Art by J.CiardielloOne of the things we are learning, as we strive to understand what “retirement policy” looks like in the 21st century, is that not everyone’s retirement income needs and capacities are the same. A variety of factors—different for every worker—affect that calculation.

Consider “family.” A worker/couple who has children will, on the one hand, have competing demands for available income—education and child raising expenses versus retirement savings—limiting her/their capacity to save. The USDA estimates the cost of raising one child at $233,610. That will put a strain on anyone’s ability to save, say, 10% to 15% of income for retirement. Think about what that strain is if you’re raising three or four children. On the other hand, even in an age in which personal autonomy is one of our highest values, the existence of family—and, particularly, working children—can provide a vital Plan B: in the worst case, you can move in with your kids. Lots of people do it. Obviously, workers/savers without children/family don’t have child-raising costs, but they also don’t have that emergency Plan B. They really need to save.

Also, some workers may have more outside resources than others: e.g., some people own their own homes; some don’t. Some workers have more flexibility than others: some people may, at retirement, be able to move to a low-cost of living community, dramatically reducing their income needs. For others, that is not an option. Some workers may have health issues that affect what they will need in retirement—on the one hand, (perhaps) larger health expenses, on the other (perhaps) a shortened life expectancy.

And, finally, some humans may simply have different views of time and life. Some may value time in their younger years more than time in their older years. I realize this notion is a controversial: conventional wisdom is that humans “hyperbolically discount” future values. But even accepting that theory, there is room for differing preferences. If you really like mountain climbing, you’re probably going to want to do it while you’re younger. For others, the goal of total financial independence and (perhaps) an early retirement is a more important value.

One of the features of 401(k) plans is that they allow different workers to make different choices about saving (and spending) levels. One worker can save a lot. Another can save nothing. This flexibility is one of the two major reasons why 401(k) plans have, for most sponsors and workers, replaced defined benefit plans, with their rigid, one size fits all design. (The other major reason is 401(k) plans’ greater transparency.)

Nevertheless, the consensus—with which I agree—is that humans have a natural bias towards spending rather than saving and thus should be encouraged to save. I’m prepared to accept some level of coercion to achieve that goal—e.g., the current levels of Social Security taxation/benefits. Beyond that, I favor non-coercive encouragement, the most effective and efficient version of which is (again, in my opinion) a “nudge”—defaulting individuals into some rough version of adequate savings, allowing those with a strong preference for a different (and conceivably lower) savings rate to affirmatively elect out of the default.

And so we come to the issue of “leakage.” I find this term problematic because it covers two very different issues that (I think) should be addressed in opposite ways. Generally, retirement savings “leakage” is understood to include (1) loans, (2) hardship withdrawals and (3) cashouts at termination of employment. Loans and hardship withdrawals are, I believe, a 401(k) plan feature, not a bug. They are part of the flexibility these plans provide, allowing cash-strained participants to take a risk on making plan contributions, knowing that if they need to access their savings they can get it.

Cashouts are another matter. They are, in effect, a “loophole” in our system of defaults. We default workers into saving (via automatic enrollment), but we don’t default them into leaving their savings in the plan when they terminate. That is actually a huge problem. As the Employee Benefit Research Institute has shown, “cashouts at job change have a much more serious impact on 401(k) savings than either plan loan defaults or hardship withdrawals. … 20% of those in the lowest-income quartile who would otherwise have enough savings to achieve a real replacement rate threshold of 80% would fall short due to these cashouts at job change.”

Efforts to address this issue, e.g., by encouraging sponsors to allow terminating employees to leave their money in their “old” employer’s plan and/or the development of a clearinghouse that can be used to implement a money-follows-the-employee regime, represent one of the most important retirement policy initiatives confronting us at this time. And, I would add a rule that mandated default to one of those two alternatives—money-stays-in-the-plan or money-follows-the-employee.

I spent all that space, at the beginning of this article, discussing the importance of flexibility in the 401(k) system, to demonstrate the importance to that system of choice within a context that encourages responsible retirement savings. We have learned—from 40 years of experience—that defaults do the best job of respecting both of those principles.

Thus we have made—through the use of defaults—a lot of progress towards getting employees to save enough (via automatic enrollment) and to invest their savings efficiently (through default investment in target-date funds). We need a robust default solution at payout.

Michael Barry is president of the Plan Advisory Services Group, a consulting group that helps financial services­ corporations with the regulatory issues facing their plan sponsor clients. He has 40 years’ experience in the benefits field, in law and consulting firms, and blogs regularly http://moneyvstime.com/ about retirement plan and policy issues.