Traditional versus Roth wiki update

Non-investing personal finance issues including insurance, credit, real estate, taxes, employment and legal issues such as trusts and wills
Lee_WSP
Posts: 4726
Joined: Fri Apr 19, 2019 5:15 pm
Location: Arizona

Re: Traditional versus Roth wiki update

Post by Lee_WSP »

The problem with 12% bracket questions is that it’s kind of hard to stay in that bracket and accumulate so much tax deferred savings as to push the rmd tax rate terribly high, so there’s not really a wrong answer.

In all other cases, it’s an easy assumption to assume one’s never going to be in that bracket again.
Topic Author
fyre4ce
Posts: 1386
Joined: Sun Aug 06, 2017 11:29 am

Re: Traditional versus Roth wiki update

Post by fyre4ce »

Lee_WSP wrote: Mon May 31, 2021 4:38 pm Another recent issue is the SECURE Act which has created a very strong preference for dying with Roth funds instead of trad funds.
Thanks. The page has commentary on the SECURE Act in the Estate planning section that seems to address your point. The Stretch IRA has more details about SECURE Act planning. You're welcome to propose any page updates you see fit.
Lee_WSP wrote: Mon May 31, 2021 4:38 pm A very very very little discussed issue (and poor assumption making by users) with holding too much trad funds is the mistaken belief that their current MFJ tax status is going to continue indefinitely. According to this paper (quick google search) survivor life expectancy is around 10 years for each spouse. http://humcap.uchicago.edu/RePEc/hka/wp ... pouses.pdf
This is mentioned in the wiki under the Tax risk. Do you want to propose new language for this section?
Lee_WSP wrote: Mon May 31, 2021 4:38 pm In addition to losing MFJ status, the survivor's IRA and RMD's will probably double in size (assuming the couple had earning and saving parity).
Are you double-counting here? Let's say each spouse had their own $1M IRA and a 5% RMD. That's $100k (2 x $50k) of income appearing on the MFJ tax return. Then one spouse dies and leaves their IRA to the other. Now one spouse has a $2M IRA. But they should still have a 5% RMD, which is $100k, right? I understand that the tax brackets are compressed and the survivor will pay a higher tax rate, but I'm seeing why the RMD would double too.
Lee_WSP wrote: Mon May 31, 2021 4:38 pm As such, IMO, the people who will be able to forecast their future tax rates and income streams the best are single people who plan to remain single until they pass.
That does remove one variable, for sure.
Lee_WSP wrote: Mon May 31, 2021 4:38 pm I've read the back and forth and the proposed wiki update with the case for going 100% trad.

I think the whole section relies on too many assumptions and generalizations and should be cut out completely.

If a person is unwilling or unable to make an assumption about future tax rates, then 50/50 or contributing the maximum $6,500 per year to Roth (through backdoor or front) are both equally valid routes since there are too many moving parts to say one method is always going to be better. Even an allocation of say 80/20 or 60/40 is equally valid.

As such, I think the less words used to simply tell the user to maximize all their tax advantaged space regardless of their allocation is arguably of better utility to the reader....
I agree it's a hard problem. There was a lot of discussion about the General guidelines and I think what we have now is a reasonable compromise between the views expressed here. I did decide to add the following two sentences, which (I hope) will have the effect you describe:
For those reluctant to contribute at all, either traditional or Roth, or any mix, is almost always a better choice than saving outside retirement accounts. In addition to providing more future income after taxes, traditional and Roth accounts also offer other benefits such as asset protection and estate planning.
Lee_WSP wrote: Mon May 31, 2021 4:38 pm If you truly want to, you can simply state that by choosing trad, you take the known deduction today and you'll have more money in your account after paying taxes this year. Whereas with Roth, you'll have less money today, but possibly/probably more money when RMD's kick in.
I would not agree (if this is what you're saying) that Roth must mean less money today, except in the case of maxing out. Someone deciding to go Roth can contribute less so their out-of-pocket cost is the same. What's more to the point is that traditional tends to be the right answer for most people most of the time, and there are some strong reasons for this, like those listed here.

