she

explained, it was not necessary for the OFCCP to separately investigate, make findings, and attempt to conciliate each additional violation by BOA because it would be impractical and inefficient since the case was already in litigation. 63

In Honeywell I, the Secretary explicitly approved the OFCCP’s authority to allege and prove post-complaint, related practices that may be considered part of the affected class. 64 The

Secretary reasoned that, “[i]n comparable situations under Title VII, courts have permitted both private plaintiffs and the EEOC to prove that acts the same as or similar to those alleged in the charge, but taking place after it was filed, have occurred.” For example, in Oubichon v. N. Am.
Rockwell Corp., 482 F.2d 569 (9th Cir. 1973), the court held that the “complaint . . . may encompass any discrimination like or reasonably related to the allegations of the EEOC charge, including new acts occurring during the pendency of the charge before the EEOC.” 65

Additionally, recent OFCCP regulations explicitly authorize the OFCCP to expand the temporal scope of compliance investigations. Effective March 24, 2014, the OFCCP implemented new regulations for the Rehabilitation Act at 41 C.F.R. § 741.60(a)(1)(i) (2015) and for the Veterans’ Act at 41 C.F.R. § 60-300.60 (2015) that state that “OFCCP may extend the temporal scope of the desk audit beyond that set forth in the scheduling letter if OFCCP deems it necessary to carry out its investigation of potential violations of this part.” When the rule was implemented effective March 24, 2014, the comments stated:

[t]he purpose of this proposal was to clarify that OFCCP may need to examine information after the date of the scheduling letter during the desk audit in order to determine, for instance, if violations are continuing or have been remedied. While the existing section 503 provision addresses the authority of the agency to conduct desk audits, it does not expressly state the temporal scope of these audits. It has been OFCCP’s longstanding

61
Id. at 4.
62
Id.
63
Id.
64
Honeywell I, 1977-OFC-003, 1993 WL 1506966, slip op. at 7.
65
482 F.2d at 571 (citations omitted).

USDOL/OALJ REPORTER
PAGE 36
 position that the agency has authority to obtain information pertinent to the review for periods after the date of the letter scheduling the review, including during the desk audit. However, in 2010 an ALJ disagreed in a recommended decision in the Frito- Lay case, in part because the parallel Executive Order 11246 desk audit regulation at issue in the case does not address the temporal scope of a desk audit. OFCCP v. Frito-Lay, Inc., Case No. 2010- OFC-00002, ALJ Recommended Decision and Order (July 23, 2010). On May 8, 2012, the Department’s Administrative Review Board (ARB) reversed this recommended decision, concluding that a desk audit authorized by the regulation permitted OFCCP to request additional information relating to periods after the scheduling letter. The ARB concluded that the regulation does not have an inflexible temporal limitation. OFCCP v. Frito-Lay, Inc., Case No. 2010-OFC-00002, ARB Final Administrative Order (May 8, 2012). OFCCP views the Frito-Lay decision as equally applicable to desk audits concluded under its section 503 authority as to those conducted under its Executive Order 11246 authority.
Nevertheless, the final rule makes the clarification explicit in the text of the regulation. OFCCP notes that paragraph (a)(1) also authorizes OFCCP to request during the desk audit additional information pertinent to the review after reviewing the initial submission. See United Space Alliance v. Solis, 824 F. Supp. 2d 68, 81-82 (D.D.C. 2011) (holding that agency’s interpretation of its desk audit regulation to authorize additional information requests when necessary was entitled to deference).[ 66]

BOA argues that even if the ALJ properly allowed the OFCCP post-1993 discovery, “it was improper to permit the OFCCP to expand the liability period beyond 1993.” 67 The

Secretary rejected a similar objection to the scope of liability in Honeywell I: “I reject Defendant’s claim that no one hired after September 25, 1975 and discriminated against by these practices or related, succeeding practices, can be considered part of the affected class.” 68 The

Secretary explained that since the defendant (Honeywell) had ample notice of the allegations against it and opportunity to prepare a defense, the OFCCP should not be precluded from seeking relief for continuing discrimination; “if women were still discriminatorily assigned to certain seniority groups after 1975. . . nothing in the record shows the government waived the right to

66
78 Fed. Reg. 58,682, 58,711-58,712 (Sept. 24, 2013)(emphasis added).
67
BOA’s Exceptions Brief at 28.
68
Honeywell I, 1977-OFC-003, 1993 WL 1506966, slip op. at 7.

