401(k) Fee Review: Why the Percentage Hides the Dollars
The 401(k) fee increase nobody voted on
Ask a retirement plan committee what their plan costs and you'll get a percentage. Twenty-two basis points. Forty-five all-in. Whatever the last benchmarking report said.
Right instinct, wrong unit.
A percentage tells you how your plan compares to a survey. It doesn't tell you what came out of participant accounts last year, who received it, or what happens to that number the next time markets have a good run.
That last question is the one worth sitting with.
The arithmetic
Take a hypothetical plan. $80 million in assets, 700 participants, paying 10 basis points for advisory services and 12 for recordkeeping. In dollars, that's $80,000 and $96,000 a year. Roughly $251 per participant in plan-level fees, before a dollar of investment expense.
Five years later the plan holds $120 million. Contributions came in, markets cooperated, headcount is unchanged. The services are identical. Same quarterly meetings, same enrollment support, same statements, same compliance testing.
Advisory is now $120,000. Recordkeeping is $144,000. Per-participant cost has gone from $251 to $377. A 50% increase for the same work.
The percentage never moved. The 408(b)(2) disclosure never changed. No committee vote was taken. Nothing in the file explains why participants are paying half again as much.
This illustration is hypothetical, does not reflect any specific plan or investment, and assumes no change in services or headcount.
Why it matters
This isn't a scandal. It's what happens when the price of a service is indexed to something the service has nothing to do with. Asset-based pricing is common, it's legal, and in some plans it's the right answer.
But it's a fiduciary event that produced no fiduciary record. And that's the part worth fixing.
ERISA doesn't require the cheapest plan. It requires that plan assets be used for the exclusive purpose of providing benefits and paying reasonable expenses of administration. Reasonable is measured against services received. And it's evaluated by looking at how the committee reached its conclusion, not just where the number landed.
The disclosure rules were built to make that possible. The Department of Labor's service provider fee disclosure regulation under ERISA §408(b)(2) took effect July 1, 2012, requiring covered service providers to describe their services, their fiduciary status, and their direct and indirect compensation.
That gives you inputs. It doesn't give you a conclusion. A disclosure will tell you a recordkeeper receives 12 basis points plus indirect compensation from three funds on the menu. It won't tell you whether that's reasonable for a plan your size, or whether the arrangement still makes sense two market cycles after it was signed.
The benchmark won't settle it either
An often-cited public dataset on plan costs makes this point itself.
Released in March of 2026, the ICI/ISS MI Defined Contribution Plan Profile report states that In 2023, the average 401(k) plan in MarketPro Retirement powered by BrightScope had a total plan cost of 0.74% of assets, while the average participant was in a plan with a total plan cost of 0.48% of assets and the average dollar was in a plan with a total plan cost of 0.30% of assets.
Three numbers, one dataset, same year. Which one is "the average" depends entirely on whether you weight by plan, by participant, or by dollar. A committee comparing itself to the wrong one can conclude it's a bargain or a laggard on the strength of a methodology footnote.
The report's authors are explicit that the material describes marketplace fees generally and isn't intended for benchmarking any specific plan against its broad averages.
A benchmark is a sanity check on a conclusion you've already reasoned your way to. It isn't the reasoning.
Four numbers to ask for
You don't need a new advisor to answer these. You need the documents your plan already has, converted into dollars.
- What did the plan pay last year, in dollars? All four layers: direct fees, investment expense ratios, indirect compensation, and any recordkeeper proprietary fund economics.
- What is that per participant?
- Who received each dollar, and for what service?
- What happens to each fee if assets grow 40% and headcount doesn't?
If your committee can answer all four in writing, you have a defensible fee file. If it can only produce a percentage, you have a benchmarking report.
Where we land
RBP generally favors flat-dollar and per-head fee arrangements over asset-based pricing, because the cost of servicing a plan is driven by participant count and scope of work rather than account balances.
That's a view, not a rule. Asset-based pricing is administratively simple, and some plan sponsors prefer that fees flex down automatically in a drawdown. One structure doesn't win.
The point is that structure is a decision your committee should have made on the record. In most plans we review, it was inherited rather than chosen.
If you need help understanding your plan costs, schedule time to meet virtually with RBP.