Written Testimony of Michael J. Eastman Vice President, Public Policy, Equal Employment Advisory Council

Hearing of March 16, 2016 - Public Input into the Proposed Revisions to the EEO-1 Report

Chair Yang and Commissioners Barker, Feldblum, Lipnic, and Burrows, thank you for the opportunity to participate in today's hearing on the Equal Employment Opportunity Commission's ("EEOC" or "Commission") proposal to revise the Employer Information (EEO-1) Report to include data regarding employee compensation and hours worked. My name is Michael Eastman and since 2012 I have served as the vice president of public policy for the Equal Employment Advisory Council (EEAC).

Statement of Interest

EEAC is the nation's largest nonprofit association of employers dedicated to the advancement of practical and effective programs to eliminate employment discrimination. Formed in 1976, EEAC's membership now includes approximately 260 of the nation's leading and largest employers, all of which are firmly committed to the principles and practice of workplace nondiscrimination. All of our members are employers subject to the compliance, recordkeeping, and reporting requirements imposed by federal statutes and regulations prohibiting workplace discrimination. In addition, nearly all of our members also are federal contractors subject to the additional recordkeeping, reporting, and compliance requirements imposed by Executive Order 11246, the Rehabilitation Act of 1973, and the Vietnam Era Veterans' Readjustment Assistance Act of 1974, and the Department of Labor's implementing regulations.

EEAC has a long track record of working closely with the EEOC to ensure that the EEO-1 Report maintains its relevance and utility to both the Commission and the employers who have to file it. Indeed, over the years, EEAC many times has been the only organization to submit public comments in response to the EEOC's invitations for stakeholder input on the burdens and utility of the EEO-1 Report under the federal Paperwork Reduction Act.¹ And for more than three decades, we have worked and communicated less formally with Commission staff to resolve practical concerns regarding the EEO-1 reporting process in ways that benefitted both the Commission and employers.

EEAC also has a long history of researching and providing feedback on other federal agency initiatives to collect compensation data from employers on a broader scale, including those implemented or proposed by the Department of Labor's Office of Federal Contract Compliance Programs (OFCCP). These include OFCCP's Equal Opportunity Survey, which collected summary compensation data from a sample of federal contractors during a five-year period between 2000 and 2004,² as well as OFCCP's more recent proposed Equal Pay Report.³

Overview and Context

EEAC strongly opposes compensation discrimination based on protected characteristics such as sex and race. In addition, we support efforts by the EEOC and other workplace regulators to focus their limited resources in ways that more effectively identify bad actors. However, as the experience with OFCCP's Equal Opportunity Survey demonstrated, the collection of summary compensation data did not provide any useful data to help that agency identify federal contractors that were engaging in unlawful compensation discrimination. Rather, the data collected by the Equal Opportunity Survey were in such a raw and aggregate form that they could not be used to compare similarly situated employees - one of the critical prerequisites to any meaningful analysis of compensation data for the purpose of detecting unlawful pay discrimination.

Moreover, as OFCCP itself has recognized, "[i]nvestigations of compensation discrimination are complex and nuanced" and such investigations require a "tailoring of compensation investigation and analytical procedures to the facts of the case based on Title VII principles."⁴ Together, these statements underscore the reality that aggregate compensation data are of negligible if any utility in identifying the factors that drive private-sector compensation, and even more importantly, whether those factors are legitimate or discriminatory.

Even OFCCP's contractor-level, site-specific compensation investigations have revealed remarkably few instances of race- or gender-based pay discrimination - especially considering the sheer volume of line-item compensation data that the agency has collected since it began using its revised scheduling letter in 2014. Indeed, despite OFCCP's devotion of resources to discovering compensation discrimination at the "local" level, it has largely been unable to turn up any violations. And yet, this methodology - as resource intensive and flawed as it is - still is more likely to identify compensation discrimination than EEOC's proposed approach in the form of the revised EEO-1 Report.

As the proposal acknowledges, the President's National Equal Pay Task Force recommended that the EEOC engage the National Academy of Sciences (NAS) to "conduct a study assessing how to most effectively collect pay data to support law enforcement efforts."⁵ This recommendation led to a report published by a panel convened by NAS (NAS panel) that reviewed options and made several important recommendations regarding actions that the EEOC, OFCCP, and other agencies should take before implementing a new survey to collect compensation data. Unfortunately, many of the important recommendations by the NAS panel have either been disregarded or not completely followed in the development of the EEOC's current proposal.

