@Gary,
You would complete an (Traditional, 401K, 403b, etc.) IRA to Roth IRA Conversion if the future Rate of Return (ROR) will be higher for the Roth IRA comparef to a tax deferred IRA. Article below discusses this in detail using this calculator (Optimal Retirement Planner).
From Article:

A onetime conversion of the entire IRA during the first year of retirement is technically feasible, sometimes practiced, but rarely part of an optimal plan.

Also, for strategic periodic IRA to Roth IRA conversions,

We find that in comparing two optimal plans, differing only in whether or not conversions are allowed, that there is in the neighborhood of a 1 percent improvement in the conversion plan’s disposable income compared to the non-conversion plan.

In the conclusion of article:

It would seem desirable to convert when asset prices are depressed because there is less tax paid and the state of the market is amenable to a recovery. Following the same logic, converting when asset prices are inflated would seem imprudent.

Measuring the Financial Consequences of IRA to Roth IRA Conversions

After 20 years, I'm guessing that your wife had at least $5K in the 401(k).
"Generally, if your account balance exceeds $5,000, the plan administrator must obtain your consent before making a distribution." IRS 401k guide.
While what's done is done, forcing your wife to take the money was likely illegal.
You have the option of recharacterizing the Roth conversion into a traditional IRA and not owe any taxes. You could then move the money back into the 401k (pre-tax) if that's where you really want it. However, you can't try to undo everything in one step by having the 401(k) take back the money (pre-tax) straight from the Roth IRA.

How can I recharacterize an amount rolled over to a Roth IRA from an employer-sponsored retirement plan?
You can only recharacterize amounts rolled into a Roth IRA from an employer-sponsored retirement plan by transferring them to a new or existing traditional IRA, and not back into the plan from which they were distributed.

IRS: IRA FAQs.
If you want the money back in the 401(k) as a Roth 401(k) (and if the plan offers this and allows the transfer), then you could move the money back via a trustee-to-trustee transfer.
Note: After having converted to a Roth IRA, you can withdraw the amount converted (but not subsequent earnings) without owning tax on that money (since you just paid that tax). But so long as your wife under 59.5, there will be an early distribtution penalty of 10% for the first five years after conversion.
Fairmark: Distributions After a Roth IRA Conversion

@msf,
>> Wait until Fidelity has another cash promotion
have not seen such; they have had?
All the $50 xfer fees were reimbursed by ML. Our 'benefit' total will be $1150 all told, plus the zero commissions thing for all ML trading, although I seldom do that.
Somewhere I have Fido paperwork from 1971 account, which I kept to show a rep at a center once, not that long ago. (Yawn, another geezer passing through.)
Yeah, I've wondered how the thou will be logged. I am a little more tax-sensitive in semiretirement than I used to be, since I've been following the Optimal Retirement Planner guru's deep looks at withdrawal strategies and taxes. Man, does Welch keep up. He may not be Kitces or Thomas, but what a free service he provides; unbelievable.

I'm not a tax guy so please double check my understanding. It sounded like the transfer has already happened from the 401K at your wife's employer to her Roth IRA at a mutual fund company. If that's true I'm not sure how easy it is to undo it because I think it would be considered a withdrawal from a Roth without waiting what I believe is the required 5 years and that would probably incur a penalty. If the transfer isn't done yet and you still have the option to change your mind, one advantage of transferring to a rollover IRA rather than the Roth IRA is that you don't pay any tax. Depending on your situation and what you believe/want to bet on about Trump's tax plans, you might potentially pay lower tax on a Roth conversion if you waited until a new tax code is in place.
As an addition to Bill's comments, I believe once you wait the required 5 years after converting to a Roth IRA, you're able to withdraw what you "contributed", or in other words what you've paid tax on, at any time without penalty. It's only the gains on what you contributed that you're not allowed to withdraw until retirement age without penalty.

I'd keep it as an IRA, not 401k. There are more investment choices available to you.
The only reasons to put it back in 401k are:
1) if you want early retirement, you can get at the funds without penalty a couple of years earlier.
2) if the 401k plan has access to some particular funds that you otherwise can't invest in.
Bill

I just earned a thou from ML for transferring three retirement accounts from Fido, leaving one. The downside is unavailability of some funds, of course, but they did transfer ones they do not 'carry'.
I asked Fido beforehand about retention / stay incentives, esp as a 50y customer, and got zilch.

