About Us  |  About Cheetah®  |  Contact Us

Experts discuss pay equity analysis and its relation to OFCCP compliance

November 16th, 2016  |  Cynthia L. Hackerott

While the issue of pay equity continues to be a focus of the OFCCP, not to mention the political landscape, what type of pay analysis is actually required under OFCCP regulations and to what extent should a federal contractor conduct any pay analysis outside of the regulatory requirements? Employment law experts John C. Fox, president of Fox, Wang & Morgan P.C. in Los Gatos, California, and Brian W. Bulger, of counsel to Cozen O’Connor in Chicago, Illinois, addressed these topics in separate presentations at the National Employment Law Institute’s (NELI) Thirty-Fourth Annual Affirmative Action Briefing in Chicago, Illinois.

What do OFCCP regulations require? The OFCCP’s regulations at 41 CFR §60-2.17(b)(3) require contractors, as an element of their Affirmative Action Programs (AAPs), to conduct a self-evaluation of their compensation systems “to determine whether there are sex-, race-, or ethnicity-based disparities.” Bulger and Fox both pointed out that the regulations do not specify how contractors must analyze their systems. Bulger further noted that, in the preamble to the OFCCP’s new sex discrimination regulations at 41 CFR Section 60-20 (which took effect on August 15, 2016), the agency states that contractors are allowed “substantial discretion” in determining their own evaluation process for complying with Section 60–2.17(b)(3) (See, 81 FR 39125-39126).

Very little is required of contractors to meet the Section 60–2.17(b)(3) requirement, Fox said. Contractors can meet the requirement via a qualitative narrative that does not contain any statistical analysis. Even a statement as simple as “we investigate any complaints about pay of our employees,” has been found in recent administrations to be compliant with the regulation, he noted.

Should contractors do anything beyond the regulatory requirement? Aside from the regulatory requirement, contractors may exercise their discretion to conduct a substantive analysis of their compensation systems, Fox said. They may decide to do so as a means to reduce litigation risk and/or prepare to defend against any OFCCP analysis of their compensation systems in an audit. Still, Fox points out that contractors really cannot accurately anticipate what type and extent of compensation analysis the OFCCP will actually undertake in any given audit.

Privilege. Both Bulger and Fox emphasized that when contractors do any pay analysis beyond the very minimal requirements of the regulations, they should do so under attorney-client privilege. Actions done pursuant to regulation or statute are not protected by attorney-client privilege, they cautioned, and information not protected by attorney-client privilege is subject to discovery in litigation. Therefore, contractors must be clear in their documentation and file-keeping whether a given action has been done voluntarily or pursuant to legal requirements, they advised. Contractors should very specifically identify what they are doing to meet the requirements of Section 60–2.17(b)(3), and label all else attorney-client privilege, Fox said.

Diagnostic tools. The two major compensation diagnostic tools are: (1) cohort analyses, and (2) multiple regression analysis, Fox noted. Cohort analyses simply compares the factors which affect pay as applied to any two given similarly situated employees to determine what legitimate non-discriminatory reasons, if any, explain a pay gap between the two.

Multiple regression analysis (MRA) is a statistical tool for understanding the relationship between two or more variables. In evaluating compensation, an MRA measures the impact of each potential explanatory variable — such as experience, time in grade, shift, education, gender, race, or ethnicity  — on the dependent variable — compensation — by holding all other explanatory variables constant among similarly situated employees. The analyses creates figures demonstrating how much, if any, of an observed disparity in compensation can be traced to a given protected category (i.e. gender, race, or ethnicity), as opposed to any other potentially explanatory variable. Put another way, the purpose of an MRA is to determine whether there are any significant differences in compensation among similarly situated employees by gender, race or ethnicity that are not explained by legitimate factors.

Statistical significance. Under the applicable legal standards, a finding of systemic compensation discrimination must be based on disparities that are “statistically significant,” i.e., those that could not be expected to have occurred by chance. The OFCCP has said that a statistically significant disparity occurs at a level of two or more standard deviations, based on measures of statistical significance that are generally accepted in the statistics profession. Size matters in putting together groups of similarly situated employees, Fox explained, and any group of less than 30 employees is too small of a sample for statistical analysis. For groups under 30 employees, cohort analysis can be used.

Bulger pointed out that the OFCCP’s new sex discrimination regulations at 41 CFR 60–20.4(a) contain a very broad, flexible definition of “similarly situated” for the purposes of evaluating compensation differences. This section of the regulations states that “[r]elevant factors in determining similarity may include tasks performed, skills, effort, levels of responsibility, working conditions, job difficulty, minimum qualifications, and other objective factors. In some cases, employees are similarly situated where they are comparable on some of these factors, even if they are not similar on others.”

Courts will likely take a narrower view on this definition than the agency, Bulger predicts. Nevertheless, given that the OFCCP indicates it will apply this broad definition in audits, a contractor should include disclaimers in its AAPs and other documentation to the effect that the contractor’s AAP job groups and pay grade groups do not mean that these groups should be considered as similar for the purposes of compensation discrimination analysis, he recommended.

What factors must be included in a MRA? Pursuant to the U.S. Supreme Court’s 1986 decision in Bazemore v Friday and other federal court precedent, a proper MRA requires a plaintiff (i.e. employee/applicant, OFCCP or EEOC) to analyze all the “major” factors which affect pay, Fox instructed. Thus, MRAs that fail to analyze all of the major factors affecting pay are a legal nullity. For example, if only four of the five major factors which affect pay are used in the regression, that MRA is useless for proving (or disproving) unlawful compensation discrimination.

