Federal supply and service (non-construction) contractors that have 50 or more employees and that hold a single contract valued at $50,000 or more are required to develop and implement a written affirmative action program (AAP) for women and people of color (POC) at each establishment (or function, for those with a functional AAP agreement).

Construction contractors, whether they hold direct federal construction contracts or federally-assisted construction contracts, are subject to the regulations implementing Executive Order 11246 Affirmative Action Program Requirements for Construction Contractors.

All employees must be included in an AAP whether or not they actually perform work on a qualifying federal contract or subcontract. New contractors (and establishments or functions that come on-line during the performance of a qualifying federal contract) have 120 days to develop and implement AAPs. AAPs must be maintained for no more than 12 months (referred to as the AAP “cycle” or “year”) after which the AAPs must be updated if the organization is still subject to the AAP requirements.

The AAP requirements for women and POC apply to supply and service (non-construction) contractor organizations that employ a total of 50 or more people (across the entire organization, not just at the establishment(s) performing contract work) and:

• Has a single contract valued at $50,000 or more; or
• Has government bills of lading which in any 12-month period totals, or can reasonably expected to total, $50,000 or more; or
• Serves as a depository of government funds in any amount; or
• Is a financial institution which is an issuing and paying agent for U.S. savings bonds and savings notes in any amount.

Note that unlike the threshold for the EO clause, OFCCP generally cannot aggregate contracts to meet the dollar threshold for AAP requirements, with two exceptions. Bills of lading can be aggregated over a 12-month period to meet the threshold. And if OFCCP can demonstrate that the organization purposefully structured its contracting to avoid the AAP threshold, the agency could still assert jurisdiction based on aggregate contract values over a 12-month period.

Note also that the term, “contractor,” includes subcontractors of any tier.  For a subcontract to qualify, it must be necessary for the performance of either the prime contract, or another qualifying subcontract (only first-tier subcontractors are required to file EEO-1 reports if they don’t otherwise have to).  Moreover, the term, “subcontract,” includes purchase orders.

Contractors are required to develop and implement AAPs no later than 120 days after becoming subject to these AAP provisions. For new contractors, this is likely calculated from the date that the qualifying federal contract or subcontract is executed.

For existing contractors, this provision applies to newly created or acquired facilities. Here, contractors have some latitude to determine when a merger or acquisition is “effective” such that AAP(s) can be developed, at which point the 120-day “clock” starts to run.

Note that AAPs run on 12 month “cycles.” Contractors must develop a new AAP no more than 12 months after the previous AAP for as long as the organization is subject to the AAP requirements. AAP cycles may be shorter than 12 months, but no longer (this is how contractors can “re-set” their AAP cycles to begin on a different start date).

Contractors subject to the AAP requirements must develop and maintain a written affirmative action plan outlining the program for each of their “establishments” (or functions, for those with a functional AAP agreement). Although OFCCP does not define the term, “establishment,” the agency has historically adopted the definition for the purpose of EEO-1 reporting:

"Establishment" is generally a single physical location where business is conducted or where services or industrial operations are performed (e.g., factory, mill, store, hotel, movie theater, mine, farm, airline terminal, sales office, warehouse, or central administrative office).

In other words, for OFCCP presumes each mailing address is a separate AAP establishment, and establishments with 50 or more employees must have an AAP, so start by identifying your 50+ locations and then decide what to do with the rest according to the exceptions detailed below.

Each employee in the contractor's workforce must be included in an affirmative action program, regardless of whether or not the employee performs work on a qualifying contract.

For establishment-based AAPs, each employee must be included in the affirmative action program of the establishment at which they physically work, except that:

1. Employees who work at establishments other than that of the manager to whom they report, must be included in the affirmative action program of their manager.
   
2. Employees who work at an establishment where the contractor employs fewer than 50 employees, may be included under any of the following three options: In an affirmative action program which covers just that establishment; in the affirmative action program which covers the location of the personnel function which supports the establishment; or, in the affirmative action program which covers the location of the official to whom they report.
3. Employees for whom selection decisions are made at a higher-level establishment within the organization must be included in the affirmative action program of the establishment where the selection decision is made.
4. If a contractor wishes to establish an affirmative action program other than by establishment, the contractor may reach agreement with OFCCP on the development and use of affirmative action programs based on functional or business units. The Deputy Assistant Secretary, or his or her designee, must approve such agreements. Agreements allowing the use of functional or business unit affirmative action programs cannot be construed to limit or restrict how the OFCCP structures its compliance evaluations.

For remote, work-at-home, or employees who work at locations not controlled by the employer (such as client sites), typically OFCCP expects contractors to apply exceptions 1 or 3 above to “map” these employees to an appropriate AAP location.

