WASHINGTON — Employers should seriously consider conducting a pay audit — and soon, experts suggested at the National Employment Law Institute's 42nd annual Employment Law Update in Washington, D.C.

As employers make headlines for closing their race and gender pay gaps — and others find themselves defending equal pay claims — it's only a matter of time before more face such compensation questions, Zina Deldar, an associate at Paul Hastings and Quenton Wright, a principal at Charles Rivers Associates, told attendees. "We are in a new age in which there is intense pressure on companies, both internally and externally, to address pay gaps," Deldar said.

And even if an employer does end up facing a claim, the fact that you conducted an audit can sometimes help. Massachusetts' equal pay law, for example, incentivizes employers to look at pay proactively, they noted.

Higher-ups will have to be ready to take action based on the results of an audit, so you may have to hold off until they're prepared to do that, Deldar said, "but hopefully you can convince them that the time is now."

Make a plan

Once you've decided to review your pay practices, you'll need to answer a few initial questions, the speakers said. Among those:

• Will the audit be conducted in a way that ensures it will be privileged should litigation arise? Set up a call with a compensation subject matter expert, an economist and counsel to keep it privileged.
• What data do you have? Talk with a person familiar with the data. Pull a snapshot on each employee that includes age, job information, productivity information and more. Check it and learn about it.
• What positions will you be comparing? Look at your job descriptions and only compare positions that have a common core of tasks. Additionally, will you exclude anyone? Union employees? Sales employees who have differing regions or quotas? Executives? Expats who are paid a premium to go abroad?
• What compensation will you study? Only base salary? Or will bonuses or something else be involved? The key, Wright said, is determining what poses the most risk to your company and analyzing that.

It's also important to understand your overall pay philosophy and how your processes affect pay rates. If you use a seniority system or a merit system, for example, your analysis needs to account for that. But you also need to be sure that you have the data to support your process.

If you’ve eschewed formal performance reviews in favor of frequent, informal check-ins, for example, you may not have any concrete ratings available. Deldar cautioned: "If you are a pay-for-performance company, how are you going to justify that if that’s your rating system?"

Likewise, if you say you use prior salary, be sure you actually have that data on hand. (Deldar noted, however, that salary history is a risky factor to use, in light of recent bans on the question.) The same goes for education and prior experience. Case law says these are legitimate factors for setting pay, but you would need to have that information documented and be confident that it's updated every time an employee's education changes.

And while past experience may make sense as a predictor of performance for new hires, be sure it really makes sense for long-term employees. Every merit review cycle, you should be making sure you're setting pay based on actual performance, if you're a pay-for-performance company, Deldar said.

If you don't have this data or it's not reliable, that might be another privileged project you want to undertake because, as the saying goes, "garbage in, garbage out," Deldar said. Do you really want to ask leadership for remediation based on bad information? Additionally, even if your analysis is privileged, your data can be discoverable in litigation, the speakers said. It's better to have it captured and clean ahead of time.

Interpretation and mitigation

When you're conducting your audit, there are two questions you're trying to answer, Wright said. First, if there is a pay disparity, what's the magnitude of the difference? In other words, do we have a concern? And then, is the disparity statistically significant? Small differences are normal; that’s just standard deviation, Wright explained; but when they're large and widespread, they're less likely to have occurred by chance.

Importantly, data can’t tell us causation, Wright said; it only shows correlation. So if you have a statistically significant difference, you'll want to try to determine whether discrimination occurred, or whether something else happened. Perhaps a control factor was missing, or maybe a group of people was a poor fit for the model.

And when you get to mitigation, your actions will depend on how your company thinks about risk, the speakers said. Are you going to remediate to zero, so that there's no differential between your populations? Or are you just going to bring it down below statistical significance? The courts have not addressed whether this is ok, but "I think it is something they will address," Deldar said.

Likewise, you'll have to decide which groups to mitigate. Some choose to give pay adjustments only to the lowest-paid women, for example; some opt to spread the remediation budget out among all women employees.

Also, consider your timing and communication carefully. Some employers choose to build adjustments into the next review process to avoid raising red flags. Others are already under so much fire about pay that the adjustment can only help. And when it's time to make that adjustment, get leadership to make a statement, Deldar recommended. But be careful that they don't share information that could waive the privilege; have counsel review their comments ahead of time.

Finally, mitigation is not a bandage that you can slap on and ignore, Wright said. You need to look at the policies and procedures that got you here. Deldar agreed: You need to be looking under the hood, she said. "If you don't, you're setting yourself up for a problem down the road."

For example, maybe your pay equity problem wasn't underpaid women, but overpaid men. Most employers aren't going to reduce employees' pay, but you can build in a process to prevent that from happening in the future. "Ensure you're learning," said Wright. "Make sure you're not going to have to do this again next year."