In summary, I think everything you say is reasonable, but the page does at least touch on these topics. You're welcome to propose any new language for areas where you think there can be an improvement.
Last edited by fyre4ce on Mon May 31, 2021 11:03 pm, edited 1 time in total.
Lee_WSP
Posts: 4726
Joined: Fri Apr 19, 2019 5:15 pm
Location: Arizona

Re: Traditional versus Roth wiki update

Post by Lee_WSP »

fyre4ce wrote: Mon May 31, 2021 10:44 pm
Lee_WSP wrote: Mon May 31, 2021 4:38 pm In addition to losing MFJ status, the survivor's IRA and RMD's will probably double in size (assuming the couple had earning and saving parity).
Are you double-counting here? Let's say each spouse had their own $1M IRA and a 5% RMD. That's $100k (2 x $50k) of income appearing on the MFJ tax return. Then one spouse dies and leaves their IRA to the other. Now one spouse has a $2M IRA. But they should still have a 5% RMD, which is $100k, right? I understand that the tax brackets are compressed and the survivor will pay a higher tax rate, but I'm seeing why the RMD would double too.




Yes. It was an off the cuff analogy. The total RMD's do remain the same, but the bracket's get cut in half, so one could conceivably be pushed into a higher one rather easily.

Lee_WSP wrote: Mon May 31, 2021 4:38 pm If you truly want to, you can simply state that by choosing trad, you take the known deduction today and you'll have more money in your account after paying taxes this year. Whereas with Roth, you'll have less money today, but possibly/probably more money when RMD's kick in.
I would not agree (if this is what you're saying) that Roth must mean less money today, except in the case of maxing out. Someone deciding to go Roth can contribute less so their out-of-pocket cost is the same. What's more to the point is that traditional tends to be the right answer for most people most of the time, and there are some strong reasons for this, like those listed [url=https://www.bogleheads.org/wiki/User:Fy ... th#Results[/url].

In summary, I think everything you say is reasonable, but the page does at least touch on these topics. You're welcome to propose any new language for areas where you think there can be an improvement.
It does mean less money in your account today. You did not defer paying the taxes today. And by today, I mean the current tax year payment cycle.

If they aren't maxing out contributions, they need to forgo Roth and maximize their tax advantaged space as I said above. Feel free to correct me if I'm wrong, but from where I stand, if you don't use it, you lose it and you don't get to go back in time and re-contribute.

* I'll admit there may be a few "exceptions that prove the rule" cases, such as a business owner able to create so much tax deferred space as it's incomprehensible to the W-2 BH, but other than those, it's a fine rule.
Topic Author
fyre4ce
Posts: 1386
Joined: Sun Aug 06, 2017 11:29 am

Re: Traditional versus Roth wiki update

Post by fyre4ce »

Lee_WSP wrote: Mon May 31, 2021 10:54 pm It does mean less money in your account today. You did not defer paying the taxes today. And by today, I mean the current tax year payment cycle.

If they aren't maxing out contributions, they need to forgo Roth and maximize their tax advantaged space as I said above. Feel free to correct me if I'm wrong, but from where I stand, if you don't use it, you lose it and you don't get to go back in time and re-contribute.

* I'll admit there may be a few "exceptions that prove the rule" cases, such as a business owner able to create so much tax deferred space as it's incomprehensible to the W-2 BH, but other than those, it's a fine rule.
I generally agree here, but many Americans are unfortunately not in a position to maximize their contributions. The median household income is around $60,000, so I'd say if this family can contribute $10,000 to their retirement, they're doing alright. So the choice becomes either $10,000 pre-tax, or $8,800 Roth. Locking in a 12% tax rate is nice, but then again if this family has low assets, there's a decent chance they could withdraw at 10% or 0. Hard to say, and depends a lot on the surrounding circumstances.

This brings up a general philosophical problem with the wiki - is it better to target to the average American (I don't want to sound American-centric but it is a US site), or the average reader? The average BH reader almost certainly has more income, assets, and much more interest in solving these financial problems. Finding that balance continues to be a challenge.
Lee_WSP
Posts: 4726
Joined: Fri Apr 19, 2019 5:15 pm
Location: Arizona

Re: Traditional versus Roth wiki update

Post by Lee_WSP »

fyre4ce wrote: Mon May 31, 2021 11:59 pm
Lee_WSP wrote: Mon May 31, 2021 10:54 pm It does mean less money in your account today. You did not defer paying the taxes today. And by today, I mean the current tax year payment cycle.

If they aren't maxing out contributions, they need to forgo Roth and maximize their tax advantaged space as I said above. Feel free to correct me if I'm wrong, but from where I stand, if you don't use it, you lose it and you don't get to go back in time and re-contribute.

