Testing Workplace Retirement Plans For Equality and Fairness
August 02, 2014
A major criticism trending in the employee benefit arena is that tax-advantage retirement plans (mostly the 401(k))-style ones) benefit higher paid workers. Progressive politicians and academics continue to propose changes to these plans. Individuals defending the current tax structure of workplace employee benefits rebuke these proposals by pointing out their technical inaccuracies. They claim that the tax features of these plans are equal and fair because they are subject to annual non-discrimination (ND) testing. This battle is not going away benefit pros, and if you conduct your own non-discrimination testing, you are right in the middle of it.
What is Non-discrimination (ND) Testing?
As you can imagine, non-discrimination testing is a complicated process that actually consists of multiple tests. However, basically, its purpose is to determine if highly paid and key employees benefit more from the plan than lower paid employees. It can be difficult to understand what these tests mean unless you perform or assist in performing them yourself. But that is not my main concern here. If non-discrimination testing is what proponents of 401(k)-style retirement plans want to use to claim they are equal and fair, let’s talk about why this may not be true for one simple reason. And what is that reason…? Well, the data used to perform the tests comes from the employer and its accuracy is questionable.
Data Troubles
I want to meet the employer willing to testify under oath that the data they gather to conduct non-discrimination testing for any of their qualified employee benefit plans is 100% accurate. I say this because I know from experience that gathering this data is not a simple process. In fact, it’s often difficult. Why? Well, it usually requires the assistance of others from the Payroll and Information Technology departments. Which means providing these folks with an understandable explanation of data needs and timetables.
This is where the problems can surface. What if their systems can’t easily supply this data and some of the data collection is manual. Say for example, you have payroll data that spans two calendar years and you have to manually remove the data you do not need. Or how about dealing with HR and Payroll data entry errors such as incorrect hire or termination dates… And let’s not even talk about the accuracy of each employee’s gross compensation and actual hours worked. Again, you would think that with today’s sophisticated human resources management systems that obtaining this data is easy and that the data is accurate. I can honestly say that I cannot imagine a ND census that is completely accurate; where there is no manipulation of numbers.
Conclusion
But maybe I am asking for too much. Does the data really need to be 100% accurate to conclude that the non-discrimination testing is fairly accurate and that the retirement plans we test are equal and fair? No, I don’t think I am asking too much when I suggest that the data needs to be 100% accurate. Employee benefit pros need to be in charge of the data they use in preparing their ND census. They need to have direct access to Payroll and/or IT systems where they can create their own reports and monitor data changes throughout the year. For now, it is hard to claim that non-discrimination testing results prove that 401(k) plans keep it equal.
How comfortable are you with the accuracy of your non-discrimination testing data census?
blog comments powered by Disqus
Denise Perkins
