John J. Brennan, chairman and CEO of Vanguard, is less trenchant than Vanguard founder Jack Bogle. But he's smart, direct and lucid. He runs an outfit that puts most money management firms to shame.

In early May he spoke to the annual conference of SABEW, the Society of American Business Editors and Writers. His subject: "The undertold 401(k) story."

It's good reading for anyone who wants a broad overview of the world of pensions, 401(k) plans, and retirement security. In it, he challenges the idea that defined benefit pensions were the proverbial "good old days" when life was safe and better for everyone. He also challenges the idea that 401(k) plans are flawed and don't cut the mustard as a retirement income tool.

He offered broad illustrations to prove his point.

That's where the trouble started.

When ERISA attorney Brooks Hamilton read the speech, he thought something was wrong. Here's a summary of what bothered him:

Mr. Brennan explained that the median age of 401(k) participants is 44. That he or she earns $59,000 a year. And that the median 401(k) balance was $24,000 at the end of 2004.

"That is a pittance to retire upon. It's enough to buy a single life annuity of less than $2,000 per year," he said.

"But the mistake that's often made by observers and commentators is to judge the adequacy of 401(k) plans based on today's median balances," Brennan continued. "Our median 401(k) participant is age 44, with another two decades or more to work before retiring. Research shows that 401(k) account balances grow as participants grow older, gain job tenure, and as their household income increases."

Mr. Brennan then asks us to imagine that the median employee will contribute at typical levels, 6 percent of pay with another 3 percent from an employer match. That's the median level for the plans administered by Vanguard.

He assumes an annualized return of 8.5 percent after costs, referring to it as a "real" return in the original speech, which is no longer available on Vanguard's website, having been replaced by a slightly modified version. And he tells us that the worker's portfolio would grow to $400,000 or $500,000 by age 65, depending on which version of the speech you got to read.

Either way, it sounds pretty nice, doesn't it?

With $24,000 a year coming from the 401(k) plan and $18,000 (in current purchasing power) coming from Social Security, that median participant would have a retirement income of $42,000 a year. "In total," Mr. Brennan concludes, "the participant would replace just over 70 percent of his or her income."

Mr. Hamilton begs to differ. He says the example isn't reassuring. "He's talking about a program that would replace about 3 percent of income a year. That just doesn't happen."

The problem, Hamilton says, is that Mr. Brennan had frozen the worker's wages at $59,000 for 20 years. At the same time, he had added 20 years of inflation as real return in his investment return assumptions. As a consequence, the worker's 401(k) plan was "successful" only if you didn't notice that the worker lost 51 percent of his purchasing power before he retired.

Had the worker's income simply risen with inflation, Hamilton figured, his final salary would have been much higher. So his 401(k) balance would replace a smaller proportion of his final income. As Mr. Hamilton said in a telephone interview: "The issue is real return. If the worker's wages don't increase with inflation, the inflation rate shouldn't be added to the investment return."

Mr. Hamilton posted his comments on www.401khelpcenter.com, a 401k watchdog website. He was soon joined by Gregory W. Kasten, a CFP at the Unified Trust Company in Lexington, Kentucky. Mr. Kasten found that the worker's total replacement rate (including Social Security) would be only 45.1 percent, not 70 percent.

Pension analysts and planners consider a 70 to 85 percent income replacement rate to be a successful retirement. (URLs for Mr. Brennan's speech and the analysis of his rebutters are below.) Query: Is this a tempest in a teapot? Is it much ado about a regrettable computation error?

Yes and no. The Employee Benefit Research Institute projects that workers who consistently participate in 401(k) plans and retire between 2030 and 2039 are likely to replace at least 84 percent of their final five-year average salary. That finding supports Mr. Brennan's optimism--- even if his example is wrong.

But the real stakes here--- what most Americans will retire to--- are huge.

So who's right?

Sadly, it isn't EBRI or John J. Brennan.

I'll explain why, using an economist's consumption smoothing tools, on Tuesday.

On the web:

Brennan speaks: The Undertold 401(k) Story (revised version)

401(k) Helpcenter: "How Successful is the Status Quo?"

Hamilton on Brennan

EBRI Issue Brief #283, July 2005