* I'll admit there may be a few "exceptions that prove the rule" cases, such as a business owner able to create so much tax deferred space as it's incomprehensible to the W-2 BH, but other than those, it's a fine rule.
I generally agree here, but many Americans are unfortunately not in a position to maximize their contributions. The median household income is around $60,000, so I'd say if this family can contribute $10,000 to their retirement, they're doing alright. So the choice becomes either $10,000 pre-tax, or $8,800 Roth. Locking in a 12% tax rate is nice, but then again if this family has low assets, there's a decent chance they could withdraw at 10% or 0. Hard to say, and depends a lot on the surrounding circumstances.

This brings up a general philosophical problem with the wiki - is it better to target to the average American (I don't want to sound American-centric but it is a US site), or the average reader? The average BH reader almost certainly has more income, assets, and much more interest in solving these financial problems. Finding that balance continues to be a challenge.
If they're in the 10% bracket, I don't think it matters much which you choose. I can't come up with any scenarios off the top of my head where it will make any meaningful difference whatsoever. The 12% bracket may make a slight difference, but contributing the maximum $6,500 to actual Roth and the rest in tax deferred is as good a strategy as any if you aren't able to maximize tax advantaged accounts.

As a writer you need to write for your audience. I do not know who the BH audience is, but based on forum members, they skew higher income and higher education.
User avatar
FiveK
Posts: 11536
Joined: Sun Mar 16, 2014 2:43 pm

Re: Traditional versus Roth wiki update

Post by FiveK »

fyre4ce wrote: Mon May 31, 2021 11:59 pm This brings up a general philosophical problem with the wiki - is it better to target to the average American (I don't want to sound American-centric but it is a US site), or the average reader? The average BH reader almost certainly has more income, assets, and much more interest in solving these financial problems. Finding that balance continues to be a challenge.
An excellent question. The statement in the "Getting stated" wiki, The Bogleheads® motto is Investing Advice Inspired by Jack Bogle. We are part of his campaign "to give ordinary investors a fair shake," suggests "the ordinary investor" as the answer.

The word "ordinary" implies "not special" so catering to only the top 10% or so wouldn't be appropriate. But Most Americans Don’t Have A Real Stake In The Stock Market so the median "ordinary investor" would have somewhat higher income, etc., than the average American (and probably the same in any nationality). Of course, half the investors would have income below the median investor so that's still a significant fraction with household income <$100K/yr.
Topic Author
fyre4ce
Posts: 1386
Joined: Sun Aug 06, 2017 11:29 am

Re: Traditional versus Roth wiki update

Post by fyre4ce »

FiveK wrote: Mon May 31, 2021 8:04 pm Modifying the "teachers" case to have the SS benefits and "other income" increase linearly by year to their final value, one gets
Image

Due to the way SS and pension benefits accrue, linear isn't correct but at least it's more correct than assuming a constant value.

Just because someone disagrees with you doesn't mean the other person lacks understanding.

And yet those spreadsheet results are not at all conclusive. E.g., see the chart above and the one here, which comes from changing the annual income from $410K to $400K:
Image

To paraphrase Inigo Montoya, you keep saying "the answer seems obvious" but I do not think these spreadsheets show what you think they show.

You did say that you "didn't take the time to make sure all the formula were generalizable if inputs were changed" so it's possible that changing the income from $410K to $400K violated some implicit assumption. But that's the sort of thing that should be fixed before saying the calculations prove anything.

The modified version of the spreadsheet is at https://drive.google.com/file/d/1ZNPkFz ... sp=sharing
I really don't mean to be rude, but you seem to me to be missing the forest from the trees with these spreadsheets. You're generating these plots that show that having extra pre-tax money helps you when your household income permanently drops to 0 anywhere from 0 to 40 years ahead of your planned retirement. No one, including me, is going to argue that that's true. But let's not lose sight of how extreme some of these cases are. One spouse getting laid off at 55 or 60 is one thing. You're talking about total permanent retirement of both household earners going back as early as your 20's. That's an extreme, and very unlikely, situation (and one that should be insured against in other ways - more on this later). You're picking a range of cases that's biased heavily in one direction. Let's say these couples start working at 22 and plan to work to 62. Why only look at a window of 0 to 40? Wouldn't 30 to 50 make more sense (ages 52 to 72)? That would seem to capture a range of more typical retirement dates. What would the answers look like then? When you start by asking the wrong question, you're not going to get a useful answer.