While we have no quarrel with the EEOC's intent to root out and eliminate pay discrimination where it exists, we respectfully submit that the EEOC's current proposal to collect pay data from employers is misguided and unlikely to produce data that will be useful either for enforcement purposes or for employer self-evaluations. Given that the proposal is unlikely to collect sufficiently useful data, the significant burdens that it will impose on employers, who will be required to report nearly three billion fields of data per year, simply cannot be justified. We therefore urge the Commission to withdraw the proposed changes.

Should the Commission nevertheless decide to move forward with its proposal, we strongly recommend that it first conduct a reliable pilot study to determine that the proposed revisions will in fact effectively accomplish the EEOC's underlying objectives before the massive new burdens associated with these revisions are imposed on employers across-the-board. In addition, we recommend that the commission not utilize W-2 wages as a measure of compensation in this reporting instrument.

As an alternative, and as we discuss in more detail going forward, using data based on employee rates of pay is likely to be more meaningful and will significantly reduce employers' compliance burden. We also recommend that the Commission permit employers to file based on calendar year data and that employers not be required to submit Component 2 of the proposed report more frequently than once every three years. In addition, we address the significant confidentiality concerns that employers might face with the proposed expansion of the EEO-1 Report to include sensitive compensation-related data. Finally, we address some of the Commission's burden estimates, which we believe grossly underestimate the actual time and cost burdens that the proposed revisions are likely to impose on employers.

EEAC remains committed to working with the EEOC as it considers whether to move forward with the proposal and, if so, what changes should be made.

A Meaningful Pilot Study Is Essential Before Proceeding

One of the most important recommendations to the EEOC and OFCCP made by the NAS panel was the need for them to conduct a pilot study before proceeding with a new compensation data collection mandate. Specifically, the NAS panel stated that the enforcement agencies "should initiate a pilot study to test the collection instrument and the plan for the use of the data. The pilot study should be conducted by an independent contractor charged with measuring the resulting data quality, fitness for use in the comprehensive plan, cost, and respondent burden."⁶

The EEOC states that it commissioned an independent pilot study to identify the most efficient means to collect pay data. However, the pilot study commissioned by the EEOC is wholly inadequate as it did not in any meaningful way analyze compensation data collected from actual employers, nor assess the utility of data collected, nor attempt to understand the actual burdens imposed on employers by the proposed collection tool.

In fact, in the 109-page Final Report prepared by Sage Computing, there is no discussion of a pilot study that used actual employer data, no discussion of a pilot study that examined the time for employers to compile and submit actual data, and no discussion of a pilot study that analyzed collected data for use in enforcement, for example, by identifying employers more likely to be engaging in systemic compensation discrimination. While the Sage report does include a section-less than one page long-addressing employer burdens, that section implies that no more than a handful of private-sector employers were contacted to gather this information, and that the limited data Sage did receive simply did not enable it to provide any meaningful estimates of the burden of the proposed changes.

Accordingly, we recommend that before moving forward with its proposal, the EEOC should seek OMB approval for a pilot study that involves a sufficient number of employers to yield accurate and meaningful estimates of the burdens and utility of the proposed revisions.

The EEOC could also conduct an analysis of actual compensation data that it already possesses through the EEO-4 Report, which provides the agency with salary information from state and local governments. We are not aware of any analysis showing that the EEO-4 Report's collection of data by pay band helps the EEOC in any way identify public sector employers engaging in systemic pay discrimination. This certainly suggests that the current proposal may not produce data sufficient to justify the costs of collection. The EEOC could conduct statistical analyses of the real data it already has from state and local governments to further evaluate the utility of expanding the EEO-1 Report in a similar way. Of course, public sector entities are more likely to establish pay structures based on rigid pay grades and steps and, in this way, are unlike most private sector employers. Consequently, tools appropriate for analysis of EEO-4 pay data may not prove effective for analysis of pay data collected through expansion of the EEO-1 Report. Nevertheless, it would provide an opportunity to test statistical models on existing data before instituting a new broad-based mandate for private-sector employers.

Annualized Compensation or Base Rates of Pay Are More Appropriate Than W-2 Wages for This Reporting Requirement

The proposal would require employers to report their establishment-level headcount and hours worked data by race/ethnicity and gender in 12 new pay bands within each of the existing ten EEO-1 job categories. Covered employers would use each employee's W-2 wages to determine the pay band in which the employee would be reported. The proposal does not define precisely how W-2 wages are to be calculated.

