Saturday, September 23, 2017

"Checks" and Balances

In about a month, the IRS will announce the new contribution and benefit limits for 2018 – and that could be good news even for those who don’t bump against those thresholds.

These are limits that are adjusted for inflation, after all – designed to help retirement savings (and benefits) keep pace with increases in the cost of living. In other words, if today you could only defer on a pre-tax basis that same $7,000 that highly compensated workers were permitted in 1986 – well, let’s just say that you’d lose a lot of purchasing power in retirement.

But since industry surveys suggest that “only” about 9%-12% currently contribute to the maximum levels, one might well wonder if raising the current limits matters. Indeed, one of the comments you hear frequently from those who want to do away with the current retirement system is that the tax incentives for 401(k)s are “upside down,” that they go primarily to those at higher income levels, who perhaps don’t need the encouragement to save. Certainly from a pure financial economics perspective, those who pay taxes at higher rates might reasonably be seen as receiving a greater benefit from the deferral of those taxes.

Drawing on the actual account balance data from the EBRI/ICI 401(k) database, and specifically focusing on workers in their 60s (broken down by tenure and salary), the nonpartisan Employee Benefit Research Institute has found that those ratios were relatively steady. In fact, those ratios are relatively flat for salaries between $30,000 and $100,000, before dropping substantially for those with salaries in excess of $100,000. (See chart.) In other words, while higher-income individuals have higher account balances, those balances are in rough proportion to their incomes – and not “upside down.”

And yet, according to Vanguard’s How America Saves 2017, only about a third of workers making more than $100,000 a year maxed out their contributions. If these limits and incentives work only to the advantage of the rich, why aren’t more maxing out?

Arguably, what keeps these potential disparities in check is the series of limits and nondiscrimination test requirements: the boundaries established by Internal Revenue Code Sections 402(g) and 415(c), combined with ADP and ACP nondiscrimination tests. Those plan constraints were, of course, specifically designed (and refined) over time to do just that – to maintain a certain parity between highly compensated and non-highly compensated workers in the benefits available from these programs. The data suggest they are having exactly that impact.

Those who look only at the external contribution dollar limits of the current tax incentives generally gloss over the reality of the benefit/contribution limits and nondiscrimination test requirements at play inside the plan – and yet surely those limits are working to “bound in” the contributions of individuals who would surely like to put more aside, if the combination of laws and limits allowed.

One need only look back to the impact that the Tax Reform Act of 1986 had on retirement plan formation following the imposition of strict and significantly lower contribution limits – as well as a dramatic reduction in the rate of cost-of-living adjustments to those limits – to appreciate the relief that came in 2001 with EGTRRA.

Anyone who has ever had a conversation with a business owner – particularly a small business owner – about establishing or maintaining a workplace retirement plan knows how important it is that those decision-makers have “skin in the game.”

While we don’t yet know what the limits for 2018 will be, gradually increasing the limits of these programs to keep pace with inflation helps assure that these programs will be retained and supported by those who, as a result, continue to have a shared interest in their success.

And that’s good news for all of us.

- Nevin E. Adams, JD

Saturday, September 16, 2017

Are You (Just) a Retirement Plan Monitor?

A recent ad campaign focuses on the distinction between identifying a problem and actually doing something about it.

In one version a so-called “dental monitor” tells a concerned patient that he has “one of the worst cavities that I’ve ever seen” before heading out to lunch, leaving that cavity unattended. Another features a “security monitor” who looks like a bank guard, but only notifies people when there is a robbery.

As an industry, we have long worried about the plight of the average retirement plan participant, who doesn’t know much (if anything) about investing, who doesn’t have time to deal with issues about their retirement investments, and who, perhaps as a result, would really just prefer that someone else take care of it.

What gets less attention — but is just as real a phenomenon — is how many plan sponsors don’t know anything about investments, don’t have time to deal with issues about their retirement plan investments, and who, perhaps as a result, would — yes, also really just prefer that someone else take care of it.

Of course, if many plan sponsors lack the expertise (or time) to prudently construct such a plan menu, one might well wonder at their acumen at choosing an advisor to do so, particularly when you consider that surveys routinely show that plan sponsors choose an advisor primarily based on the quality of the advice they provide. One can’t help but wonder how that advice is quantified (certainly not in the same way that investment funds can be), and doubtless, that helps explain why so many advisors are (apparently) hired not on what they know, but on who they know.