And, I noticed you seem to be tweaking the income in the direction that makes the best-looking plots. I changed the teacher's income back to the original $110k (you raised it to $150k), and the long-term planning breaks even after just 12 years - that's permanent retirement at age 34! Raising the "splitting" couple's income to $420k rather than lowering it to $400k gives a break-even at just 19 years, and a ~4% benefit at a full career length vs. a best-case savings of ~2% with a permanent retirement after working just 5 (!) years. Again, I hope this doesn't come across as rude, but I'm worried you're trying to reason toward a desired conclusion, rather than following the evidence wherever it leads.

Another issue I have is that your proposed solution to unplanned early retirement is a really inefficient way to do it. With basically any set of inputs, a Future Value prediction in early career with 0 payments is going to return a very low number. So, your analysis is going to see a future 0% bracket and peg 100% traditional in almost all cases, until things start to level out after ~3-8 years. That's a problem, because one's early years are often the best time to go Roth - incomes are usually lower and/or you get the biggest benefit when maxing. More than anything, this is what should be taken away from the spreadsheets. That's why I highlighted those 100% traditional years in red. Frankly, I'm all for the idea of keeping some extra traditional as "dry powder" to be spent as Roth conversions in occasional unplanned low-income years, or even in a more extreme loss-of-income situation. But wouldn't it be a better, more balanced strategy to build it up more gradually, maybe over ~10-20 years, to still take advantage of those great Roth opportunities? Maybe, a rule like "tie goes to traditional until you have enough to fill a 0% bracket with 4%" or something?
FiveK wrote: Mon May 31, 2021 8:04 pm a) no disagreement that results from a plan that matches the actual future may outperform one that does not match the actual future.
b) A study and some data below.
camillus wrote: Sat May 29, 2021 6:56 pm --According to the AARP, half of workers are forced out of a job in their 50s. https://www.aarp.org/work/working-at-50 ... ement.html
Actually, that's an interesting article. I didn't know the numbers were that high. But it's worth noting that they count "job loss" as 6 months out of work, or taking a lower-paying job, not necessarily the permanent loss of income you're talking about. And, this doesn't consider the other direction of outcomes, which is that raises, promotions, inheritances from aging parents are likely to come in this age range too, and these probably won't (and shouldn't) be part of someone's plan. Again, you're looking at only one type of risk when the reality is that many things can happen that will move someone in either direction tax-wise. But it also tells me that people should be careful when planning. Avoiding saving anything in your 20's and 30's because you can make the numbers work working til 70 is not smart. Maybe a planned retirement age of 60 makes sense. But then work your long-term plan around this date.
FiveK wrote: Mon May 31, 2021 8:04 pm The most recent "Number of Returns Classified by: Size of Adjusted Gross Income, Marital Status, and Age of Taxpayer" in SOI Tax Stats - Individual Statistical Tables by Size of Adjusted Gross Income | Internal Revenue Service shows ~2/3 of all returns in the 0%-12% brackets, ~1/4 in the 22%, and ~1/10 in all higher brackets combined.
c) See the transcript of Mike Piper: Delaying Social Security Not Always a Good Deal | Morningstar: "But I think it's important to be realistic in that if you are in your 20s, or you're in your 30s, any detailed analysis to calculate what your marginal tax rate will be in your 60s or 70s, it's pointless, it's just a guess." Granted, the word "detailed" is subjective, but....
Sure, I'd agree, but we should still do the best job we can. Tax-planning around a forced early retirement at age 28 would not be the best one can do, I think.
FiveK wrote: Mon May 31, 2021 8:04 pm
Can you tell me what would change your mind?
Sure. Either of the following would suffice:
1) demonstrated ability to predict future tax law and individual circumstances, or
2) data showing that more people are underpredicting their retirement marginal tax rates than overpredicting them, with a root cause analysis of why and a plausible fix for that root cause.
#1 is, frankly, ridiculous. Anyone who claims to be able to predict future tax laws is lying or trying to sell you something. For #2, you're setting the bar too high. Hardly anyone tries to predict their future tax rate down to a number, even on BH. But in my experience, it's very roughly split between people who end up in a better financial situation than they expected vs. worse. I'm not seeing this heavy bias toward catastrophe, and even if I did, the right answer would not be "100% traditional in early career" as your formula would suggest.
FiveK wrote: Fri May 28, 2021 1:54 pm
To my mind, 90% of this whole T vs R question comes down to that future rate estimate, and it's so important to get it accurate as possible. Everything else is known, or straightforward to figure out. As I mentioned before, I'm working on a spreadsheet that automates this calculation, and I have a draft almost ready. The situation in that thread seems like a prefect candidate for using it. But, this update should be completed first.
Seems better to "try it out" several times before changing the wiki. Either you'll find modifications needed or more people will understand why it is a better approach.
It's an interesting idea, but I think it would confuse the basic concepts with the software implementation.
Topic Author
fyre4ce
Posts: 1386
Joined: Sun Aug 06, 2017 11:29 am