“Major” factor.” Accordingly, a point of contention in OFCCP compensation audits may be  whether a factor which affects pay is a “major” factor. So, what constitutes a “major” factor?  The term is not well defined in applicable federal case law, Fox said. The courts generally define the term as any “influential” variable affecting pay. Thus, to be “major,” the factor affecting pay has to make a noticeable difference, one way or the other, to the pay in question.

Consistent application. A contractor may have a policy to apply legitimate, non-discriminatory factors such as skill, education, work experience, position, level of function, tenure in a position and performance ratings in setting compensation, but if that contractor does not apply such factors consistently, the OFCCP may consider them to be “tainted” by discrimination, Bulger noted. “Companies get tripped up here by bad record keeping,” he said.

Pay systems. Further, the relevant major factors must be applied in terms of each pay system examined, and this is a factual question, where one size does not fit all, the attorneys explained. It is a “huge mistake” for a contractor to think it has only one pay system, Fox warned. Any contractor that allows its supervisors discretion to adjust pay has more than one pay system, he pointed out. Likewise, Bulger noted the “wunderkind” scenario where an executive will bring in extraordinary talent at a higher pay rate than an “ordinary” hire would receive.

Digitizing data. Both attorneys pointed out that few federal contractors are in a position to do a true multi-variant regression analysis because their data is not complete, accurate, or sufficiently digitized. Only about one out of 100 federal contractors have data sufficiently digitized to undertake a MRA of their compensation systems, Fox estimated.

Most employers need to clean up their data so that it is free of things such as typos and spaces that don’t belong, and they need to update their data as to employee name changes and other developments that may have occurred since an employee was first hired, Fox observed.

Furthermore, the majority of contractors have not digitized many of the majors factors that affect pay, he points out. He listed some examples of major pay factors including: job title, related education, time in related service, related training, time in grade, location of work, work shift, performance, premium pay, such as overtime, merger or acquisition, productivity. “How many of you have digitized job history or prior job level?” Fox asked the audience.

Still, Fox points out that even if a contractor’s data were digitized, typically, 95 percent of that  contractor’s work force is probably not even susceptible to MRA for two reasons. First, in most companies, there are not many jobs of 30 or more similarly situated employees that will allow for a statistically significant analysis. Second, the pay of unionized employees, which is driven by seniority, is legally immune from the laws regarding pay equity.

OFCCP compensation analysis in audits. There is not any regulatory obligation to digitize pay data, Fox advised, and thus, the OFCCP cannot force a contractor to digitize its data. Accordingly, contractors cannot be sanctioned if they advise the agency that they do not have all the major factors that affect pay digitized. Nevertheless, the OFCCP may still conduct an incomplete analysis using only a few factors.

Moreover, even though the applicable law requires the OFCCP to analyze discrete pay decisions, the agency wrongly focuses on current pay, both attorneys pointed out. This flawed focus is also one of several deficiencies with the EEOC’s finalized changes that will add the collection of summary pay data to the  EEO-1 Report, Fox noted. For further insight into those changes, he referred to the audience to a recent Employment Law Daily story on this topic at https://shar.es/1EkPEp, which includes his commentary, among that of other experts.

Vendors. If a vendor sells you a big ticket MRA, and you do not have your data sufficiently clean or digitized for a proper analysis—i.e. one that includes all major factors that affect pay—that analysis will be useless, Fox cautioned the audience. A computerized MRA will only be as good as the data put into the program; “junk in, junk out,” he stated.

A regression analyses itself is relatively inexpensive, ranging from about $10,000 to $20,000, Fox said, explaining that the expensive part of the process is cleansing and digitizing the raw HR data. Contractors should not pay a law firm, HR vendor, or labor economist hundreds of dollars an hour to do the tasks of data cleanup and digitization that can be performed by HR personnel at significantly less cost, he cautioned.

Importantly, Fox points out that actual compensation disparities due to discrimination are rare in federal contractor workplaces, as evidenced by the OFCCP’s own enforcement data. Most contractors should ignore the “siren calls” of some vendors trying to sell MRA products and services because most contractors do not have a discriminatory compensation problem to fix and most are not yet in a position to properly analyze compensation even if they wanted to do so, he said.

Actions to consider. Given that most contractors are not ready to conduct a MRA, and there is no legal requirement to do so, what might contractors pro-actively do aside from the very minimal requirements of 41 CFR §60–2.17(b)(3)? Fox suggests contractors consider the following actions:

(1) Conduct cohort analyses of selected similarly situated employees in jobs where HR has some concern that the pay disparities there could be problematic. HR often has a “gut feeling” about which supervisors merit concern in how they make pay decisions, he observed. Keep in mind that because cohort analysis is not required by OFCCP regulations, it should be done under the protections of attorney-client privilege (as explained above).

(2) If you have decided to eventually undertake a MRA, set as an annual “good faith effort” a project to catalogue all the “major” factors which affect pay at least in those job titles with 30 or more employees that have five or more men or women and/or employees of color. Fox points out that such a project will probably take at least five years for most contractors.

(3) Set as an “action-oriented” program under an Executive Order 11246 AAP a project to clean-up HR compensation data over the next several years.

(4) Another “action-oriented” program could be a project to digitize HR compensation data in those, likely few, jobs susceptible to regression analysis. This project will take about 3-5 years for most contractors, he said.

NELI’s Thirty-Fourth Annual Affirmative Action Briefing was held in Chicago on October 20-21, 2016. For more information on NELI, including its publications and future programs, call (303) 861-5600 or go to NELI’s website at: www.neli.org.