In 2017 and 2018, employers with 100 or more employees were required to participate in what was effectively a grand experiment. In addition to filing an annual EEO-1 form with the Equal Employment Opportunity Commission (EEOC), which collects certain demographic data, broken into job categories, related to their workforces, they were required to report a plethora of additional data pertaining to employees’ wages. This “Component 2” data was collected with the intent of enabling the EEOC to identify disparities and address inequities in pay based on sex and race/ethnicity.
A recent report from the panel tasked with analyzing the usefulness of the additional data the EEOC collected is concerning in what it found and what it recommends. It found that much data was gathered, but it is of little use for the intended analytical purposes. It recommends gathering massive amounts of additional data — effectively subjecting every employer to a detailed, annual or bi-annual pay equity audit — as a long-term solution.
How We Got to This Point
The initial collection of the Component 2 data was halted by the agency in 2017, with the change in presidential administrations. A subsequent lawsuit filed by the National Women’s Law Center reinstated the regulatory requirement. This led to employers belatedly reporting, beginning in 2019, Component 2 pay data for reporting years 2017 and 2018.
The EEOC then asked the National Academies of Sciences, Engineering, and Medicine to examine the quality of the data for its intended use and to provide recommendations for future data collections. A panel of the National Academies reviewed the data to determine its usefulness in three contexts:
- 1. as an initial step in EEOC’s assessment of individual charges;
- 2. to examine pay differences at the national level; and
- 3. to assist with employer self-assessment.
The panel has issued a nearly 300-page report, which essentially found that, as executed, the 2017 and 2018 Component 2 data is of limited use in meeting any of the EEOC’s three objectives.
Viability as an Assessment Tool
Limits identified by the panel in terms of the scope and range of data collected suggest it is a poor tool for initial assessment of whether to file an EEOC charge against an employer, and an equally poor tool for employers to engage in their own self-assessments. In particular, the panel noted the following flaws, which hinder pay equity analysis:
- Only wage data was collected, not overall compensation;
- Reporting is by pay bands, which is less useful than individual-level pay data, especially for discerning anomalies among small employers and in the highest and lowest paying occupations;
- Pay data is broken down into reporting for 10 job categories used by the EEOC for EEO-1 reports, but those categories are outdated and overly broad for purposes of making effective comparisons;
- Hours worked data was not helpful. It required extensive cleaning, more than was possible for the panel to complete for purposes of its report. Also the hours worked did not delineate full-time, part-time and seasonal employees, or when employees were on paid leave, all of which affect the calculation of annualized pay;
- Data was overly generalized demographically, notably with only a binary designation for gender and with delineations that combine race and ethnicity into a single list, allowing no separate designation for those of Hispanic or Latino national origin who identify as white. Also, to the extent there are additional protected characteristics (pregnancy, age, disability, and veteran status) that fall within the EEOC’s enforcement remit, no data was collected for evaluation; and
- Data did not reflect education, tenure, performance or other legitimate causes of pay differences, which the report noted “diminishes the robustness of data” for purposes of initial EEOC investigations.
The panel concluded that the data collected on the current Component 2 forms could be used to identify potential outliers, but only as an initial step to prioritize investigations and the allocation of resources. In other words, it was better than nothing, but not very helpful in actually determining whether employees are being paid disparately for discriminatory reasons.
Viability to Examine National Pay Differences
The panel also noted several flaws in the quality of the data that limit its viability for examining pay differences at the national level. Employers with fewer than 50 employees could opt out from the data reporting and many did. Other smaller employers provided a more summary version of the data, known as “Type 6” (as was permitted under the EEOC’s regulations). Type 6 reports listed salary data based on the total number of employees, without specifying sex, race/ethnicity, occupation or pay bands. The panel found the resulting product so unhelpful that it excluded all “Type 6” data from its analysis.
Further flaws included numerous reporting errors, so much so that the panel excluded from its analysis more than one-third of the data provided. In addition, the proportion of reporting establishments for 2017 that could not be aligned to reporting establishments for 2018 was statistically much greater than Census Bureau data on the proportion of establishments that had opened or closed in that time period. The panel concluded part of the problem were inconsistencies in the identification numbers that establishments used on their reports, which hindered matching. Also, professional employer organizations (PEOs) that were reporting for multiple, otherwise unrelated client entities, sometimes submitted the reports under the identification codes and industry categories applicable to the PEO itself, instead of breaking down the data to align with the codes and categories for their various clients.