Re: Traditional versus Roth wiki update

Post by fyre4ce »

FiveK wrote: Mon May 31, 2021 8:05 pm
fyre4ce wrote: Sat May 29, 2021 4:30 pm This move has a similar character to another financial product: disability insurance. You pay a "premium" (the tax cost for going traditional when the long-term planning says you should go Roth), and if you become disabled, you get a "benefit" (the tax savings of withdrawing your money at a low or zero rate). So let's look at this move as though it were a disability policy.
Let's not, because that is an inappropriate comparison.

If instead you want to do some sort of statistical modeling to look at the Expected value of the t/R choice, that could be appropriate. Because the future marginal rate is a guess, one could be "scientific" about it by estimating the probability of various future rates, depending both on tax law and individuals' likely career paths, then multiplying those probabilities by the after-tax outcomes of those marginal rates. One could do that, but probably better to use Occam's razor and not do that.
Especially after this, I honestly can't see how anyone can have lingering doubts about this. If you still do, I'm sorry I haven't made a better case. I'm happy to have any mistakes in my analyses pointed out and will write corrections if necessary. If you have a proposal for how we can resolve this disagreement differently, I'll listen. I don't enjoy heavy-handed editing and will do it only as a last resort, but my priority needs to be getting the most accurate and helpful content on the wiki, and on this point, the answer is clear.
Because the disability insurance comparison isn't applicable, any conclusion based on it likewise isn't applicable.
If you're not seeing the connection, let me connect them for you. :-)

Readers should not be expected to do any kind of Monte Carlo or probability-weighted analysis on their own, I agree. That's asking way too much. Maximizing a future expected value, though, is a generally good approach to many problems, including this one. As a short-cut, if one assumes some symmetry in the values (ie. amount of taxes gained or lost based on various strategies are roughly comparable) and some bell-shaped range of possible outcomes (ie. tax rates will most likely stay the same, small chance they go up or down), then this can be simplified to just planning around the most likely outcome (which is what I've been saying all along). But can we try to look at this as an expected value? Oh, I think we can:

Link - image hosting doesn't seem to work

Exact values are going to be different for each person's situation, but plug in any reasonable numbers and you'll get the same answer. The "plan on likely income" gives better results for two reasons. First, you're "tuning" your planning to the most likely outcome, which will save you the most taxes in the most likely case. Second, by centering your plan around the middle of the range, you're removing the possibility of really bad outcomes, like planning for one extreme and getting the other. This method is about as non-controversial as it gets, and I'd even be shocked if FiveK had an issue with the basic concept.

But I can foresee two possible objections. One is that the probabilities are too centered, when they should actually skew heavily toward the bottom of the table. If that's true, my response would be that you have too aggressive a financial plan if you can't meet or beat it with some reasonable likelihood. If you have an unreliable income, it definitely makes sense to plan around some low-income years or maybe even an involuntary retirement like FiveK proposes (although maybe not in your 20's). If you're in a career that doesn't typically have a growing income, maybe plan on a decreasing real income by 1%/year. If you expect a growing income, plan on your real income staying flat. Don't count on any inheritances, even if you're a beneficiary. Per the AARP article FiveK linked, maybe rather than planning on working til 65 or 70, you plan to work until 55 or 60, and any extra is gravy. But a reasonable plan, by definition, should be one that you have a decent chance of following. Dialing back your plan will cause you to lean more toward traditional, but not in the extreme way as FiveK's formula where you're 100% traditional for all your early years.

The second is that this maximizes expected value when it should maximize expected utility. In non-analyst lingo, that means: sure, you're planning for a likely outcome, but if you have a financial catastrophe, then you'll need money a lot more badly than if you have some windfall. So, it's better to plan around the rare worst-case scenario and cover yourself there as best as possible. This is where disability insurance comes into play, which is almost a perfect comparison to what's being proposed here. The problem with the "utility" approach is that it's costing you a bunch of taxes in most scenarios. But rather than pay that almost-guaranteed cost, you can spread the risk around with a disability policy, which insurance companies price to reflect the low likelihood of the permanent disability FiveK is talking about (in the policy I referenced, less than 1% chance). Rather than adopting a tax strategy that will very likely cost you money, you can pay a much smaller premium to the insurance company to cover that risk, and still get the tax benefits of planning for your normal career. To boot, you also get real protection against this kind of extreme catastrophe (monthly cash until you're 67), rather than just a few percent tax savings on the money you already have saved. That's why insurance is applicable here.