Contractors must develop an AAP for every establishment with 50 or more employees who physically work at the location, prior to the application of exceptions #1 and #3 (but including any remote employees that “report in” to the location). For the “under-50” locations, exception #2 provides contractors the option of developing “small” AAPs, “rolling up” the entire location into a related AAP, or “rolling up” individual employees based on their reporting structure. Historically, audits of under-50 AAPs do not go beyond the scope of a compliance check [ADD LINK] because the headcounts are usually too small for meaningful statistical analysis.

Note that strict application of exceptions #1 and #3 could result in employees that should technically be included in multiple AAPs for analysis purposes, which is not what OFCCP expects. Follow either the line of management (exception #1) or the line of selection decisions (exception #3, if different) and be consistent. If the affirmative action program is made “stronger” by applying exception #1 in some instances, and exception #3 in others, be prepared to explain the rationale to OFCCP in an audit.

If employees are included in an affirmative action program for an establishment other than the one in which the employees are located (they are “reporting in”), the Organizational Profile and Job Group Analysis reports of the affirmative action program in which the employees are included must be “annotated” to identify the actual work location of such employees. If the establishment at which the employees actually work maintains an affirmative action program (they are “reporting out”), the Organizational Profile and Job Group Analysis reports of that program must also be annotated to identify the program in which the employees are included.

“Annotated” simply means that the report somehow provides notice of the employees who report “in to” or “out” of the AAP, indicating either the physical location they came from or the AAP they are reporting to, respectively. OFCCP does not prescribe a particular form for these annotations. See BCGi’s sample Workforce Analysis and Job Group Analysis reports for one example of how to annotate these reports.

An affirmative action program must include the following quantitative analyses:

• Organizational Profile
• Job Group Analysis
• Determining Availability
• Comparing Incumbency to Availability
• Placement Goals

An AAP is not complete without the following components:

• Designation of Responsibility for Implementation
• Identification of Problem Areas
• Action-Oriented Programs
• Periodic Internal Audits

There are a few "hidden" affirmative action obligations that are relatively difficult to suss out of the OFCCP's regulations but are critically important. Specifically, federal contractors are required to evaluate their employment selection processes in accordance with the Uniform Guidelines on Employee Selection Procedures (external link).

Essentially, the obligations under the Uniform Guidelines can be broken down into three main things:

1. Solicit sex and race/ethnicity self-identification from applicants and employees;
2. Perform statistical analyses of employment selection processes comparing the selection rates of various protected groups to reveal potential discrimination; and
3. Perform statistical analyses of employment tests comparing the passing rates of various protected groups and validate the tests if necessary.

Contractors are also required to evaluate their "compensation systems" for potential discrimination, typically through statistical analyses of compensation data.

Employers that are subject to EEO-1 reporting requirements are already obligated to solicit sex and race/ethnicity self-identification from employees for reporting purposes. In order to perform the ststistical tests required under the Uniform Guidelines, federal contractors subject to the Executive Order (E.O.) 11246 equal opportunity clause requirements are also required to solicit self-identification from job applicants.

As a technical matter, contractors are not required to solicit self-identification from all job applicants, just those who meet the OFCCP's definition of an "Internet Applicant," because only Internet Applicants must be included in analyses of the selection processes. However, most contractors include self-identification in the standard application process for all job applicants, primarly because it is easier logistically and requires fewer resources.

Note that the OFCCP's regulations specify outdated race/ethnicity categories, but the agency does allow contractors to use the same race/ethnicity categories and definitions as used for EEO-1 reporting purposes.

Note also that federal requirements currently provide just two options for sex: female or male. Although several state and local jurisdictions may now require or allow employers to solicit self-identification in non-binary gender options, the federal government does not.

Further, there is no "unknown" sex or race/ethnicity category on the EEO-1 or for OFCCP reporting purposes, so employers are required to fill-in these designations for any and all employees who decline to self-identify, including new hires regardless of whether or not they successfully on-boarded. Longstanding EEOC guidance instructs employers to do this by "visual" inspection (an educated guess by a manager or supervisor based on the person's appearance, which is inherently problematic) or other available information (without providing examples).

If job applicants decline to self-identify, they simply fall out of the analysis for affected protected groups. So, for example, if an applicant self-identifies their sex but not their race/ethnicity, they will appear in the analyses comparing the selection rates of men and women, but will not be in the pool for any race/ethnicity comparisons. And that is fine (well, it is what it is). But if such an applicant is selected, as noted above the employer has an obligation to count them in one of the prescribed categories, and their applicant record should be updated with this information so that a selected applicant does not fall out of the analysis pool.