Great find @Ted.
I’m always interested in what T Rowe Price is doing. Interesting that they had a Retirement Income Fund for many years, but decided about 5 years ago to rename it Retirement Balanced.
Now a new Retirement Income fund? Modeling its performance expectations on their current 2020 Target date fund would make it somewhat more aggressive than their previous Retirement income fund (TRRIX). In hindsight, rebranding the old fund must have been Price’s way of “clearing the deck” for this new one. Brings to mind, “What’s in a name ...”. When a company reaches the point where there are no longer enough names to go around due to their offering so many funds, what does that say? :)
Still reading this story. Not entirely clear whether there’s a glide slope with this one - but probably not.
(Actually, their website says there is a glide slope). I don’t understand where the firm is going with the launch of so many new funds in recent years. This one is a real puzzle (unless their goal is just to attract more and more assets). Dodge & Cox seems to do just fine with only 5 or 6 funds.
One thing that would steer me clear of this one - In order for it to work as intended, an investor would seem to have to entrust his/her entire retirement nest egg to this fund. Diversifying into several other funds would appear to thwart the fund’s intended goal.

FYI: The latest target-date funds focus on the task of managing a nest egg once retirement has started, including RMDs. Will they catch on?
Regards,
Ted
https://www.wsj.com/articles/new-target-date-funds-are-geared-for-withdrawal-time-1507515120?tesla=y

Hi Guys,
Being an "aimer" is very common. We all have goals. The challenge is how realistic those goals are. What are the odds of achieving those goals? Given market uncertainties, a 100% success goal is very bushy-tailed. But it is achievable if an investor is flexible to changing circumstances.
An early step in that process is to identify the likely odds of success. Monte Carlo simulators provide one useful tool to make those estimates. That tool not only yields the odds for success, but also makes estimates of end wealth, and a timeline for the portfolio failures. If those failure times are well beyond likely life expectancy, the failures are far less significant for planning purposes.
I well understand why only a 95% portfolio survival projection would be troublesome for some folks. It was for me. However, once that estimate is known, an investor could think about adjusting his withdrawal plan to alleviate that possibility.
When planning my retirement, I programmed my own version of a Monte Carlo code. I frequently calculated portfolio survival likelihoods in the mid-90% ranges. What to do? To improve that survival rate, I modified the code to reduce drawdowns by an input value (like 10%) if annual market returns were negative for some specified years. Withdrawals were increased once the markets turned positive in the simulations.
I explored many such portfolio survival issues by using my easily modified Monte Carlo code. Portfolio survival rates of very near 100% could be projected when very modest withdrawal rate flexibility was allowed. Like in so many other life situations, flexibility is a winning strategy.
My Monte Carlo calculations identified the frequency of shortfalls, the magnitudes of any shortfalls, and the timeframe of those shortfalls. These were all useful inputs for my retirement decision. The very modest adjustments needed to alleviate that unacceptable outcome gave great comfort. These additional Monte Carlo simulations cemented my retirement decision.
I believe (alternately, IMHO if you dislike "I believe") that Monte Carlo simulations would help many MFOers when considering their retirement decision.
Thanks for the Kitces reference. Be aware that he has a vested interest in the subject matter of that referenced article. He closes his piece with the following declaration:
"Michael Kitces has a financial interest in the US distribution of the Timeline app."
I have no such vested interest in Monte Carlo codes. I merely advocate that they be included as one tool to support investment decisions. They permit easy multiple sensitivity analyses. Of course, they depend on good input data ranges. They do not stand alone.
Best Wishes

I always wonder what the practical effect of such fine distinction-making is.
'For the particular kind of [prostate, breast] cancer you have, the new data show that watchful waiting outcomes are as good in terms of mortality and life quality as treatment, often better, and the number needed to treat is yada yada. Discuss with your doctor whether treatment or monitoring is right for you.'
'Return-sequence risk is always significant and badly down years at the start of your retirement can be deleterious to all of your planning. Discuss with your adviser the consequences of not planning yada yada ...'
And then what? What is the discussion? What can it change besides (dis)comfort level and moves toward drastic preventive actions? How wise is it to have 'just get rid of it' surgery or go to all laddered CDs? In the worst case, plenty wise. So is the discussion necessarily education in likelihood of worst cases?