-----

I'm not sure what else I can say. I appreciate your enthusiasm on this topic, but every angle we look at it from gives the same answer: create a reasonable financial plan, make financial decisions based on that plan, insure against catastrophe if you can't self-insure, and adjust the plan when necessary. I won't have time to write lengthy posts like this again for a while. If your passion on this still burns, I'd encourage you to think more carefully about the concerns that have been raised about your proposal, like: What other possible scenarios besides early retirement does it make sense to include in the analysis? Which is better for protecting against disability, insurance or tax-planning, and why? If someone wants to lean traditional, is 100% traditional in early years the best time, or are there better options? Which method is better for handling the case of early-career Roth conversions? Which method is better for handling cases of increasing income (doctors and others)? Is the "assume no future income" recommendation applied universally, or are there exceptions? If so, are exceptions based on expected job security, or other factors? If you think through all the possible ramifications of what you're proposing, and these questions, I think the path forward will become clear. Or, if I'm missing something, you or anyone else are welcome to post new thoughts here and I'll check back periodically.
User avatar
FiveK
Posts: 11536
Joined: Sun Mar 16, 2014 2:43 pm

Re: Traditional versus Roth wiki update

Post by FiveK »

fyre4ce wrote: Tue Jun 01, 2021 3:13 am I really don't mean to be rude
Yet somehow you manage.
but you seem to me to be missing the forest from the trees with these spreadsheets. You're generating these plots that show that having extra pre-tax money helps you when your household income permanently drops to 0 anywhere from 0 to 40 years ahead of your planned retirement. No one, including me, is going to argue that that's true.
The point was to look at the behavior of the function. It's a good way to troubleshoot numerical models in general and spreadsheets in particular. That's how the errors in your original version came to light. Yes, the first few years are irrelevant for practical purposes.

The main points are
1) there isn't a significant difference in results regardless of which approach is taken. Thus neither is "obviously better".
2) small changes to inputs (e.g., income) seem to have inordinately large changes in results, just based on the various examples presented. And that's assuming constant returns, etc. Given the realities of life, that suggests anything more than a simple, approximate, approach is likely a fool's errand.
Frankly, I'm all for the idea of keeping some extra traditional as "dry powder" to be spent as Roth conversions in occasional unplanned low-income years, or even in a more extreme loss-of-income situation. But wouldn't it be a better, more balanced strategy to build it up more gradually, maybe over ~10-20 years, to still take advantage of those great Roth opportunities? Maybe, a rule like "tie goes to traditional until you have enough to fill a 0% bracket with 4%" or something?
Whether the Roth opportunities will really have been great will be known only in hindsight. But something like "consider using 100% Roth for your first few (up to you to define "few") years of employment, regardless of what the marginal rate estimates suggest" could work. Again, there isn't likely to be a significant difference is results. Maybe Lee_WSP or someone else can suggest the wording to minimize "bickering". ;)
Again, you're looking at only one type of risk when the reality is that many things can happen that will move someone in either direction tax-wise. But it also tells me that people should be careful when planning. Avoiding saving anything in your 20's and 30's because you can make the numbers work working til 70 is not smart. Maybe a planned retirement age of 60 makes sense. But then work your long-term plan around this date.
Yes, that's the type of risk that would be most damaging if forced. It's also not uncommon for people in early career to think they'll like to work for a long time, yet find that after 25 or 30 years it sure would be nice not to. Those whose finances turn out much better than expected can cry all the way to the bank.
FiveK wrote: Mon May 31, 2021 8:04 pm
Can you tell me what would change your mind?
Sure. Either of the following would suffice:
1) demonstrated ability to predict future tax law and individual circumstances, or
2) data showing that more people are underpredicting their retirement marginal tax rates than overpredicting them, with a root cause analysis of why and a plausible fix for that root cause.
#1 is, frankly, ridiculous. Anyone who claims to be able to predict future tax laws is lying or trying to sell you something. For #2, you're setting the bar too high. Hardly anyone tries to predict their future tax rate down to a number, even on BH.
Hey, you asked and I answered. But it's good to see that you don't want to get too detailed about predicting the future. :) That isn't what you have been saying:
To my mind, 90% of this whole T vs R question comes down to that future rate estimate, and it's so important to get it accurate as possible.
Lee_WSP
Posts: 4726
Joined: Fri Apr 19, 2019 5:15 pm
Location: Arizona

