For years the OFCCP has placed a heavy emphasis on the challenging arena of systemic compensation discrimination. OFCCP analyzes compensation discrimination under the same theories it uses in the analysis of hiring discrimination. This would work better if compensation discrimination, like hiring, grew out a set of decisions made in a discreet window of time for all potential parties. For example, a hiring case is usually about all of the hiring for vacant positions in a particular job group or title that were made either during the AAP year or during the two year period of potential liability. The pool of applicants during that period represent the pool of people who are to be examined and compared in the analysis of discrimination.
Compensation discrimination takes a snapshot of the pay of everyone during a discreet period of time but the actual pay decisions that created that snapshot may have occurred over an extensive period of time and may have been influenced by a variety of factors including market factors, economic realities in the company, individual work performance, and individual pay deciders at the company. The snapshot may also have been influenced by pay decisions that occurred before the employee ever joined the company if starting salaries are impacted by prior salaries. These decisions may not have been made in the same window of time as the potential violation since each paycheck affected by a discriminatory pay decision restarts the clock for purposes of remedying a discriminatory compensation disparity. In addition, it may not have been a single pay decision that resulted in the disparity but rather the cumulative impact of a series of pay decisions. Further complicating matters is the fact that when the pay decisions were made there would likely have been a different pool of comparators than when the OFCCP takes the snapshot perhaps years later. These differences have presented numerous challenges to effective enforcement. Therefore, despite the OFCCP’s perennial focus on the issue, systemic compensation discrimination has remained a tough nut to crack.
The imperfect fit of the hiring discrimination paradigm is evident from the very outset of the enforcement process, the setting of benchmarks or triggers to identify situations that warrant further investigation. In hiring, these thresholds are set based on the recognized standard of statistical significance which is 2 or more standard deviations from the normal expected outcomes of a selection process or, less and less frequently, the 80% rule which sets the trigger when the selection rate of the disfavored group is less than 80% of the selection rate of the favored group.
OFCCP’s trigger for compensation disparities warranting further investigation does not rest on as firm a foundation. OFCCP has used as a benchmark average pay differences of 2% or $2,000 between comparative job groups. This benchmark was intentionally set so as to capture most Federal contractors since the agency had no better benchmark. The thought was that by setting the benchmark low, the agency would begin to collect enough data to help it figure out where a more precise benchmark should be. This is also the impetus behind the proposal to create a compensation tool. At present there is no screen that convincingly separates the potentially non-compliant from the likely compliant contractor. What this means for contractors is that virtually any contractor could be subject to extensive follow-up investigations in the area of compensation.
Contractors who trigger the threshold may be sent a letter requesting an additional 11 to 17 items of data. Regional Offices are given some discretion in who gets the follow up request. Lower level, non-union positions with sufficient numbers of favored and disfavored incumbents are most suited to the enforcement tools available to the OFCCP. It is easier to show that these positions are similarly situated, the absence of a collective bargaining agreement eliminates the terms of such an agreement as a defense, and the diversity of the incumbent pool assures that there are sufficient favored and disfavored employees to increase the reliability of the statistical evidence. More complex, non-union, upper level positions are harder to prove using statistics, especially if the numbers of comparators is low but they are inviting targets for potential cohorts. Unionized jobs are the least likely to be pursued as compensation cases provided that the collective bargaining agreement controls the rates of pay.
The follow up request for data provides the information that the Regions send to the National Office Division of Statistical Analysis (DSA) to be used in mini-regression analyses. Commonly, after the mini-regression, the Region is told that DSA did not identify any indicators from the data sent and suggests that the Region see if it can put together any SSEGs, meaning similarly situated employee groups, that can be used for comparison of compensation. Constructing SSEGs is not an exact science and there is no policy guidance or court case that fully describes what characteristics must be present for a group of employees to be considered a similarly situated employee group.
As a practical matter, employees who have the same title have a higher likelihood of being viewed by OFCCP as a similarly situated employee group. It is important for the regulated contractor to understand why the agency has grouped a particular set of employees together for the purposes of pay comparison and to examine whether the grouping is warranted. Some contractors argue that none of their employees are similarly situated with respect to compensation analysis. The agency is not receptive to this argument. The obligation to audit and monitor presupposes that the contractor has some means of auditing and monitoring, and if you say none of your employees can be properly compared for purposes of pay, from the agency perspective, this sounds exactly like an admission that you are not monitoring or auditing for compensation discrimination.
It is important for the contractor to remember that many of the federal officials who will be evaluating your pay policies and practices have either never worked in the private sector or have not for a long time. The pay system of the Federal Government is very different from private sector pay. Federal pay scales are published. Federal employees at the same grade, step, title, and agency are usually comparable for purposes of pay. Federal employees are used to a very cut and dry pay system. The compliance officers and the careerists who supervise them have worked for years in this pay system. In order to begin to help them understand how a private sector pay system works, the Federal contractor will have to be very clear when explaining their systems and not take for granted that either the compliance officer or their management truly understands how your pay system works. It is vital that you give them a clear understanding of your pay system and practices in order to eliminate the kind of assumptions they will naturally bring to the investigation of your very different pay system and in order for them to properly examine the issue of pay discrimination. This may not be as easy as it sounds because both parties are familiar with similar pay lingo but not necessarily with the all of the factors that actually go into your pay decisions.
In order to provide the Federal Government with a clear understanding of your pay system and practices and in order for you to properly monitor it to eliminate discrimination, you will need to understand your pay system and how it actually operates. It is not enough to know how it is supposed to operate, you must understand how it actually operates in practice.