The panel concluded that, once reviewed and cleaned up (which it repeatedly referenced as a necessary step), the reported data could be used to estimate raw pay gaps at the national level by sex, race/ethnicity and job category. Here, too, the data collected thus fell short of achieving desired objectives.
The panel proffered a series of recommendations for improvements that could be made in the short-term to maximize the efficacy of the data being collected. These recommendations included:
- Better outreach to enhance compliance;
- Use of statistical weighting of data for analysis and reporting on a national or sub-national basis;
- Combining the components of the EEO-1 reporting into a single data-collection instrument with a standard reporting period;
- Carefully reviewing and cleaning the data before assessment;
- Eliminating reporting for employers with less than 50 employees, but continuing to require multi-establishment firms to file a consolidated firm-level report that includes entities with fewer than 50 employees;
- Requiring PEOs to submit data separately for each firm they represent, using the client firm’s industry code;
- Allowing a method for employers to download and review responses before submission for quality control;
- Collecting W-2 box 5 total compensation data, instead of box 1 wage data;
- Adopting narrower pay bands, with more pay bands for top earners; and
- Allowing a demographic category for individuals with more than one race and finding measures that recognize gender as non-binary.
Before any revisions are made to the form, the panel cautioned that field tests be conducted to assess the burden, data availability and questionnaire design.
Creating More Robust Data Collection
The panel suggested that, in the longer term, the EEOC can more effectively achieve its stated objectives if it reconsiders its current approach to data collection and implements substantial changes to its measures. Rather than directing employers to aggregate their employee data into what it described as “legacy” aggregated job categories, the panel recommended a series of changes that would essentially amount to a pay equity audit of every covered employer.
This would be achieved through four key changes. First, job categories would be delineated using the more detailed Standard Occupational Classification system for classifying occupations. Second, individualized data would be distinguished based on status:
- exempt and non-exempt;
- part-time and full-time; and
- seasonal or year-round,
with hours worked collected only for non-exempt employees. Third, employers would be required to report individual-level data relevant to pay disparity analyses, including:
- job experience; and
Fourth, the range of demographic data would be expanded to include age, disability and veteran status.
Tackling Pay Inequities
The panel is correct that, were the EEOC to adopt most or all of the panel’s four recommended actions, it likely would have the data necessary to achieve its objectives of identifying which employers to target for enforcement actions and could examine pay differences on a national level. But at what cost?
The panel theorized that the administrative burden of more individualized data reporting might prove to be less than that posed by the current EEOC job categories, because the EEOC’s categories do not align to any other government or employer reporting system. While potentially saving employers one step, though, the panel’s suggestions would add numerous additional data fields, many of which are not currently tracked in a meaningful way by employers’ information reporting systems. Recognizing that possibility, the panel suggested that bi-annual reporting might suffice and it stressed that any additional data reporting requirements should be field tested before they are adopted.
Beyond the administrative burden to employers, it is questionable whether the EEOC remotely has the capacity to massage all that raw data into a meaningful analysis. Having lots of information is not helpful if you are not able to pull it together and extrapolate from it.
Most significant is the data privacy concern. How do we as a society feel about providing the federal government with individualized data on the total compensation of each employee, together with their demographic data, performance ratings, skills and experience, tenure and other factors? Even anonymized, the data gets to a level where some individuals will be identifiable.
More alarming still, the panel’s final recommendation was that, while protecting for confidentiality, the EEOC should strengthen its data sharing with the public and other government agencies. So employers would not only be entrusting all this data to the EEOC, but should expect it would be shared with other government agencies, advocacy organizations of various sorts, and the general public.
Final Worrisome Thoughts
Currently pending before the Office of Federal Contract Compliance Programs (OFCCP) is a broad request under the Freedom of Information Act for all Component 2 EEO-1 reports filed by federal contractors from 2016 to 2020. As there are a great many federal contractors, the OFCCP responded to this request by posting a notice in the Federal Register on August 18 and granting employers exactly one month to object to their data submission being released. No individualized notice is being provided to potentially impacted employers, few of whom likely monitor that which is posted in the Federal Register.
The OFCCP’s handling of the currently pending FOIA request for Component 2 data does not bode well for employers or for employee data privacy, were the EEOC to broaden the pool of data that it collects. Grounded in current, demonstrated government actions, employers and employees have reason for concern.
The objective of achieving employee pay equity is laudable, but the approach of providing massive quantities of data to the EEOC for purposes of analysis and enforcement is fraught. Employers should watch for further action by the EEOC in response to the panel’s report.
By Tracey I. Levy