Some of it certainly is education about worst case probabilities. There's a general belief that outcomes are better if treatment is more aggressive. Sometimes that's true, often it's not, especially given possibilities of false positives (not ill when tests say otherwise).
For example, mammograms are not too reliable with dense breast tissue. Here's a page from the American Cancer Society suggesting in that case you talk with your doctor, because on the one hand you might want to also have an MRI. But it also says that MRIs produce false positives leading to more tests and biopsies (which have their own risks).
https://www.cancer.org/cancer/breast-cancer/screening-tests-and-early-detection/mammograms/breast-density-and-your-mammogram-report.html
Similarly, PSA tests are not especially reliable and can lead to biopsies with risks.
A good part of the conversation can simply be an exploration of what's really important to you. In some other thread was a link to an article on how the usual financial planning questions are not helpful, e.g. "would you risk a 20% loss if 2/3 of the time you'd gain 15%?" People don't know what they want or how they'd react if actually (not hypothetically) faced with a 20% loss.
So IMHO talking with a planner at length about what really concerns you and discussing the cost/benefits of different risk mitigators (e.g. immediate annuities, long term care insurance, greater cash allocations, etc.) is a good use of time.
Different people place different values on a given outcome. Worse, most people don't even have a clear idea of how they value each possible outcome.
At one end of the spectrum you have women who will have radical mastectomies because they have a genetic risk of cancer. They choose life, regardless of its quality, over all else. At the other end of the spectrum, you have men who will decline prostate cancer treatment even when faced with certain death, rather than assume any risk of impotence due to treatment.
There are real people like that. I think I can appreciate their perspectives even if I don't agree with them.
Personally, I don't want to go broke, period. In that financial sense, I take an extreme position. A magic number of, say 95% chance of success tells me nothing. I need to know what the 5% of paths look like. Then I can explore possible followup actions that would increase success to 100%.
Likewise with that doctor talk. It's difficult to follow radiation therapy with surgery if the radiation is unsuccessful, while the reverse is much easier. That's a consideration in selecting choice of treatment, if one is willing to live with the much higher likelihood of a degraded quality of life due to multiple therapies.
Knowing not only the odds but the paths of outcomes enables you to plan for dealing with possible failures. And for not doing something just because the expert, whether physician or advisor, felt it was best in his not so humble opinion.

Hi msf,
Thank you for reading my post, but an especial thanks for alerting me to my error in making a double reference. The Portfolio Visualizer website provides an excellent Monte Carlo simulator, but it doesn't merit a double endorsement.
In my initial post, I meant to acknowledge a Monte Carlo simulator that is presented on the MoneyChimp website. It is very simple to use. I have referenced it in earlier postings. Here is the corrected, initial Link:
http://www.moneychimp.com/articles/volatility/retirement.hem
Thanks again for alerting me to my mistake.
When doing any financial analysis, I too am often puzzled over how extensive the data set should be in terms of timespan. The marketplace is in constant change and the relevance of data going back numerous decades is suspect.
Far too much has changed. We have morphed from an agricultural dominated economy to an industrial powerhouse. Market investing is now dominated by institutional agencies and not by individual players. Factors like these should influence the data set selection.
Depending on the scope and purpose of the research, the timeframe for the selected supportive data sets must be prudently chosen. "Garbage In, Garbage Out" applies. For most of the limited analyses that I do, I favor data from after WW II. I recognize the shortfall in numbers that that decision introduces, but I believe these data are more relevant. When doing any statistical analysis, tradeoffs of this kind always must be made. Decisions like these are not always obvious or easy.
To excite some additional responses I'll close,with this decision making quote from Donald Trump: "I have made the tough decisions, always with an eye toward the bottom line. Perhaps it's time America was run like a business."
I anticipate and await acerbic comments.
Best Wishes

I always wonder what the practical effect of such fine distinction-making is.
'For the particular kind of [prostate, breast] cancer you have, the new data show that watchful waiting outcomes are as good in terms of mortality and life quality as treatment, often better, and the number needed to treat is yada yada. Discuss with your doctor whether treatment or monitoring is right for you.'
'Return-sequence risk is always significant and badly down years at the start of your retirement can be deleterious to all of your planning. Discuss with your adviser the consequences of not planning yada yada ...'
And then what? What is the discussion? What can it change besides (dis)comfort level and moves toward drastic preventive actions? How wise is it to have 'just get rid of it' surgery or go to all laddered CDs? In the worst case, plenty wise. So is the discussion necessarily education in likelihood of worst cases?