But for many plan fiduciaries, the obstacle to hiring a retirement plan advisor is financial, not intellectual. Particularly for a plan sponsor who has not previously employed those services — or, more ominously, in the case of one who has hired an advisor that didn’t hold up their end of the bargain — the additional costs of hiring an advisor can be problematic. The question asked of a prospective advisor may be, “Why should I hire you?” But one can well imagine that the question that is often unarticulated, and perhaps the real heart of the matter is, “Why should I pay you (that much)?”

There are ways, of course, to quantify the value of those services, ways that quantify not only what that advisor is worth, but why those fees are what they are. Some advisors promote their services as a shield against litigation, or at least some kind of buffer against the financial impact of such an event, but in my experience, while most employers are glad to get/take the “warranty” (implied or explicit), they often aren’t willing to pay very much extra for it.

In the most obvious case, you can walk in and demonstrate the ability to save a plan money by upgrades to the menu, a change in providers, or perhaps even a better negotiation of the current arrangement. That’s clearly added value, and value that is readily measured (though it has a finite shelf life). Assuming, of course, that the plan sponsor is ready, willing (and in some cases, able) to act on those recommendations.

Indeed, most of the attempts to affix a value to having an advisor tend to focus on investment returns or cost savings for the plan. Both are valid, objective measures that can have a real, substantive impact on retirement security for participants, and fiduciary peace of mind for the plan sponsor.

Similarly, the ability to increase plan levels of participation, deferral, and investment diversification also adds value — quantifiable value, particularly measured against goals and an action plan for achieving them that is clearly articulated, and updated, up front. Ultimately, it’s about more than identifying issues and problems, it’s about having a plan for the plan, and being willing to be an agent for change in pursuit of it.

The “monitor” ad closes with the admonition, “Why monitor a problem if you don’t fix it?”
Indeed.

- Nevin E. Adams, JD

See also: “The Value of Good Advice.”

Saturday, September 09, 2017

A "Real Life" Example

In addition to the books, reference guides, and a few personal “knick knacks,” I have for years had in my office a couple of model cars – but not for the reason people generally think.

These models happen to be Studebakers (a 1950 Champion, a 1953 Starliner and a 1963 Avanti). I’d wager that a majority of Americans have never even heard of a Studebaker, and the notion that a major U.S. automobile maker once operated out of South Bend, Indiana would likely come as a surprise to most. I keep them in my office not because I have an appreciation for classic cars (though I do), but because of the role the automaker played in ERISA’s formation.

Born into a wagon-making family, the Studebaker brothers (there were five of them) went from being blacksmiths in the 1850s to making parts for wagons, to making wheelbarrows (that were in great demand during the 1849 Gold Rush) to building wagons used by the Union Army during the Civil War, before turning to making cars (first electric, then gasoline) after the turn of the century. Indeed, they had a good, long run making automobiles that were generally well regarded for their quality and reliability (their finances, not so much) until a combination of factors (including, ironically, pension funding) resulted in the cessation of production at the South Bend plant on Dec. 20, 1963. Shortly thereafter Studebaker terminated its retirement plan for hourly workers, and the plan defaulted on its obligations.

At the time, the plan covered roughly 10,500 workers, 3,600 of whom had already retired and who – despite the stories you sometimes hear about Studebaker – received their full benefits when the plan was terminated. However, some 4,000 workers between the ages of 40 and 59 – didn’t. They only got about 15 cents for each dollar of benefit they had been promised, though the average age of this group of workers was 52 years with an average of 23 years of service (another 2,900 employees, who all had less than 10 years of service, received nothing).

ERISA did not create pensions, of course; they existed in significant numbers prior to 1974, as the workers at Studebaker surely knew (they certainly had reason to – Studebaker-Packard had terminated the retirement plan for employees of the former Packard Motor Car company in 1958, and they got even less than the Studebaker workers wound up with in 1964). But armed with the real life example of those Studebaker pensions, highlighting what had been a growing concern about the default risk of private sector plans (public sector programs weren’t seen as being vulnerable to the same risk at that time) – well, it may have been a decade before ERISA was to become a reality, but the example of Studebaker’s pensions provided a powerful and on-going “real life” reminder of the need for reform.

