Why Auto-enroll 401(k)s May Reduce Retirement Savings
For the past decade, policymakers and pension experts have encouraged employers to increase worker participation in 401(k) plans by automatically enrolling their employees in these retirement programs. And it works. One study concludes that participation among new hires nearly doubles—from less than half to nearly 90 percent—when workers are auto-enrolled.
But important new research by the IMF's Mauricio Soto and Urban Institute’s Barbara Butrica finds there may be a downside: While more employees enroll thanks to the opt-out design, their employers are likely to match less of their contributions. And that might actually reduce the level of overall pension contributions for some workers. No good deed, as they say, goes unpunished.
In many ways, auto-enrollment makes a lot of sense. Because inertia is so powerful, new workers often don’t bother to fill out the paperwork to participate in 401(k)s. But if their employers sign them up, few workers take the trouble to disenroll even though they are allowed to do so. The 2006 Pension Protection Act as well as 2009 Internal Revenue Service regs are intended to further encourage employers to auto-enroll their workers.
The idea was catching on even before those policy changes. The percentage of 401(k) plans with auto enrollment boomed from just 4.2 percent in 1999 to almost one-quarter in 2006. More than 40 percent of firms with 5,000 or more employees automatically enroll their workers. Employers typically match 50 cents for every dollar their workers contribute, up to 6 percent of salary. And, according to a 2006 survey, half of firms said their main reason for offering auto enrollment was to encourage retirement savings.
But Mauricio and Barbara found that employer match rates are about 7 percentage points lower for opt-out plans. They can’t say for sure whether auto enrollment causes lower match rates. But it sure is possible. After all, if more employees participate, their employers will have to spend more to match their contributions. Whatever the cause, it seems that while auto-enrollment may increase the number of workers with 401(k)s, it won't necessarily boost their retirement savings.
I found what I used to be looking for. excellent write-up, thanks
This is the same comment that I posted at the Wall Street Journal, on 8:22 am December 28, 2009, in response to an article written there.
SINCE THEN, THERE IS A SEPARATE STUDY DONE BY THE EMPLOYEE BENEFITS RESEARCH INSTITUTE THAT SHOWS EMPLOYES WHO ADOPT AUTOMATIC ENROLLMENT DO NOT REDUCE THEIR MATCH.
“… Normally, I keep my concerns about professional studies to myself. However, I had a significant number of issues after reading the paper in terms of the author’s conclusions.
There are many other, perhaps more likely reasons for these results. In fact, I believe the results are specifically attributable not so much to employer actions (though they may be one cause), but primarily due to the limitations of the data the authors used.
Issue #1: The authors calculate an “average” match rate using 2007 Form 5500 data instead of polling employers to:
1. Split out the portion of employee contributions that are not matched, and, just as importantly,
2. Split out the portion of company contributions that are not matching contributions.
Because the study is focused solely on the likely impact to company matching contributions from adopting automatic enrollment features – the data, and the variable they construct with averaged data is a major limit on the value of the study. The authors justify using the average match because it helps to summarize the step-wise match formulas a few employers use and because it includes non-elective, non-matching contributions to pass ADP/ACP tests.
The authors defended the creation of this variable by stating ”… Nontheless, we feel that this estimate of the effective match rate (total company contribution divided by the participant contributions as reported on the 2007 Form 5500) is a better indicator of employer’s behavior — it is one of the levers that employers can control.”
Simply, averages are deceiving, they hide all manner of sins. The authors’ limited experience in designing a 401(k) plan or filling the role of a plan sponsor seems obvious – studies show that a reduction in the rate of match (to reduce company spend or to keep total company spend level over a greater participant base?), would negatively impact the perceived value of the plan among participants and non-participants alike. The change in average contributions is much more likely the result of a change in participant savings.
Issue #2: The study did not adjust for the significant number of plans in the data base that limit match to, say the 1st 6% of pay contributed. For example, today, well over half of the associates at my company contribute an amount in excess of 6% of pay, even though my company does not match contributions in excess of 6% of pay. Similarly, plans that also apply auto increase to a level that includes unmatched contributions, such as the plan at my employer, introduce other issues to the study.
Issue #3: The study ignores the effect of other plan provisions. A plan with great attributes (low cost fees, automatic increase to include non-matched contributions, flexibility, effective payout features, target date models, etc.) might attract a disproportionate amount of contributions – relative to a modest match and/or the use of automatic enrollment. That is, individuals might be more willing to contribute here, even without a match, or over and above a match, if the plan has very attractive features.
Issue #4: The authors confirm they have no way to identify profit sharing contributions (remember, 401(k) plans are qualified under the Internal Revenue Code not as pension plans but as profit sharing plans). Those company contributions may or may not be a function of the individual’s contributions. Similarly, profit sharing contributions may be more or less prevalent in plans that incorporate automatic features – I don’t know.
