Last week, the comment period closed on the EEOC’s proposed revision of the Employer Information Report. Also known as EEO-1, the Employer Information Report records the demographic information of employees. The EEOC’s revision seeks to expand data collection to include compensation and hours worked. Opponents of the revisions claim that the collection is not specific enough yet invasive of workers’ privacy, but for those concerned about improving ethnic and gender diversity, this expansion is a welcome move.
Background and proposed changes
Under Section 709(c) of Title VII, employers are required to collect and report data in order to determine whether discrimination has occurred in the workplace. Since 1966, EEO-1 has been the designated form for the Joint Reporting Committee, composed of the EEOC and OFCCP.
The EEOC uses EEO-1 data as evidence in its civil rights actions, while the OFCCP uses it to determine which employers to target for compliance reviews. EEO-1 data remains otherwise confidential under section 709(e) of Title VII, though aggregate data is provided to examine broad trends in diversity, and a number of firms have chosen to make their individual forms public.
In 2010, the President’s Equal Pay Task Force, consisting of the EEOC, DOJ, DOL and OPM, recommended increased collection of salary data, particularly from federal contractors. A subsequent National Academy of Sciences survey, commissioned by the EEOC, suggested the potential of using the EEO-1 for collecting salary data.
The EEO-1 has undergone changes in the past. In 2007, the EEO-1 divided the “Officials and Managers” category into executive and midlevel categories, in addition to creating separate race/ethnicity categories for “Hispanic,” “Asian,” “Pacific Islander” and “two or more races.” There are two key changes in the proposed revision: the addition of pay data and the addition of hours worked.
Specifically, where Section D of the EEO-1 previously consisted of one table that counted employees by race, gender, and job category (e.g., number of white female technicians), it now would consist of two tables. The first table would ask for the number of employees by race, gender, job category, and annual salary. The second table would ask for the number of hours, again by race, gender, job category, and annual salary category. Salary data would be based on W-2s. Only employers with more than 100 employees would be required to enter salary and hours information, unlike the current form, which is filled by employers with more than 100 employees and federal contractors with 50 to 99 employees.
The EEOC states that if implemented, these changes would allow the EEOC to more effectively and proactively find potential discrimination.
Is opposition to the proposal justified?
Testimony regarding the proposed regulations occurred on March 16. While only the written testimonies are currently available, opponents of the proposal claim that:
- How data is aggregated in the proposed EEO-1 is insufficient enough to pinpoint industry/employer discrimination;
- There are privacy concerns with the data collection as a whole;
- In light of concerns (1) and (2), data collection is not worth the “millions of hours” additional compliance will take.
I address these concerns below.
- There are legitimate concerns with data aggregation, but the revisions are an improvement over the status quo
With regard to data aggregation, there are two main concerns. First, critics claim that because W-2 data captures other payments besides wages, it would be difficult to uncover wage discrimination. According to this argument, wages best measure equal pay for equal work because bonuses are offered on the basis of differences in performance. Thus, the current proposal might show discrimination where there is none – differences in performance would lead to differences in bonuses and in turn, overall W-2 pay.
Yet, as Emily Martin suggests, there could be significant disparities in bonuses which could be due to factors other than pure merit, such as implicit bias. As for an appropriate wage metric, counting the number of hours allows a rough estimation of how much workers are being paid overall per hour, which is useful information in any case.
The second concern related to data aggregation is that the proposed salary categories may not account for the types of jobs within each industry. That is, what may look like wage discrimination in an industry may simply be, as in one hypothetical offered by the Chamber of Commerce’s expert witness, due to differences in other factors between men and women. For example, if a company has a pay scale based on work experience, and male employees have on average more experience than women in one particular job and salary category, then an analysis would show statistically significant gender differences in pay even without discrimination. But while this concern is somewhat founded, it is not enough to justify not collecting the data at all. The salary categories might lead to some false positives, but altogether more data is more likely to reduce false positives and negatives compared to the existing EEO-1.
- Privacy concerns will be amplified, but the existing proposal is better than other alternatives
It is true that at small businesses, a minority or female worker’s identity – and pay – may be more easily identifiable. Michael Eastman of the Equal Employment Advisory Council provides an example: “[. . . I]f there is one Hispanic Female First / Mid-Level Official and Manager in a facility, her salary range, as well as her race and ethnicity, would be easily identified under the proposed reporting structure.” This is certainly a problem, given our existing discomfort with discussing money.
However, the scope of the information that is being revealed is not as invasive of privacy as opponents claim. EEO-1 data remains confidential by law, and under the existing EEO-1, the race and ethnicity of the Hispanic Female First / Mid-Level Official and Manager in the above example would have already been revealed because she was the only person with that demographic background and role. Additionally, salary disclosure is also not as invasive as suggested: it is still encapsulated within a number range. Moreover, sites like GlassDoor offer similar information about employee salary ranges.
To that end, claims that the consequences flowing from a potential data breach of the updated EEO-1 data would resemble that of the “OPM hack” are questionable. A hack on the EEOC (rather than on the employer) would only reveal what would be on the EEO-1 form, and again, the EEO-1 form does not collect data at the individual level or confidential information like an employee’s name.
Finally, opponents are using (1) and (2) as a type of catch-22. That is, their argument is that only individualized data can provide ideal evidence of pay discrimination because potential differences may depend on individualized circumstances. But individualized data would obviously amplify privacy concerns even further and be politically unfeasible. In fact, the proposed data aggregation appears to have been chosen based on concerns about data privacy. While privacy is a legitimate concern, the easiest way to make the EEO-1 less invasive for workers would be to have more women and people of color in managerial roles such that they are less individually identifiable.
- In light of concerns (1) and (2), data collection remains worthwhile
It is true that EEOC’s revisions are not perfect. For instance, not all jobs require employees to count how many hours they work each day, so one outstanding concern (for which the EEOC is soliciting comments) is how hours will be measured and collected.
However, in addition to the concrete benefits of data collection – such as a better understanding of the structural causes of gender and race wage inequality, like differences in experience – there is symbolic value to collecting pay data, and that may justify increased compliance. While one should be skeptical of the claim that individual employees could use the aggregate data as a negotiating tool given the problems with aggregated data discussed earlier, increased data collection is likely to create a social norm for pay transparency and for pay equality, in the spirit of the Lilly Ledbetter Fair Pay Act.