Has ERISA “worked”? Well, in signing that legislation – 43 years ago this past weekend – President Ford noted that from 1960 to 1970, private pension coverage increased from 21.2 million employees to approximately 30 million workers, while during that same period, assets of these private plans increased from $52 billion to $138 billion, acknowledging that “[i]t will not be long before such assets become the largest source of capital in our economy.” His words were prophetic; today that system has grown to exceed $17 trillion, covering more than 85 million workers in more than 700,000 plans.

The composition of the plans, like the composition of the workforce those plans cover, has changed considerably over time, as has ERISA’s original framework. Today much is made about the shortcomings in coverage and protections of the current system, the projections of multitrillion-dollar shortfalls of retirement income, the pining for the “good old days” when everyone had a pension (that never really existed for most), the reality is that ERISA – and its progeny – have unquestionably allowed more Americans to be better financially prepared for retirement than ever before.

It’s a real life example I think about every time I look at those model cars – and every time I have the opportunity to explain the story behind them.

- Nevin E. Adams, JD

Saturday, September 02, 2017

Storm "Warnings"

Watching the incredible, heart-rending coverage this past weekend of Hurricane Harvey’s devastation, I was reminded of a personal experience with nature’s fury.

It was 2011, and we had just deposited our youngest off for his first semester of college, stopped off long enough in Washington, DC to visit our daughters (both in college there at the time), and then sped home up the East Coast with reports of Hurricane Irene’s potential destruction and probable landfall(s) close behind. We arrived home, unloaded in record time, and rushed straight to the local hardware store to stock up for the coming storm.

We weren’t the only ones to do so, of course. And what we had most hoped to acquire (a generator) was not to be found – there, or at that moment, apparently anywhere in the state.
What made that situation all the more infuriating was that, while the prospect of a hurricane landfall near our Connecticut home was relatively rare, we’d already had one narrow miss with an earlier hurricane and had, on several prior occasions, been without power, and for extended periods. After each I had told myself that we really needed to invest in a generator – but, as human beings are inclined to do, and reasoning that I had plenty of time to do so when it was more convenient, I simply (and repeatedly) postponed taking action. Thankfully my dear wife wasn’t inclined to remind me of those opportunities, but they loomed large in my mind.

Retirement Ratings?

People often talk about the retirement crisis in this country, but like a tropical storm still well out to sea, there are widely varying assessments as to just how big it is, and – to borrow some hurricane terminology – when it will make “landfall,” and with what force.

Most of the predictions are dire, of course – and while they often rely on arguably unreliable measures like uninformed levels of confidence (or lack thereof), self-reported financials and savings averages – it’s hard to escape a pervasive sense that as a nation we’re in for some rough weather, particularly in view of objective data like coverage statistics and retirement readiness projections based on actual participant data.

Indeed, in a recent op-ed in the Wall Street Journal, Andrew Biggs, Resident Scholar at the American Enterprise Institute offers what is certainly a contrarian perspective these days: assuring President Trump (and us) that there is no retirement crisis (subscription required), at least not a “looming” one. And yet his point is basically little more than things are better than many would have us believe, based largely on data that indicates things are better now than they were before we worried about such things. Biggs, who previously testified before Congress that the use of the term “crisis” to describe the current situation was an “overblown non-solution to a non-crisis,” maintained that 75% of today’s retirees are “doing well,” and that Boomers are having a better retirement than their parents, who ostensibly lived during the “golden age” of pensions.

Another reassuring perspective was published earlier this year by the Investment Company Institute’s Peter J. Brady & Steven Bass, and Jessica Holland & Kevin Pierce of the Internal Revenue Service – who found, based on IRS tax filings, that most individuals were able to maintain their inflation‐adjusted net work‐related income after claiming Social Security. Not exactly an affirmation that the income was enough to sustain retirement expenses, but at least it showed that individuals were maintaining (and most improving upon) their pre-retirement income levels, at least for a three-year period into retirement.