Collecting W-2 wages is likely to be problematic, however. First, W-2 wages are among the measures of compensation least likely to reflect discriminatory practices, given that they reflect a mixture of employer and employee decisions. They therefore are a poor choice if the Commission is seeking to maximize the utility of data it intends to collect. Second, it is far more burdensome for employers to report W-2 wages than it is for them to report annualized compensation or base rate of pay.

More specifically, W-2 wages are a poor choice for this reporting requirement because they reflect numerous components driven primarily by the choices of individual employees and other factors that have no bearing on the employer's decision as to how much an employee earns. These include, but are not limited to:

• Parking stipends;
• Mass transit stipends;
• Relocation and travel stipends;
• Military stipends;
• Expense reimbursements;
• 401(k) contributions;
• Certain insurance premiums;
• Sick pay;
• Severance payments;
• Back wages;
• Deferred compensation; and
• Profit sharing.

Reporting W-2 wages also conflicts with the "snapshot" basis of the EEO-1 Report. Whereas employers report each employee's EEO-1 category as of a certain date within the reporting window, the proposed revisions would have employers also report an entire year's worth of each employee's earnings, at least for those employees who had been with the employer for an entire 12-month period. For employees who changed EEO-1 categories during the prior year, their EEO-1 category on the snapshot date wouldn't necessarily match their prior year's earnings, skewing the aggregate data and inviting incorrect conclusions to be drawn. We submit that rates of pay or annualized salary would correct for this potential mismatch, as they would be drawn from the same snapshot date as the EEO-1 category.

Using rates of pay or annualized compensation to determine pay band placement also would reduce the burden on employers because pay rates are typically stored in the same human resources information system (HRIS) in which current EEO-1 Report data are stored, whereas wages earned data typically are stored in separate payroll systems, which will require reporting employers to retrieve, analyze, tabulate, and reconcile data from entirely separate systems at significant difficulty, burden and expense. In addition, reporting by pay rate would remove the requirement to report hours worked, which would effectively halve the burden imposed on employers by the proposed revisions.

Importantly, the NAS panel explicitly recommended collecting data on rates of pay, not actual earnings or pay bands. Reporting by pay rate is also consistent with the recommendation of the NAS panel that enforcement agencies should evaluate pay in the manner in which employers actually look at compensation. Rates of pay thus provide better data points to evaluate compensation for indicators of unlawful discrimination.

If W-2 Wages Are Collected, the Commission Should Consider Permitting Employers to File Reports Based on a Calendar Year

Employers must currently file EEO-1 Reports by September 30 each year. When doing so, employers are required to report data based on a "snapshot" taken of the employer's workforce between July 1 and September 30. The Commission's proposal maintains the current due date and three-month window during which the employer may take its "snapshot." The proposal then requires employers to report on W-2 wages and hours worked for the 12-month period prior to the snapshot date. However, this requires employers to extract W-2 wage data and hours worked over two different tax years. For many employers, it would be significantly less burdensome to report W-2 data and hours worked for a single calendar year, and the Commission should provide employers with this option if it does not accept our recommendation for using rate-of-pay information to assign employees into the appropriate pay band.

It should be noted that the Commission's current regulations establish a due date for the report, but not the reporting period,⁷> and we believe the Commission could provide for this option through this proposal. Indeed, at one time the EEO-1 Report was due on March 31 and was based on a reporting period in the first three months of the year. It was changed to September because the Commission believed the third quarter of the year was "less affected by the variation in seasonal employment, such as employment in the construction industry, … and will provide employment figures which reflect average annual employment more closely than" reporting in the first quarter.⁸ These concerns are not relevant to the reporting of annual W-2 wages, and in the interest of minimizing burdens on employers the Commission should provide employers with the option to file using calendar year data.

Of course, if the Commission were to change the reporting requirement from W-2 wages to base rate of pay or annualized compensation, the reporting period would no longer be a concern.

Confidentiality

Many employers consider the data reported on the current EEO-1 Report to be highly sensitive and proprietary. Inclusion of compensation data will only heighten this concern. In fact, the NAS panel recognized this when it observed that the EEOC shares data with agencies that do not have the same level of confidentiality protections and which are not covered by the same penalties for unauthorized disclosure that apply to EEOC employees. Accordingly, the NAS panel recommended that the EEOC seek legislation to extend Title VII's penalties to those with whom it shares data.

