401(k)s Could Work Much Better

I have to participate on a panel that looks at the current state of the U.S. defined contribution system.  This event forces me to reassess my views on 401(k) plans.  I start with the premise that the move away from defined benefit plans to 401(k)s in the private sector is not reversible.  Employers do not want a pension system where they bear all the risk in terms of investment and longevity.   Moreover, while final earnings defined benefit plans are great for employees who spend their entire career with a single employer, they produce haphazard income for employees who shift jobs frequently. 

Moreover, defined contribution plans in the private sector help balance the overall risks to the retirement system.  In the pay-as-you-go defined benefit Social Security system, the risks are primarily demographic.  Lower fertility and increased life expectancy explain why Social Security costs will rise sharply.  In contrast, in a funded defined contribution system, the main risk – which is borne by employees – comes from fluctuations in asset returns.   Longevity risk remains, but it is also borne by the employee rather than the plan sponsor.  Given that the two pension arrangements are exposed to different risks, it makes sense to have defined contribution plans as a component of the nation’s retirement system.

Thus, the problem is not that the U.S. ended up with defined contribution plans in the private sector but rather that 401(k)s are the most extreme individualistic form of a defined contribution plan.  People have to decide whether to join the plan, how much to contribute, how to invest those contributions, whether to roll over accumulations when they shift jobs, how much to invest in company stock and how to withdraw money at retirement. 

People make serious mistakes at each step, producing low coverage rates and small balances.  Building on the work of behavioral economists, Congress passed the Pension Protection Act of 2006 to make 401(k)s easier and more automatic.  The legislation encouraged automatic enrollment and automatic escalation in the default contribution rates and made target date funds an acceptable default investment. 

The problem is that less than 50 percent of firms have adopted automatic enrollment and only about a third of those have automatic escalation in the default contribution rate.  The other problem is that fees are very high.  The result is only 80 percent of those eligible participate and balances are modest.  The median balances for households approaching retirement (age 55-64) are only $120,000 – $575 per month if the household purchases a joint-and-survivor annuity.  Some argue not to worry about low participation and balances because the system is relatively new, coming into existence in the early 1980s.  My view is that the “new” argument is getting a little old. 

Participation and balance questions are only part of the story.  Many of those approaching retirement will be reliant entirely on 401(k) assets for income in retirement.  Yet, most will have no idea how to draw down their assets to balance the risks of running out of money too early and depriving themselves of necessities once they stop working.  Annuities would solve this problem, but people hate annuities and don’t buy them.  

The question is what to do with a large defined contribution system that is falling far short of its potential.  The obvious answer is “fix it.”  This raises both political and legal questions.  The big political issue is the fact that the employer-provided pension system is voluntary, so legislators are reluctant to make the burden too onerous.  The legal question is how many changes can be made within the 401(k) construct.

My recommendation for immediate fixes is to pass legislation with three changes.  First, make automatic enrollment and automatic escalation in the default contribution rate integral parts of the definition of a 401(k) plan, rather than leaving the choice of whether to adopt these provisions up to the individual employer.  Second, require – at least as a default – that all 401(k) investments be in the form of indexed ETFs or low cost indexed mutual funds and limit fees for rebalancing.  (My reading of the literature is that active management is not worth the additional fees.)  Third, require – again, at least as a default – automatic annuitization of some portion of the balances. 

In the longer run, my view is that an ideal defined contribution system should have extensive risk sharing, similar to that in the Dutch collective defined benefit plans.  Instead of each employee having his own account, assets would be pooled so individuals would be freed from the burden of selecting their portfolios.  Benefits would be paid as an inflation-adjusted stream of lifetime payments.  If the plan suffers losses, adjustments could be made by an increase in employee contributions, a reduction in cost-of-living adjustments, and/or lower benefit accruals for current workers.  Similarly, if returns come in better than expected, participants would share the bounty. 

Such risk-sharing arrangements may not be consistent with American culture, much less with the 401(k) system.  But making 401(k)s automatic and cheap should be an achievable goal.