Life is full of uncertainty, and events and circumstances, as often as not, happen with little if any warning. Even though hurricanes are something you can see coming a long way off, there’s always the chance that they will peter out sooner than expected, that landfall will result in a dramatic shift in course and/or intensity, or that, as with Hurricane Katrina – and apparently Harvey – the most devastating impact is what happens afterward. In theory, at least, that provides time to prepare – but, as I was reminded when Irene struck, sometimes you don’t have as much time as you think you have.1

Doubtless, a lot of retirement plan participants are going to look back at their working lives as they near the threshold of retirement, the same way I thought about that generator. They’ll likely remember the admonitions about (and their good intentions to) saving sooner, saving more, and the importance of regular, prudent reallocations of investment portfolios. Thankfully – and surely because of the hard work of advisors and plan sponsors – many will have heeded those warnings in time. But others, surely – and particularly those without access to a retirement plan at work – will find those post-retirement years (if indeed they can retire) to be a time of regret.

Those who work with individuals trying to make those preparations know that the end of our working lives inevitably hits different people at different times, and in different states of readiness.
But we all know that it’s a “landfall” for which we need to prepare while time remains to do so.

- Nevin E. Adams, JD

Footnote
  1. As it turned out, we got lucky. An apparently random and unexpected delivery of generators happened at our local hardware store where my wife had only hoped to be able to stock up on batteries. We got it home, and in record time learned enough to run it, managed to get in a supply of gasoline (before the pumps and cash registers ran out of juice), got our windows covered with plywood, and hunkered down for what still feels like the longest night of my life.

Saturday, August 19, 2017

Hypothetically Speaking

A new academic study offers some insights on taxpayer preferences for pre-tax versus Roth savings – at least in certain conditions.

The study – which carries the somewhat unwieldy title “The Relative Effects of Economic and Non-Economic Factors on Taxpayers’ Preferences Between Front-Loaded and Back-Loaded Retirement Savings Plans” – takes a look at the various factors influencing preferences for paying taxes “up front” on retirement savings (this is termed “back-loaded by the researchers, in that the tax advantages come in retirement) versus pre-tax treatment as with a 401(k).

Writ large, and pretty much across the board, the researchers – Andrew D. Cuccia, associate professor and a Grant Thornton faculty fellow at the University of Oklahoma, and Marcus M. Doxey and Shane R. Stinson, both assistant professors at the University of Alabama – found that individuals preferred Roth (back-loaded) – even in circumstances in which they thought a rational determination would favor a pre-tax option.

“Consistent with prior research, our results suggest that individuals, on average, do not respond rationally to the relative economic incentives associated with alternatively structured plans,” they wrote. And while “at least part” of the “failure to connect relative tax rates – those paid now versus those in the future – was attributed to “a lack of awareness and/or understanding,” the researchers found individuals largely reluctant to embrace the pre-tax approach, even when education specifically designed to help frame that understanding was employed. Or, as the researchers explained, “… although errors can be reduced with increased awareness, our evidence illustrates that individuals systematically incorporate non-economic factors into their retirement plan choices, often leading to a preference for [pre-tax deferrals] even when such a choice is economically adverse.”

The researchers determined that “participants do not incorporate expected tax rate changes into their plan choice without an explicit explanation of the impact tax rate changes have on relative after-tax returns,” and “even when participants were educated about the tax rate change-return relation, 49% who reported that they expected their tax rates to be lower in retirement nonetheless elected to make their contributions to a back-loaded (Roth) plan.”

Now, in fairness the research wasn’t based on actual administrative data – rather they constructed several scenarios to test responses to various factors, and ran a group of online survey respondents through those scenarios to evaluate and weigh those responses. So, it was basically asking individuals about their (hypothetical) response to a variety of conditions regarding (hypothetical) tax rates, market conditions, as well as non-economic attitudes and preferences.

While they found a general preference for the back-loaded Roth accounts, they found “mixed evidence regarding whether individuals appropriately weight expected tax rate changes in their plan choices,” even though tax rates were seen as the primary factor driving the relative after-tax returns of front- and back-loaded plans. Indeed, the certainty of knowing the tax rate that would be paid, even if paid “now” seemed to outweigh the concerns associated with the uncertainty of future tax rates, though those who did expect to pay higher taxes in the future were – as one might expect – inclined toward the back-loaded (Roth) option.

If the researchers seemed puzzled about some of the preferences for the Roth option, they also found that a sense of urgency regarding saving for retirement was “positively associated” with savings rates, and that perhaps what they saw as “the current crisis in retirement preparedness” suggested to them that “current marketing and education campaigns are not sufficiently stoking investors’ sense of urgency.”