Accordingly, many contractors will include a notice in self-identification forms used for employees letting them know that, while self-identification is still voluntary, if they choose not to self-identify the identification will be made for them by the company as required by federal law. The idea is that if an individual knows they are leaving their designation up to the company they are more likely to self-identify. Such notice is not typically included in self-identification forms for applicants because employers are not required to identify the sex and/or race/ethnicity of all applicants.

The Uniform Guidelines require contractors to perform statistical analyses of various employment selection processes such as hiring, promotion, and termination processes. Within those braod categories, which are not the only types of selection processes subject to the Uniform Guidelines, there may be multiple "flavors" of selection processes. For example, the selection process for entry-level data entry positions is likely to be significantly different that the process for selecting senior executives. Though not explicit, the Uniform Guidelines anticipate "apples-to-apples" comparisons to the extent possible.

Moreover, any one instance of an employment selection process (such as a single requisition) may not involve sufficient numbers for meaningful statistical analysis. Often, contractors will group several instances of the same or substantially similar selection processes together for the purpose of these analyses. As a result, most contractors (and the OFCCP) typically start by analyzing all of the individual instances of a particular type of selection process in the previous AAP cycle (year) together.

The analyses themselves involve comparing the selection rates of men and women, and by race/ethnicity, and identifying differences in selection rates that are "significant." This involves identifying the "pool" of people who could have been selected and the people who actually were selected. The OFCCP provides the "Internet Applicant" rule to help determine who should and should not be in those pools, particularly with regard to hiring. If an applicant did not meet the basic qualifications for the job, were not actually "considered" for a particular position, or withdrew themselves from consideration, they could not have been selected and therefore can and should be excluded from the analysis pools.

The Uniform Guidelines, written decades ago, prescribes the "80% test" under which if the selection rate of one group is less than 80% of the selection rate of the other group, it is considered to be "significant." Since it was published, federal courts have recognized more stringent statistical tests designed to determine the point at which the difference in selection rates can no longer reasonably be attributed to random chance and are therefore "significant."

The most common tests involve either calculating the “standard deviation” or an “exact binomial.” You are most likely familiar with standard deviation (also referred to as a chi-square test) and the common maxim that 2 or more standard deviations are considered statistically significant (technically, the threshold for significance is 1.96). However, a standard deviation test is only considered useful when the sample size is at least 30 people and there are at least 5 expected selections. For smaller groups, contractors will typically employ an exact binomial test.

There is a mistaken belief that an exact binomial calculation is only appropriate for small sample sizes. This is simply not true. The calculations are more complicated and require more computing power, so when organizations were doing these calculations on mainframe computers and those resources were expensive and limited, a “simpler” test like a standard deviation was preferred when the results were as accurate and meaningful as the more processor-heavy exact tests. But today exact binomial and similar calculations are available in common, off-the-shelf database and spreadsheet applications such as Microsoft Excel and Access and can be used exclusively.

The Uniform Guidelines require contractors to compare the selection rates among the various race/ethnicity groups and provides a particular method for doing so.

First, the analyses are limited to race/ethnicity groups that comprise at least 2% of the aggregate analysis population. For example, if there is a total of 1,000 people in the analysis pool, each race/ethnicity group must have at least 20 applicants to make it into the analysis.

Then, among the race/ethnicity groups that comprise at least 2% of the overall analysis population, you must identify the "favored group" for each analysis. This is the race/ethnicity group with the most favorable (highest) selection rate by percentage.

Finally, you compare the selection rate of the "favored group" to all other race/ethnicity groups that also comprise at least 2% of the overall analysis population.

There is some question as to whether or not "two or more races" is a race/ethnicity group for these purposes, or simply a reporting convention for EEO-1 purposes. BCGi recommends treating the "two or more" category as a legitimate race/ethnicity group. However, if that is the "favored group," consider also running a set of analyses identifying the race/ethnicity group with the next highest selection rate as the favored group, particularly when the "two or more" group is just over the 2% threshold.

Note that the OFCCP still runs analyses comparing Whites to all POC in the aggregate, even though the Uniform Guidelines have no provision for such a grouping. BCGi recommends that contractors also perform analyses of POC in the aggregate in order to see what the OFCCP would see in an audit (and because it is absolutely possible for hiring managers to have a preference for Whites over all people of color that could be masked by strictly following the Uniform Guidelines). In these analyses, all groups are included whether or not they individually comprise 2% or more of the overall analysis population.

When multiple instances of a selection process are grouped together for analysis purposes, the analysis acts as if everyone in each selection pool is in one big pool and could have been selected for any of the specific selection instances. In the hiring context, imagine you are grouping together ten requisitions; the analysis treats the applicants from those ten requisitions as if they all applied to a single requisition, which was not the case. If the demographics by sex and/or race/ethnicity are significantly different from requisition to requisition, that can skew the initial analysis. One way to address this is through a "weighted" analysis in which the demographics of each requisition is weighted in accordance with the proportion of overall selections made from it. So the demographics of a requisition that resulted in a single hire will be weighted 1/10 of a requisition that resulted in ten hires.