Additionally, the proposed revisions would greatly increase the likelihood that the compensation and race/ethnicity of individual employees would be disclosed whenever there are only a few employees in a particular job category, race/ethnicity group, and pay band at an establishment. For example, if 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. It is critically important that any new data collection effort include robust new protections to ensure the privacy of personally identifiable information as well as information which employers rightly deem to be proprietary and confidential.

The EEOC Has Seriously Underestimated the Likely Reporting Burden

The EEOC's estimate of the burdens imposed by the proposed revisions is flawed and needs to be significantly revised. Among the problems with the EEOC's estimates are the following:

• Failing to consider alternatives that are less burdensome than requiring employers to submit as many as three billion data fields of compensation related data based on a workforce of about 158 million people;
• Improperly changing the analysis of burden from the number of forms filed to the number of entities filing them without accounting for the majority of filers who do not use the data file upload method;
• Failing to properly explain the rationale for the estimate of burden hours per entity resulting in an unreasonably low burden estimate;
• Assuming the vast majority of time spent compiling and filing reports will be by "administrative support" employees earning an average of $24.23 per hour, despite the fact that the proposal will require complex analyses and system adjustments that can only be made by more highly paid professional and senior-level employees; and
• Underestimating the burdens associated with developing systems to communicate between payroll and HRIS systems or otherwise query and report payroll data.

While we anticipate filing more detailed comments on the EEOC's flawed estimates at a later stage of this proceeding, the following offer some initial reactions.

Are Three Billion Data Cells Necessary?

The Commission reports that in 2014, 1,482,810 EEO-1 Reports were filed and that 672,420 of those reports were "Type 6" summary reports. This means that employers filed 810,390 "full-grid" Reports of one type or another containing the number of employees in each job group by demographic category. Currently, the full grid includes 180 data fields.

The EEOC's proposal expands the size of the grid from 180 data fields to 3,660 data fields: 1,830 of these data fields would be used to report the number of employees in each of 12 pay bands in each job group by demographic category. An additional 1,830 data fields would be used to report the hours worked by the employees reported in each of the first 1,830 data fields. If covered employers continue to file EEO-1 Reports in the same number and in the same manner as they did in 2014, in 2017, employers will report on up to 2,966,027,400 data fields. It should be noted that this estimate does not discount the number of federal contractor filers with fewer than 100 employees who will not be required to file compensation data. While we know that in 2014 there were 6,260 such filers, we do not know the number of EEO-1 Reports that they filed.

A proposed reporting requirement that would force employers to complete or skip over nearly three billion data cells each year, when there are only approximately 158 million workers in the entire civilian non-institutionalized labor force, is simply not an effective or efficient reporting framework. It also reveals that the vast majority of these data cells (and by vast majority we mean in the billions) will be zero, even when aggregated across the entire workforce. Yet, nowhere in the EEOC's proposal does it explain alternatives that could significantly cut down on the number of unnecessary data fields.

In addition, the proposal does not address several suggestions made by the NAS panel for reducing the burden, such as less frequent collection or collecting data on a rotating sample. The EEOC should carefully evaluate these and similar suggestions for reducing burdens beginning with our recommendation that the EEOC not mandate that employers complete Component 2 of the report, which collects headcount by pay band and hours worked, more frequently than once every three years. ⁹

Burden Calculation Methodology Improperly Ignores Majority of Filers Who Use the EEOC's On-Line Portal

In prior years, the EEOC has estimated the burdens imposed by the EEO-1 Report based on an estimate on the amount of time to complete and submit each report. In the past, the Commission has estimated that these tasks took employers 3.4 hours. In its proposal, the Commission has changed its methodology. According to the proposal:

employers now rely extensively on automated HRIS to generate the information they submit on the EEO-1 Report. As a result, each additional report filed has just a marginal additional cost. To accurately reflect the manner in which employers now collect and submit the data for filing, the estimated reporting burden … is calculated per form, rather than per report. This burden calculation is based on the time spent on the tasks involved in filing the survey, rather than on "key strokes" or data entry. As such it more accurately reflects how virtually all employers actually complete the EEO-1 …¹⁰

The EEOC supports this changed methodology by observing that in 2014, 1,449 firms filed EEO-1 Report by uploading a data file, accounting for 704,654 of the EEO-1 Reports filed in that year. However, this does not account for the 478,392 "full-grid" EEO-1 Reports that were manually entered by more than 59,000 firms (almost 98%) using, primarily, the EEOC's on-line portal. Data entry and the number of key strokes are indeed essential parts of determining the true burden for this overwhelming majority of EEO-1 filers.