Those of us who work with retirement plans – and retirement plan participants – might not be quite so perplexed by the notion that individuals don’t always act in their financial best interest.  Additionally, while the researchers seemed to be quite thorough in outlining (and doubtless executing) their test scenarios, it is arguably one thing in a “laboratory environment” to make a choice with someone else’s money, and perhaps something else again to make those same choices with your own retirement savings.

Still, those concerned about a negative response by participants to the imposition of a Roth choice, might find some comfort in these findings.

Hypothetically speaking, of course.

- Nevin E. Adams, JD

Saturday, August 12, 2017

Pay Me Now, or Pay Me Later

Many years ago, there was a commercial (for car oil filters, as I recall) that cautioned, “You can pay me now, or pay me later” – in other words, spend a little now on an oil filter, or pay lots later on to fix the damage done by not doing so. It’s a mantra that I’ve heard employed to encourage retirement savings – but these days it might have a new twist.

We now have a second survey of plan sponsors expressing concern about the impact that switch from the current pre-tax preferences accorded 401(k)s would have on participation.

That member survey by the Committee on Investment of Employee Benefit Assets (CIEBA) found that 78% of the 61 member respondents believed that a switch to an all-Roth system would negatively affect participation rates in their 401(k) plans. In that sense, it roughly mirrored the findings of a survey by the Plan Sponsor Council of America (PSCA) which found that more than three-quarters of the 443 employer respondents to the survey said they strongly agreed with the statement that eliminating or reducing the pre-tax benefits of 401(k) or 403(b) retirement savings plans would discourage employee savings in workplace retirement plans.1

While at least two other employer surveys are reportedly in the field and/or pending release, we (still) don’t know how participants will actually respond. However, it doesn’t require a massive leap of imagination to think that there might be a negative response of some magnitude to the federal government “taking away” the benefit of saving on a pre-tax basis that is, after all, what Section 401(k) of the Internal Revenue Code was all about.

Roth Rising?

Ironically, we find ourselves at a time when the availability of the Roth option in plans is at an all-time high, when providers like T. Rowe Price note that they have seen the biggest one-year increase in Roth contribution offerings in its clients’ 401(k) plans in 2016 – 61% of the plans for which it provides recordkeeping services – while Vanguard notes that two-thirds of the plans it recordkeeps now offer the feature, compared with 49% as recently as 2012.

Now, I realize that there is a difference between having the opportunity to contribute on a Roth basis, and having no option but to contribute on that basis. I’ve no doubt that there are individuals living paycheck-to-paycheck who would find the loss of the here-and-now tax preference to be a hardship. Individuals who, confronted with a Roth mandate, might indeed reduce their retirement savings in order to put food on the table, pay the rent, or put gas in the car so they can get to work.

Those concerns aren’t new, of course. For years they were – and in many cases still are – invoked as reasons to go slow, or go “low” on embracing automatic enrollment. Real as they may be, we also know what that means for retirement security.

Indeed, the surveys that have asked individuals about tax preferences – to the extent they are specific at all – nearly always focus on one particular aspect: deferring current taxes on contributions. The Roth advantages of not paying taxes on the accumulated earnings and the freedom from being forced to take RMDs aren’t even mentioned. Nor do most discussions about post-retirement drawdowns acknowledge that some large chunk of those retirement savings will be due Uncle Sam.

That said, it’s not as though the Roth doesn’t have its own set of tax preferences – and the closer one gets to retirement, the better they look. The odds that tax rates in retirement will be lower, particularly for younger workers, these days seems a quaint notion. Not surprisingly, Vanguard notes that nearly a third (30%) of Roth participants in Vanguard plans were in the age cohort of 34 or younger – and that’s without being defaulted in that direction.

Don’t get me wrong – like most of us, I’d rather have the choice than not. Nor would I diminish the communication challenge ahead if the long-standing 401(k) pre-tax preferences were capped or eliminated.

But of late, every time I see one of those reports about the average 401(k) account balances of those in their 60s, I can’t help but think that somewhere between 15% and 30%, and perhaps more, won’t go toward financing retirement, but will instead go to Uncle Sam and his state and municipal counterparts. And on a frequency dictated by the required minimum distribution schedules of the IRS.

And I can’t help but wonder how many plans for retirement don’t factor in that tax “cut.”

Nevin E. Adams, JD
  1.  I draw comfort from the findings in both surveys that very few employers indicate that they would discontinue or diminish their current programs if a shift, full or partial, to Roth would occur.

