It seems likely that most hands went up, understandably. A broad consensus holds that America faces a retirement crisis because people don’t save enough, and young people are the worst savers. Workers in their twenties are by far the least likely to participate in employee retirement saving plans, investing giant Vanguard reports, and in the past five years they’ve become significantly less likely to participate. They’re squandering the opportunity to let their savings compound for the longest possible time. That’s why automatic enrollment in employer-sponsored 401(k) accounts is such a good idea.

But what if that thinking is wrongheaded, and young people are behaving just the way hyperrational economic models say they should? New research by eminent experts finds that many young workers are doing exactly that. In a just-released working paper, the authors conclude that “for liquidity-constrained young adults who anticipate significant earnings growth, optimal retirement saving is zero.”

This is beyond startling. It’s sacrilege in a financial advice industry that has long counseled maximum feasible saving for retirement. But the authors of this new paper base their counterintuitive argument on the indisputable reality that there’s more to life than saving. 

A large literature in economics specifies happiness rather than savings as the quantity to be maximized. The focus of the new paper is “total life satisfaction from material consumption,” as one of its authors, Stanford University professor John Shoven, tells Fortune. He’s a longtime authority on retirement economics, Social Security, Medicare, and pensions. 

Life satisfaction is hardly a new concept in economics; it’s what economists call utility. The economic model created by Shoven and colleagues assumes income can be spent or saved, and if it’s saved it can go into a retirement account like a 401(k), with various rates of employer matching, or into a taxable investment account. The future utility of consuming today’s savings is discounted, says Shoven, for three reasons: “People display impatience; they may not be alive at distant future dates; and they may not be in good health at future dates, such as suffering from dementia.”

The model assumes a worker who was born in 1995, starts work at age 25, and retires at 67 with Social Security replacing 34% of after-tax income; it also includes required minimum distributions from the retirement account.

In a previous era, when interest rates were higher and most of the workforce consisted of high school graduates whose inflation-adjusted incomes would increase only slightly during a career, the traditional advice—starting to save for retirement at the beginning of employment—made sense under Shoven’s model.

 The conclusions begin to shift, however, when you add a couple of plausible assumptions based on today’s environment—namely, that a college graduate’s income at age 25 will be only 42% of their peak income at 45 or 50, and inflation-adjusted interest rates, a gauge of safe investment returns, are extremely low, around 0%. Under those conditions, a rational total-life-satisfaction maximizer won’t start saving for retirement until age 41, and that’s if their employer matches 50% of their 401(k) contributions; without matching, they won’t start saving until age 44.

They would rationally start saving earlier if the safe inflation-adjusted return were higher, 3%, but even then they wouldn’t start until age 37 with employer matching of 401(k) contributions and age 40 without.

These conclusions will strike many people as crazy—the opposite of rational. When rates of return are lower, shouldn’t workers start saving earlier, not later? Similarly, shouldn’t they start saving earlier if their employer isn’t matching their 401(k) contributions? But remember, the objective isn’t maximum savings; it’s maximum total life satisfaction. When investment returns are low, workers have little incentive to sacrifice today’s consumption in favor of barely greater consumption years from now. Employer matching enables workers to achieve a given level of saving while sacrificing less of their current consumption, so they’re willing to start earlier.

The authors acknowledge that real life is more complicated. People save for reasons in addition to retirement, such as buying a home or building a rainy-day fund; saving before having kids may be wise because saving will be more difficult afterward. Nonetheless, they say, “Our general argument still applies to retirement saving that occurs in an illiquid employer-sponsored pension account.”

Which brings us back to automatic enrollment in 401(k)s, the past decade’s biggest trend in retirement saving. Vanguard says half the plans for which it keeps records now require participants to opt out rather than opt in. That structure is almost universally lauded, but this new study suggests a significant caveat. The authors’ conclusion: “Automatic enrollment that applies to workers of all ages could be nudging young people to make—rather than avoid—a mistake.”