Re: Traditional versus Roth wiki update

Post by Lee_WSP »

FiveK wrote: Tue Jun 01, 2021 5:00 am [
Frankly, I'm all for the idea of keeping some extra traditional as "dry powder" to be spent as Roth conversions in occasional unplanned low-income years, or even in a more extreme loss-of-income situation. But wouldn't it be a better, more balanced strategy to build it up more gradually, maybe over ~10-20 years, to still take advantage of those great Roth opportunities? Maybe, a rule like "tie goes to traditional until you have enough to fill a 0% bracket with 4%" or something?
Whether the Roth opportunities will really have been great will be known only in hindsight. But something like "consider using 100% Roth for your first few (up to you to define "few") years of employment, regardless of what the marginal rate estimates suggest" could work. Again, there isn't likely to be a significant difference is results. Maybe Lee_WSP or someone else can suggest the wording to minimize "bickering". ;)
Unless your company offers a Roth, you don't actually have much opportunity unless you utilize a traditional IRA and do conversions.

As such, filing up the base Roth and then focusing on trad is as good advice as any. You could modify it to contribute 50/50 or pro rata per maximum inputs.

IMO excess Roth contributions are a high earner problem.

I also think these theory discussions usually go too far into the weeds and start discussing fringe cases that are unlikely to ever come up
Topic Author
fyre4ce
Posts: 1386
Joined: Sun Aug 06, 2017 11:29 am

Re: Traditional versus Roth wiki update

Post by fyre4ce »

There is a consensus that this set of changes is an overall improvement to the page, and there were no suggested edits proposed in the last couple weeks, so the update has been moved live:

https://www.bogleheads.org/wiki/Traditional_versus_Roth

Edit: Two other pages went live in parallel with this update:

https://www.bogleheads.org/wiki/Retirem ... n_analysis
https://www.bogleheads.org/wiki/Tax_analysis

They don't add any new content, but rather collect analysis and formula derivations from this page and also elsewhere on the wiki, the goal being to move content that will only be relevant to a tiny minority of readers off main pages while still keeping it easily accessible.


There was one dissenting opinion on the question of whether reasonably-expected future contributions should be included in this analysis as a baseline. The basic points that including them results in more expected income after taxes (link) and that insurance is better protection against job loss (link) were not in dispute, and should be dispositive. But, if someone wants to explore this question further, I suggest a focused topic on this point in the Personal Finance forum that might bring in some new and more diverse opinions. I would participate to the extent my schedule allows. If there's then a consensus that they should not be included, the page can be edited again at that time.

Thanks to all who participated!
Last edited by fyre4ce on Wed Jun 09, 2021 11:42 am, edited 1 time in total.
User avatar
FiveK
Posts: 11536
Joined: Sun Mar 16, 2014 2:43 pm

Re: Traditional versus Roth wiki update

Post by FiveK »

I'll clean up a few things over the next few days, but will make the edits a little bit at a time to make it easier on anyone who wants to compare.
Topic Author
fyre4ce
Posts: 1386
Joined: Sun Aug 06, 2017 11:29 am

Re: Traditional versus Roth wiki update

Post by fyre4ce »

FiveK wrote: Wed Jun 09, 2021 1:22 am I'll clean up a few things over the next few days, but will make the edits a little bit at a time to make it easier on anyone who wants to compare.
Cleanup is fine, but please do not revert this change wholesale. For one, the two new analysis pages link to it and vice versa. (I should have included those links yesterday, but I went back and edited that post today to include links.) The changes made here have been discussed since January, advertised in the Theory forum, were clearly detailed, and withstood review from many people. With the possible exception of future contributions, feedback from all editors was able to be incorporated, or at least a reasonable compromise reached. If you have any significant editorial issues, I suggest a topic in the PF forum focused on that issue, and it can be explored to whatever depth is necessary.
Lee_WSP
Posts: 4726
Joined: Fri Apr 19, 2019 5:15 pm
Location: Arizona

Re: Traditional versus Roth wiki update

Post by Lee_WSP »

I think the section quoted below needs to be a bit clearer that this is only a consideration if you even have access to say a Roth 401k. Otherwise, it's not a consideration unless the reader had trad IRA and wants to do conversions.
Those who can't or won't make an estimate of future tax rates should consider:

Contributing 100% traditional, because it's the best choice for most people most of the time
The second clarification I'd suggest is to add the caveat that you should still fill up the annual Roth bucket each year if possible.