USDOL/OALJ REPORTER
PAGE 37
 seek relief for them and all other new hires in perpetuity.” 69 The Secretary reasoned that

“[g]iven the findings of the compliance review and the history of negotiations discussed above, Defendant cannot claim it did not have notice of the nature of the violations OFCCP alleged in the complaint, and litigated at the hearing.” 70 Similarly, in this case, BOA cannot seriously

claim it lacked fair notice of the OFCCP’s allegations by the time this case was finally tried in 2008. 71 Indeed, given BOA’s general familiarity with OFCCP compliance reviews, 72 BOA’s

conciliation negotiations following the Notice of Violations in this case, and its ongoing affirmative action responsibilities as a government contractor, the evidence demonstrating that BOA was engaged during 2002-2005 in the same conduct for which it was cited in 1993 is all the more telling. While BOA need not have completely eliminated racial inequalities to avoid charges of ongoing disparate treatment, it should have been on notice, at a minimum, of its responsibility to take measures to police misconduct in connection with the affected class the OFCCP identified in the 1997 Administrative Complaint. In any case, under the circumstances of this case, including the long delays occasioned by litigation of numerous issues collateral to the merits, as well as the failure of BOA to keep records relevant to the charges for many years, the ALJ did not err in holding BOA liable for continuing unlawful discrimination during 2002 through 2005.

The changed circumstances of the hiring process between the 1993 and the 2002-2005 periods is identified by my colleague as a principal reason why BOA may not be held liable for allegations of ongoing discrimination, despite agreement that discrimination was proved in 1993.
He cites, for example, undisputed facts that the two recruiters in 1993 were replaced by 58 different recruiters by 2002. Certainly, if the theoretical concept of disparate treatment is viewed as based upon individual bias or intent, any discrimination which occurred in the 2002-2005 period with different actors could not be considered as a “continuation of the 1993 claim.” However, Supreme Court precedent of systemic disparate treatment does not rely on finding individual instances or even several instances of disparate treatment, but may be proved by statistics and other evidence of entity disparate treatment or disparate treatment in the aggregate. 73 Intentional individual discrimination, or even deliberate indifference, does not

69
Id.
70
Id. at 8.
71
See Liability R. D. & O. at 1-2. On August 17, 1995, the OFCCP issued a Notice to Show Cause why enforcement proceedings should not be initiated against NationsBank for violations including disproportionate hiring of minorities in job groups 5A2 and 5F2. On July, 18, 1997, the OFCCP filed a complaint against NationsBank for violating the nondiscrimination requirements of EO 11246 and its implementing regulations “since at least January 1, 1993.” 72
See OFCCP’s Response Brief at 96.
73
Further, as the Supreme Court noted in Teamsters, the “pattern or practice” language in Title VII was not intended as a term of art. The Court quotes Senator Humphrey’s explanation of the

USDOL/OALJ REPORTER
PAGE 38
 define the limits of liability under systemic disparate treatment law. As Professor Green, a frequent commentator on employment discrimination explains:
Entity liability is established in systemic disparate treatment cases not vicariously, based on a finding of an individual instance or even several instances of disparate treatment but directly and systemically, based on a finding that individual instances of disparate treatment are so widespread within the organization (as shown by statistics and other evidence of disparate treatment in the aggregate) that the entity is in some part to blame.[ 74]

This ability of disparate treatment law to expose and address systemic discrimination–as opposed to discrete acts of intentional individual discrimination–is a critical regulatory tool for enforcement of remedial laws such as EO 11246 that work to counter increasingly subtle but ostensibly persistent forms of workplace discrimination. As Professor Green advises: “a systemic disparate treatment law that captures disparate treatment in the aggregate serves a role in the regulatory scheme that a law focused on individual instances of disparate treatment does not.” 75

The ALJ similarly conceptualized disparate treatment law: “As the OFCCP has argued, it does not allege that particular individuals made discrete, biased decisions, but that bias pervaded the Bank’s hiring practices. These allegations are against the Bank, not individual decision makers for discrete acts of discrimination.” 76 For related reasons, the ALJ correctly rejected