Issue #5: There are plans that offer a non-elective company contribution to all participants (a company contribution that is not a function of associate’s deferrals). So, if a plan adopted auto enrollment, non-elective contributions would skew the data up (because a considerable portion of the company contribution was not a function of associate contributions); and similarly, if a plan does not have auto enrollment, that would skew the study data down – the comparison group without automatic features would look better. I just don’t know the prevalence of non-elective contributions.
Issue #6: The authors also conclude that offering a defined benefit pension plan tends to lower the match rate but that the results are not statistically significant. That would be inconsistent with my 30+ years experience. A more likely conclusion is that 401k company contributions, and sometimes the match, are higher in companies where the defined benefit pension plan was frozen – in total or frozen to exclude new hires after a certain date. Consider recent Watson Wyatt (now Towers Watson) studies of the changing prevalence in defined benefit and defined contribution plan offerings in the Fortune 100, and study those companies in detail.
Issue #7: Even though the authors had specific company data, they never surveyed employers who adopted automatic enrollment and asked them what specific actions they took, if any, with respect to changing the company matching contributions coincident with or following a change to adopt automatic enrollment. They do confirm that some employers did not adopt automatic enrollment because of the cost, but the fact that some did not adopt automatic features does not lead to a conclusion that some employers who did adopt automatic features reduced their match. What might have been telling would be to survey those who did not adopt automatic enrollment but who did reduce their 401(k) match or freeze their defined benefit pension plans.
Issue #8: If the authors took total contributions and subtracted out the match, it isn’t clear how rollover amounts were treated.
Issue #9: The 2007 Form 5500 data are, for most employers, based on pre-PPA-2006 designs, or, only include a partial year using a PPA-2006 automatic design. That is, the future of automatic enrollment is likely to be much different from the past.
Thank you for the opportunity to comment.
I don't agree with two of your points, and have some issues with a third. Vesting periods are shorter these days. Inflation can erode the value of a benefit, but it can do the same for a defined contribution plan. Depending.
Plan termination is a real risk. Bankruptcy is a risk independent of termination only because of poor government regulation. Employers are supposed to fund the plan adequately.
The main problem with defined contribution plans is that it only works for strong people. If you are an Ayn Rand wannabe with a little bit of luck, they're okay. But Ayn Rand wannabes usually think that most people are chumps. I'm afraid that, for pension plans, they are empirically correct. If left to their own devices, people undercontribute.
If the employer simply shifts down the matching amount while upping the percentage — say, the first 25 cents up to 10% of the salary (presumably the default percentage of salary is 5% or similar) — they could incentivize more retirement saving while keeping their own costs lower since relatively few people will put aside the full 10%, although there is a new incentive to do so.
Defined benefit plans are great for the winners, but they are highly overrated.
Defined benefit plans sound good until you consider the risks. First, you can lose the job before being vested. Second, the plan can be terminated at any time, probably before the steep increase in benefits that occurs in the final years of work. Third, inflation can seriously erode the value of the benefit. Fourth, the company can go bankrupt, reducing or eliminating the payment. Companies rely on all 4 of these factors to allow them to promise more benefits than they plan to deliver.
Fifth, the government may not keep its promise to ensure even the capped benefit.
I'd rather trust a retirement account that's portable and that I can invest to provide some degree of inflation protection. Portability also benefits the economy as a whole by reducing job lock-in.
I would much rather be automatically enrolled in a defined benefit plan. I suspect that the accounting and advisory sector went after the “full funding” issue to get employers to shift from these plans (which do quite well on a pay-go basis) to defined contribution plans which produce commission revenue for what amounts to the other side of the house.
Getting people into index funds is not necessarily a good idea. I would much rather that people were enrolled in insured funds featuring employee ownership of the employing firm – with the insurance consisting of a third of the shares being traded to a fund holding stock in a variety of employee owned firms. Such a fund would normally not vote its shares unless prodded to by a quarter of the employee shareholding pool – in which case they could examine the issues and decide whether to throw in with the 25% and fire management.
Wide adoption of such a system of ownership would no bode well for Wall Street – however it would be a godsend for workers.
I agree it sounds like you are desperate to kick what is really very good news, more people saving for retirement. I'm not sure the folks already saving are going to notice the 7% drop, and, well, if you are so interested in picking nits, how will more people saving for retirement effect future govt expenditures (and tax rates) when the retired are on average better prepared for retirement?
I'm not sure how you are defining “their retirement savings” in that last sentence. Certainly the people who wouldn't have been enrolled at all are now going to have dramatically higher retirements savings. If you happen to be one of those individuals who keeps all their receipts and compares them against their credit card statement then working for an employer with auto-enrollment may not be in your best interest. It's hard to say from the numbers you quote, but it even seems likely that total employer contribution has gone up (7% reduction in match vs. almost 100% increase in eligibility for match). From a policy perspective, this change seems to have been very successful with only minor negative effects on the individuals who weren't targeted.
If enrollment nearly doubles and the match rate is 7% less, then overall an employee's retirement savings increase – more from the employee and slightly less from the employer. This is a GOOD thing.