Be mindful that AAP job groups are created solely for the purpose of determining availability and may or may not be appropriate groupings for these analyses. Often, AAP job groups will contain multiple positions that, while similar, may nonetheless have significantly different selection criteria. In such instances, an initial analysis by AAP job group might be useful to identify areas for further investigation through refined analyses, but it can also "mask" potential issues that do not show up in the aggregate. When designing your own statistical analyses, be mindful of grouping "apples with apples" and whether and to what extent the group structure should take into account different functions, departments, decision-makers, etc.

Under the Uniform Guidelines, if the overall selection process returns statistically significant differences in selection rates, contractors must then attempt to identify the particular step or steps in the process that represent a barrier to the particular protected group.

For instance, if the typical hiring process involves an application review, phone screen, skills test, manager review, then a manager interview, each of those are a "step" in the process that should be analyzed in the same way as the overall selection process.  For each step, you must identify the "pool" of potential candidates and the "selections," those who advanced to the next step.

Step analyses rely on your ability to identify the candidates through each step. Today, that is most often accomplished through the "applicant disposition codes" maintained in your online applicant tracking system. The applicant disposition codes should ideally capture both the reason someone fell out of the process and the step at which that occurred. Because there can be so many step and reason combinations, ideally you will have disposition codes for each.

Regardless, the completeness, accuracy, and reliability of your applicant disposition codes is critical. If you end up having to re-create the process from applications, resumes, notes, etc., you are in for an amazingly time- and resource-intensive process that may not yield a satisfying or meaningful result.

BCGi is currently developing more robust materials regarding impact ratio analyses, so check back soon and watch for release announcements on our Affirmative Action News blog.

If a selection process involves an employment test, such as a skills test, those tests are subject to the UGESP as well and must be analyzed separately from the overall selection process.

Here, the "pool" is everyone who took the test, regardless of whether or not they are ultimately an "Internet Applicant." If the test results indicate a statistically significant difference in selection rates, the test must be "validated."

In a nut shell, that means that the employer has conducted a proper "validity study" showing that the test is job-related and a business necessity.

BCGi is currently developing more robust materials regarding test validation, so check back soon and watch for release announcements on our Affirmative Action News blog.

"Pay equity" is currently a hot topic and, for federal contractors, the OFCCP requirement to evaluate "compensation systems" for potential discrimination is one way to get a look into an organization's pay practices. The OFCCP's regulations require contractors to evaluate pay for potential discrimination, but are silent on how contractors should go about it.

The OFCCP's other guidance documents are not much more helpful. The agency's current "compensation directive" (external link) does not provide much of a roadmap for contractors that want to rest assured that their compensation analyses are "compliant."

The OFCCP analyzes compensation through a Title VII "lens" in which the agency is required to identify a difference in pay between groups based on protected characteristics, and identify the discriminatory policy, procedure, or decision that resulted in an impermissible diference in pay. Over the years, the OFCCP has employed a number of strategies for analyzing compensation information provided by federal contractors, to varying degrees of success and utility. Most famously, the agency once employed a "threshold test" under which differences in pay that amounted to less than a certain percentage and/or a certain dollar amount (most often 2% or $2,000) were ignored. Today, however, the agency has been known to investigate very small differences in pay and contractors should not rely on the old threshold tests.

Recently the OFCCP has begun to also conduct analyses contemplated under the federal Equal Pay Act of 1963 which prohibits pay disparities based on sex. But the OFCCP does not enforce the Equal Pay Act; enforcement is handled by the Equal Employment Opportunity Commission (EEOC). Nor does the OFCCP have any parallel provisions in the agency's regulations. What the OFCCP does have is a "memorandum of understanding" with the EEOC under which the two agencies can share information with one another in investigations. So, if the OFCCP finds a potential Equal Pay Act violation, the agency can simply hand-off that portion of the investigation to the EEOC.

There are also myriad state and local laws and regulations regarding pay equity to contend with, each with their own particular "flavor" in terms of scope, analysis methods, etc.

How you structure your compensation analyses is up to you, and much like with IRAs, the onus is on you to share with the OFCCP your methodology (and, if that methodology is reasonable, the OFCCP should follow it).

BCGi is currently developing more detailed Resources pages regarding the various compensation analysis challenges faced by federal contractors and non-contractors alike. In the meantime, many organizations rely on third-party experts to advise on and perform necessary compensation analyses for federal contract compliance purposes and beyond.

BCGi is currently developing more robust materials regarding compensation analyses, so check back soon and watch for release announcements on our Affirmative Action News blog.