BOA’s argument that the OFCCP may not rely on statistical analysis alone, but must proffer anecdotal evidence of specific instances of intentional discrimination. Both Hazelwood and Teamsters countenance proof of a pattern or practice of discrimination by statistical evidence alone. 77 And neither case requires a showing of “purpose to harm.” In the seminal case, Segar

term: “(A) pattern or practice would be present only where the denial of rights consists of something more than an isolated, sporadic incident, but is repeated, routine, or of a generalized nature. There would be a pattern or practice if, for example, a number of companies or persons in the same industry or line of business discriminated, if a chain of motels or restaurants practiced racial discrimination throughout all or a significant part of its system, or if a company repeatedly and regularly engaged in acts prohibited by the statute.” Teamsters, 431 U.S. at 336 n.16 .
74
Tristin K. Green, The Future of Systemic Disparate Treatment Law, 32 BERKELEY J. EMP. & LAB. L. 395, 428 (2011) (emphasis added).
75
Id. at 434.
76
Liability R. D. & O. at 57.
77
Teamsters, 431 U.S. at 339 (quoting United States v. Ironworkers Local 86, 443 F.2d 544, 551 (9th Cir. 1971)), the Court recognized that “[i]n many cases the only available avenue of proof is the use of racial statistics to uncover clandestine and covert discrimination by the employer or union

USDOL/OALJ REPORTER
PAGE 39
 v. Smith, the D.C. Circuit explained that in certain cases, if a plaintiff’s statistical methodology is appropriate and generates evidence of discrimination at a statistically significant level, neither anecdotal evidence, nor gross disparity is necessary to prove discrimination. 78 The OFCCP’s

statistical analysis of BOA’s hiring practices during the 2002-2005 period is such a case, and I would affirm the ALJ’s finding that BOA is liable for disparate treatment during that time period.
The OFCCP was entitled to investigate whether violations it identified during the course of its compliance review were continuing or had been remedied. However, because of long delays occasioned by litigation of numerous issues collateral to the merits, as well as the failure of BOA to retain applicant data for the years 1994 to 2001, the OFCCP was precluded from investigating the possibility of continuing violations until it received BOA data for the 2002- 2005 period. After analyzing that data, the OFCCP’s expert, Dr. Crawford, concluded that, whether controlling for job title or group, the data revealed a statistically significant departure from race neutrality. The shortfall was statistically significant at 4 standard deviations, and therefore “extremely unlikely to have occurred by chance.” 79

As the ALJ found and the plurality opinion affirms, the OFCCP proved disparate treatment of African-American applicants for entry-level job groups in the Charlotte facility in 1993. The OFCCP offered a statistical analysis that demonstrated discrimination against many of the same class of applicants, 80 in the same facility, between 2002 and 2005. As the ALJ

found, BOA neither successfully attacked the OFCCP’s statistical method nor demonstrated that the statistical disparities could be explained by non-discriminatory factors. I would affirm the ALJ’s finding that the OFCCP established by a preponderance of the evidence that BOA’s illegal practices against African-American candidates for entry level positions continued over an
involved.”); Hazelwood, 433 U.S. at 307-308 (stating that “[w]here gross statistical disparities can be shown, they alone may in a proper case constitute prima facie proof of a pattern or practice of discrimination”).
78
Segar, 738 F.2d at1278 (“It is not difficult to understand that discrimination might exist even when affected individuals can point to no specific instances of an employer’s discriminatory conduct.
The days of Bull Connor are largely past; discrimination now works more subtly. Yet its effects are no less pernicious.”).
79
Liability R. D. & O. at 17.
80
“For the years between 2002 and 2005, the positions that were in the 5A2 and 5F2 groups became part of EEO job groups 5A and 5B.” Liability R. D. &O. at 5. The positions in the 5A2 group were most comparable to the 5A positions in 2002-2005. Remedy R. D. & O. at 12, n.12. Job Group 5A2 contained three jobs titled: Part-time Tellers, Prime time Tellers and Teller I. BOA’s Exceptions Brief at 13. “But in this case, the jobs involved were all entry level, in a certain geographic location, and did not require specialized knowledge or skills, and the hiring process and criteria were similar for all of the jobs.” Liability R. D. & O. at 53.

USDOL/OALJ REPORTER
PAGE 40
 extended period of time as demonstrated by the statistical disparity of 6.9 standard deviations for entry level positions between African-American and white job applicants during 1993, and a statistical disparity of 4 standard deviations for similar entry level positions during 2002-2005.
In the interest of concluding this case, I reserve decision on whether the remedy determined by the ALJ for the 2002-2005 period may be upheld. I note only that the ALJ fashioned a remedy which appeared to be appropriately specific to the continuing violation actually proven during the period 2002 to 2005. 81 The OFCCP requested substantial additional

injunctive relief to prevent further illegal behavior, but the ALJ found “no evidence to suggest that the Bank is persisting in illegal discriminatory hiring” and declined to award any such injunctive relief. 82

JOANNE ROYCE
Administrative Appeals Judge

81
Remedy R. D. & O. at 21.
82
Id. at 22.

USDOL/OALJ REPORTER
PAGE 41