While OFCCP has stated that it will use a variety of methods in its enforcement “arsenal”, the regression analysis is still the most relied on approach to analyzing systemic compensation discrimination. The regression analysis is intended to account for the major factors that influence pay and identify any remaining disparity after accounting for all known major factors. If there is a disparity in pay that remains unexplained, the agency will attribute that disparity to discrimination. Thus, you need to know what factors really make a difference in how pay is determined in your company. If an important factor is not included in the regression, it may show a significant unexplained pay disparity because the factor driving pay was not in the analysis. Further, you need to test your assumptions about what drives pay by running your own regression analyses if you have the capacity to do so. This will help you understand better what really is driving the differences in pay.
In addition to common factors such as length of service and educational level, it may be helpful to look at the pay history of the individuals being included for comparison. It is possible that the differences in pay have existed because one employee came in at a higher salary either because the market was different when they were hired or because the circumstances that brought them into the position were manifestly different than the circumstances at work in the pay decisions of others in the apparently similar job. Hopefully, these circumstances have been documented so that you can show that these were the legitimate non-discriminatory reasons that account for the observed differences in pay.
It is not uncommon for factors such as redefining the proper group for comparison, or corrected data submissions, to eliminate the apparent statistical significance of the job group or job title analysis. When this happens the agency might continue to look for cohorts. Cohorts are groups of employees for comparison which are too small to produce a reliable statistical indicator of discrimination. One thing to remember in this situation is that the loss of statistical significance removes the inference of discrimination necessary to a pattern and practice case and removes the level of disparity required for a disparate impact case. This leaves only a standard disparate treatment argument upon which to rest any allegation of compensation discrimination. The question then becomes who had the requisite intent that must be present in an individual disparate treatment case? When and by whom was the discriminatory pay decision made? It is easy for the agency to forget that the inference it usually enjoys because of the statistical significance of the indicator is no longer there. OFCCP is accustomed to this inference because the overwhelming majority of its discrimination findings rely on it. It may need to be reminded that when the statistical inference is gone, this element of the case must be proved some other way. With the statistic gone, it becomes important to know if the pay decisions for the remaining few employees who are being offered for comparison were made by the same decision maker. It becomes important to identify who had an opportunity to set the pay for each of these individuals and set them differently on a prohibited basis. It may be tempting to simply settle at this point because the dollar remedy is now so low; however, there is no reason to settle if the government has not and cannot prove its case. If you are able to explain this clearly to the investigator and his or her superiors, it is possible that the agency will drop the allegation altogether. You have to explain it in a way that makes it easy for the field investigators to explain it to the powers that be in easily understandable terms.
If the statistics hold and the regression analysis continues to result in a large unexplained disparity in pay, the agency will most likely charge you with a violation and it will be your expert against theirs. The agency avoids the use of its in house experts in litigation so you will likely be dealing with an expert secured specifically for the purpose of litigation should the case get that far.
I have been asked how a contractor should go about correcting pay disparities if it finds them before the agency does. I cannot give you legal advice on this question but I do suggest that it be done very carefully and with guidance from your law department or outside counsel. There is some risk that pay adjustments for one group may lead to discrimination allegations from members of the favored group who happen to pay comparable to the pay level being remedied for the affected group. It may be less risky, though I cannot guarantee it, to have a system in place that monitors and corrects unexplained differences in pay regardless of membership in traditionally disfavored groups. This would correct for overlooked pay increases that should have been made under the terms of your pay policy as well as any disparities that may have their genesis in some form of discrimination.
Compensation discrimination will continue to be an emphasis at the OFCCP for the foreseeable future. Even if the compensation tool is delayed, it is likely that the agency will fully utilize the powers it already has to request as much data as possible from contractors in the hope of building the long sought after proof pattern for systemic compensation discrimination cases and to devise a meaningful benchmark for compensation screening.
Trying to build your compensation compliance system by guessing at the agency’s next move is likely to be frustrating. Instead, I suggest that you build a system that is manageable and will at least pick up on gross disparities in pay. If you do not have one, it may be helpful to have an articulated pay policy that lends some uniformity to common situations such as the percentage of pay increase to be associated with particular performance levels, or whether or not starting pay is negotiable and whether that fact is to be communicated to all candidates. Some specificity is more useful than a generalized, lofty commitment to pay equity. You should also review your incumbents and decide who it makes sense to group together if you really wanted to be sure that people are being paid fairly and equitably. If there are no current comparators, you should look at others who have held the position perhaps in the past to see how their pay compares to the pay of the current incumbent. Identify the major factors that you actually believe drive pay and see if they really do by running an appropriate statistical test. Educate those who manage the pay system to recognize outliers and bring them to the attention of your EEO staff and managers. Document major events that may have impacted the pay of a large group of people such as transitioning and integrating an acquired company into your pay system, or unusually high salaries based on the demands of the market for particular skill sets, or retained pay for formerly higher graded employees resulting from placing them in a lower position during a downsizing. Having a coherent system in place that a reasonably intelligent person can understand will go a long way toward keeping you out of trouble. Focus, in other words, on what will tell you something meaningful about whether your pay system is discriminatory or otherwise unfair. You may not review the exact things the agency will review, but until the agency has landed more consistently on an approach to systemic compensation, your efforts will serve a meaningful purpose for you and may, in fact, improve your chances of ensuring non-discriminatory pay and of explaining yourself should you be subject to a compliance evaluation.