For these filers, the expansion of the EEO-1 Report's data grid from 180 data fields to 3,660 data fields will be particularly significant. At a minimum, they will need to tab through the data fields or search for the data fields into which their data should be reported, a process that will invariably take longer with so many more data fields to chose from.

Burden Estimates Are Not Properly Explained

As discussed above, in the past the EEOC had conducted its burden estimates per report filed. For example, in 2014, the EEOC concluded that the burden of the EEO-1 Reporting requirement was $19.8 million, based on an estimate of 307,103 reports filed and an average burden hour per report of 3.4 hours. In its proposal, the EEOC acknowledges that it is changing its methodology from a per report analysis to a per employer analysis. Astonishingly, it estimates that the burden per employer for Component 1 will be 3.4 hours, the same amount of time that the EEOC had previously estimated per report. The only explanation for this change is that "each additional report filed has just a marginal additional cost" because most employers now have automated HRIS systems to generate information needed. It should be emphasized that some employers file not just one or two EEO-1 Reports, but thousands. Assuming the per-employer cost is now the same as the per report cost from two years ago strains credulity.

The EEOC has also erred in its estimates for those employers who will be required to complete both components one and two. First, the agency estimates that it will take employers one hour to read the instructions for the new form. This estimate cannot be evaluated because no draft of the proposed instructions has been provided for notice and comment. Second, the estimate for collection, verification, and reporting data is estimated to be 5.6 hours per employer. The proposal provides no explanation of how the EEOC arrived at this estimate.

Finally, the proposal estimates that employers who must file both components will incur one-time costs of just over $23 million to develop queries related to the new compensation reporting requirements and that such queries are estimated to take 8 hours per filer at an average wage rate of $47.22 per hour. Based on feedback from our members who have attempted to determine what changes will be required to their systems, these assumptions appear entirely unrealistic.

We hope to have more accurate estimates of the burdens that will be imposed by these requirements at later stages of this process.

Conclusion

Thank you very much for the invitation to participate in today's hearing. As I stated at the outset, EEAC stands ready to work with the EEOC to come up with a collection tool that will actually produce meaningful data while not imposing a heavy and unnecessary burden on reporting employers.

---

Footnotes

¹ See, for example, the supporting documents maintained by the Office of Management and Budget related to EEOC's 2014, 2011, and 2009 information collection requests for approval of the EEO-1 Report, available at: http://www.reginfo.gov/public/do/PRAViewDocument?ref_nbr=201412-3046-001, http://www.reginfo.gov/public/do/PRAViewDocument?ref_nbr=201104-3046-003, and http://www.reginfo.gov/public/do/PRAViewDocument?ref_nbr=200901-3046-001.

² OFCCP formally repealed the EO Survey in 2006. Office of Federal Contract Compliance Programs, Final Rule, Affirmative Action and Nondiscrimination Obligations of Federal Contractors and Subcontractors, Equal Opportunity Survey, 71 Fed. Reg. 53,032 (Sept. 8, 2006).

³ Office of Federal Contract Compliance Programs, Notice of Proposed Rulemaking, Government Contractors, Requirement to Report Summary Data on Employee Classification, 79 Fed. Reg. 46,561 (Aug. 8, 2014).

⁴ Office of Federal Contract Compliance Programs, Interpretive Standards for Systemic Compensation Discrimination and Voluntary Guidelines for Self-Evaluation of Compensation Practices Under Executive Order 11246, Notice of Proposed Rescission, 76 Fed. Reg. 62, 63 (Jan. 3, 2011).

⁵ Equal Employment Opportunity Commission, Proposed Revision of the Employer Information Report (EEO-1) and Comment Request, 81 Fed. Reg. 5113, 5114 (February 1, 2016).

⁶ National Research Council, Collecting Compensation Data from Employers, at 88 (2012).

⁷ 29 C.F.R. § 1602.7.

⁸ Equal Employment Opportunity Commission, Recordkeeping and Reporting Under Title VII and the ADA, Final Rule, 56 Fed. Reg. 35,753 (July 26, 1991).

⁹ EEO-4 data is not collected annually, but every other year.

¹⁰ 81 Fed. Reg. at 5120.