With respect to tools that use real data as opposed to hypothetical random data (that downplay the possibility of multiyear bears found in historical data) being somehow inferior or limited to merely hundreds of sequences, here's some what what Kitces writes:

There’s never been any way to illustrate those alternative assumptions [e.g. "spending changes based on varying goals or changing needs"], as even the best financial planning software is still built around straight-line assumptions and Monte Carlo analysis.
Until now, as in the past year, two new software solutions for advisors have come forth ...

While those two tools are designed for advisors, what I want to highlight is that they are both based on historical data spanning a century (one using montly CRSP data back to 1920) or more (the other using DMS global data of a score of countries going back to 1900). Not small data sets, and as I read Kitces, better than any existing probability-based tools.
He goes on to note that even these tools are best used for educational purposes, not planning:

[They] are better viewed as a mechanism to teach and illustrate safe withdrawal rates, the sustainability of (steady) retirement withdrawals in the face of various market return sequences, and the impact of asset allocation ... on the sustainability of portfolio distributions. In other words, they can set the groundwork for initial client education about sequence of return risk and its consequences

FYI: Insuring a secure retirement with retirement experts Kim Lankford and Chris Blunt.
Regards,
Ted
http://wealthtrack.com/two-retirement-experts-address-the-rising-costs-of-retirement-and-how-to-conquer-them/

Hi Guys,
Wow! This assembly of articles by financial writer Johnathan Clements is a real tour-de-force. It touches all the bases that a retiree should consider before retirement, during the decision process, and after retirement. In general, these short pieces yield excellent insights and advice.
But this impressive body of work has a significant shortfall in the area of examining alternate possible financial,scenarios. Based on my earlier submittals, I'm sure many of you can make an accurate guesstimate of what I beleve that area is.
The sequence of investment returns is critical in determining the value and survival of any portfolio during both its accumulation and withdrawal phases. In examining likely financial scenarios, Clements uses the FIRECalc tool kit to explore a limited number of those sequences. I say limited because FIRECalc deploys real historical data with various starting points. This technique generates hundreds of scenarios that have been experienced.
Given the uncertainty of future returns, even hundreds of cases that have been already recorded is simply not enough data when considering all future uncertainties. Thousands and thousands of simulations provide a more complete set of possibities. Projections must be made in terms of likelihoods and odds. Nothing more definitive is possible.
Monte Carlo tools and numerous analyses with those tools are the best way to provide those odds. These tools are readily accessible and easy to use. You need not be a mathematician to productively deploy them. In earlier posts, I have referenced these tools; some are more elegant than others but all serve their basic function well. Here is a Link to a very easy input one:
https://www.portfoliovisualizer.com/monte-carlo-simulation
Another Monte Carlo simulator that requires a few more inputs is listed below:
https://www.portfoliovisualizer.com/monte-carlo-simulation
Sorry for my repeated references, but perhaps some MFOers missed my past postings. Both these tools do good work.
I hope you visit them and try a few cases that represent your special,circumstances. It will be a learning experience that will illuminate some possible dangers if you do a couple of parametric input scenarios. Good luck! Doing these analyses will reduce your need for luck and will help guide your investment and retirement decisions.
Best Wishes

FYI: If you just entered the workforce, it’s time to start preparing for retirement. Over the next four decades, you might pull in tens and perhaps hundreds of thousands of dollars every year. An October 2012 Census Bureau study estimates that those with a bachelor’s degree have average lifetime earnings of $2.4 million, figured in today’s dollars.
Of course, it’s lucky you have all that income coming in, because ahead of you lies life’s toughest financial task: amassing enough money so you can retire in comfort. In dry economic terms, your working career is about accumulating enough financial capital, so that one day you’ll no longer need the income from your human capital. This, alas, is a task that most Americans are not good at.
Want to do better? As you ponder how to pay for retirement, it’s helpful to think about your life in three stages—your 20s, 30s and 40s, your 50s and early 60s, and age 65 and beyond—which is how this part of the guide is divvied up.
Regards,
Ted
http://www.humbledollar.com/money-guide/retirement/

A factor that affects returns about as much as expense ratio is turnover. Here's an old but readable column citing research from a decade ago supporting the thesis that trading costs amount to at least as much as costs included in the ER:
https://www.advisorperspectives.com/pdfs/newsltr27-2.pdf
I tend to lean toward managed funds, but at the same time I also keep an eye on expenses, both documented (ER) and implicit (trading costs via turnover). FWIW, with the exception of a few placeholder positions I have (in case I ever want the funds), and one major bond fund holding, all of my funds have less than 50% annual turnover. Of those, all but one fund has under 40% turnover.
M*'s portfolio analysis claims that my weighted average ER is about a half percent below a "similarly weighted hypothetical portfolio". So while my expenses are not rock bottom, I'm satisfied with my average expense ratio too.
One minor gripe concerning the cited AAII article - it uses RMDs as withdrawal amounts. The main problem with this is that people's need for cash in retirement is fairly stable, as opposed to RMDs. RMD calculations require you to take 1 / (remaining lifetime), which grows exponentially as you get older and your expected lifetime shrinks toward zero.