Saturday, July 29, 2017

Why Your Recordkeeper Might Not Be an Automatic Enrollment Fan

I recently wrote about what’s wrong with automatic enrollment. Turns out there’s more – and it has to do with when things actually go “wrong” with automatic enrollment.

See, it’s one thing to say that eligible employees should be automatically enrolled – and yet another to actually get them enrolled automatically. Even the Internal Revenue Service (IRS) goes so far as to acknowledge that two common errors found in 401(k) plans are: (1) not giving an eligible employee the opportunity to make elective contributions; and (2) failing to execute an employee’s salary deferral election.

Now, as it turns out, both are “fixable” – through the Employee Plans Compliance Resolution System (EPCRS). But that’s only the start of things. See, in both of those situations you’re looking at a corrective contribution of 50% of the missed deferral (adjusted for earnings) for the affected employee. And then fully vesting the employee in those contributions – contributions that are subject to the same restrictions on withdrawal that apply to elective deferrals.

The only difference in the correction for the two situations outlined is the calculation of the amount of the missed deferral. In the case of an erroneously excluded employee, the missed deferral is based on the average of the deferral percentages (ADPs) for other employees in the employee’s category (for example, non-highly compensated employees), whereas if the error involves failing to implement an employee’s election, the missed deferral is based on the employee’s elected deferral percentage, or in the case of missed automatic contributions, the automatic contribution percentage. For plans with automatic contributions, however, the corrective contribution for the missed deferrals is reduced to 0% or 25% of the missed deferrals, depending upon how soon the error is corrected.

Your head starting to hurt?

Hold on – those corrective contributions also need to be adjusted for earnings – from the date that the elective deferrals should have been made through the date of the corrective contribution.1 In all cases the employer must contribute any missed matching contributions, adjusted for earnings.

‘Tell’ Tales

Now, in addition to the regular array of plan notices that will now be required for those new participants, automatic enrollment has its own special set of notices. While most larger plans rely on what is called an eligible automatic contribution arrangement (EACA), smaller programs may have in place what is called a qualified automatic contribution arrangement (QACA), a type of automatic enrollment 401(k) plan that automatically passes certain kinds of annual required testing (generally referred to as a safe harbor plan). A QACA must include certain features, such as a fixed schedule of automatic employee contributions, employer contributions, a special vesting schedule and specific notice requirements.

The automatic enrollment notice details the plan’s automatic enrollment process and participant rights. The notice must specify the deferral percentage, the participant’s right to change that percentage or not to make automatic contributions, and the default investment. The participant generally must receive the initial notice at least 30 days, but not more than 90 days, before eligibility to participate in the plan or the first investment. Subject to certain conditions, the notice may be provided, and an employee may be enrolled in the plan, on the first day of work. An annual notice must be provided to participants and all eligible employees at least 30 days, but not more than 90 days, prior to the beginning of each subsequent plan year. And guess what happens if those notices don’t go out when they are supposed to?

So, it’s not as though you just have to flip a switch on payroll and you’re done.

Even When It Works

There are, of course, issues, even if there are no processing missteps. Cost, particularly as it relates to the match – which may have been designed to encourage workers to sign up, and which, with automatic enrollment, may no longer need to – and which may have been budgeted for a 70% participation rate that, thanks to automatic enrollment, may be more like 95%. Turnover can leave behind smaller 401(k) balances, which incur additional recordkeeping costs, and which can prove to be a real administrative burden with ongoing notice and communication requirements. Which again, if those notices aren’t going out when they should…

The bottom line is that automatic enrollment is an important component of helping more Americans save, and save effectively. But as is often the case, it’s not as easy as it sounds, and plan sponsors looking to embrace this design – and advisors who tout it – should do so with a full awareness and appreciation of all the implications.

For some other issues, see “Why Doesn’t Every Plan Have Automatic Enrollment?

Nevin E. Adams, JD
  1. Not that that’s not an improvement from how it used to be. Under Rev. Proc. 2015-28, if the error is detected within 9½ months after the year of the failure, no corrective qualified non-elective contribution (QNEC) is required to an affected participant’s account for the missed deferral opportunity, as long as the person is enrolled within the 9½ month period (or earlier if the affected employee notifies the employer of the mistake).