@ Edward: Thanks for the above article, in this month's commentary. I was wondering if anyone is using the above mentioned , indexing & ETF's, to lowering their costs & therefore put a few more bucks into one's pocket. After rolling two 401-ks I'm thinking of doing something like this with 1/3 of my assets.
Thanks for any & all replies,
Derf

In addition, I looked through my stack of funds; and, I have two that bettered the 8.9% objective over a ten year period. They were FDSAX and SPECX. However, remember the 2007 and 2008 returns from the Great Recession that brought the average down will soon be coming off at the end of next year. My broker has the thinking that a balanced portfolio will return somewhere between 6% to 8% on average over the next ten year period depending on it's equity allocation and positioning. He is not looking for great things from bonds.
I think what the article was trying to establish is that market returns are not going to meat investor expectations.
I wish all ... "Good Investing."
Old_Skeet

I can't quite tell if you're talking nominal or real rate of return. Supposedly (though I have my doubts as noted above) the 8.9% objective is real return.
Using the BLS figures here for annual CPI-U (inflation) annual amounts for the 10 past July's, I computed a cumulative inflation of 18.07% over the past decade.
M* reports that $10K invested in FDSAX a decade ago would be worth $25,064 today, in nominal dollars. Adjusting for inflation (dividing by 1.18066) gives a real value of $21,229. Annualized, that $10K grew at a rate of 7.82%/year to become $21,229 in real dollars. Still terrific, but not quite 8.9% real return.
While 2008 will soon drop off the 10 year chart, that won't magically make your expected returns better, at least if you're looking long term. No more so than 2000-2002 dropping off the 5 year chart helped investors when 2008 came along. Bears will still come along sooner or later, you just don't know when. Looking at bull market data alone is IMHO misleading.
Not trying to be a wet blanket here, just trying to be, from my perspective, realistic. An aging population suggests slower growth, as does the fact that companies are hoarding cash (rather than putting the money to work).

Here's the full Natixis 2017 global survey report:
http://durableportfolios.com/global/understanding-investors/2017-global-survey-of-individual-investors-retirement-report
and the full Natixis press release on the US slice of that survey:
https://ngam.natixis.com/us/resources/2017-global-individual-investor-survey-press-release
(note that the table at the bottom of that US press release is global data, not data limited to US participants)
Just looking at the figures in the excerpt Ted quoted, my reaction was: what are these people smoking?
The historical real return of the US large cap market over the last century has been 7%. Depending on your source, bonds (10 year Treasuries) have returned between 2% and 6% less than stocks.
[See the stock link above: risk premium of stocks over bonds of 6%, historical nominal bond return of 5% with inflation average of 2%-3%, or simply the difference in nominal returns of stocks and bonds, which has been 2% or greater over the past 90 years.]
So even if the markets produce average real returns going forward (not expected over the next decade), you'd need a very aggressive (nearly all stock) portfolio to get to the 5.9% real return that advisors are supposedly predicting. (The 5.9%/advisors and 8.9%/investors figures are not in the Natixis releases, so they must come from the full survey.)
The FA Mag article says that there's a disconnect (51% difference) between investors and advisors, based on these two figures. If there is this disconnect, what does that say about the job that advisors are doing in educating and guiding their clients?
But there is another possibility. Investors may not understand what real return means, and are simply reporting nominal return expectations. That 3% difference would fall within a reasonable range of inflation possibilities. The Natixis report seems to support this interpretation of the data, as it observes that only 1/6 of Millennials (17%) "have factored inflation into their retirement savings planning." (The next sentence of the release hypothesizes a 3% inflation rate.)
Finally, note that the survey may not be representative of American households - just ones with money. It surveyed only investors with over $100K in investable assets. (About 30% Gen X, 30% Gen Y, 30% Boomers, 10% Retirees.) Most households don't have nearly that much in net worth let alone investable assets, though that's a whole 'nuther story.