Edit: I'd change the wording to

Prioritize traditional. Fill up your traditional accounts and then contribute to Roth.
Last edited by Lee_WSP on Wed Jun 09, 2021 5:54 pm, edited 1 time in total.
User avatar
grabiner
Advisory Board
Posts: 29868
Joined: Tue Feb 20, 2007 11:58 pm
Location: Columbia, MD

Re: Traditional versus Roth wiki update

Post by grabiner »

fyre4ce wrote: Wed Jun 09, 2021 12:45 am
Edit: Two other pages went live in parallel with this update:

https://www.bogleheads.org/wiki/Retirem ... n_analysis
https://www.bogleheads.org/wiki/Tax_analysis

They don't add any new content, but rather collect analysis and formula derivations from this page and also elsewhere on the wiki, the goal being to move content that will only be relevant to a tiny minority of readers off main pages while still keeping it easily accessible.
This is useful, as it gets the complicated calculations off of this and several other pages (such as taxation of Social Security benefits; most investors care about their marginal tax rate, but not about the derivation of the formulas).
Wiki David Grabiner
User avatar
FiveK
Posts: 11536
Joined: Sun Mar 16, 2014 2:43 pm

Re: Traditional versus Roth wiki update

Post by FiveK »

grabiner wrote: Wed Jun 09, 2021 5:10 pm
fyre4ce wrote: Wed Jun 09, 2021 12:45 am
Edit: Two other pages went live in parallel with this update:

https://www.bogleheads.org/wiki/Retirem ... n_analysis
https://www.bogleheads.org/wiki/Tax_analysis

They don't add any new content, but rather collect analysis and formula derivations from this page and also elsewhere on the wiki, the goal being to move content that will only be relevant to a tiny minority of readers off main pages while still keeping it easily accessible.
This is useful, as it gets the complicated calculations off of this and several other pages (such as taxation of Social Security benefits; most investors care about their marginal tax rate, but not about the derivation of the formulas).
Agreed - a simpler main page is better.
User avatar
FiveK
Posts: 11536
Joined: Sun Mar 16, 2014 2:43 pm

Re: Traditional versus Roth wiki update

Post by FiveK »

Made a few edits toward the beginning of the article. Sincerely hope they are well received, or at worst acceptable.... :)

Revision history of "Traditional versus Roth" - Bogleheads
Topic Author
fyre4ce
Posts: 1386
Joined: Sun Aug 06, 2017 11:29 am

Re: Traditional versus Roth wiki update

Post by fyre4ce »

FiveK wrote: Wed Jun 09, 2021 11:38 pm Made a few edits toward the beginning of the article. Sincerely hope they are well received, or at worst acceptable.... :)

Revision history of "Traditional versus Roth" - Bogleheads
Thanks for taking another scrub through. I'm not nuts about the appeal to "feels right" but I can live with it, and the other changes too. As Lee_WSP pointed out, those are the only choices many readers will have anyway, $19,500 pre-tax plus match in their 401k, and $6,000 into a Roth IRA (possibly through backdoor). And I am a big fan of encouraging readers to try to make a numerical estimate, so your text about encouraging that is spot-on if you ask me. I still would like to publish a tool that will help, maybe in the next month or two, as my time allows.
User avatar
FiveK
Posts: 11536
Joined: Sun Mar 16, 2014 2:43 pm

Re: Traditional versus Roth wiki update

Post by FiveK »

fyre4ce wrote: Wed Jun 09, 2021 11:43 pm I'm not nuts about the appeal to "feels right" but I can live with it....
Neither am I, but the "traditional 401k and Roth IRA" is a common and not unreasonable suggestion in many forum threads. Very open to using a different rationale to support it.

Good to see convergence so far. Hope for more to come. :beer
Topic Author
fyre4ce
Posts: 1386
Joined: Sun Aug 06, 2017 11:29 am

Re: Traditional versus Roth wiki update

Post by fyre4ce »

FiveK wrote: Thu Jun 10, 2021 12:04 am
fyre4ce wrote: Wed Jun 09, 2021 11:43 pm I'm not nuts about the appeal to "feels right" but I can live with it....
Neither am I, but the "traditional 401k and Roth IRA" is a common and not unreasonable suggestion in many forum threads. Very open to using a different rationale to support it.

Good to see convergence so far. Hope for more to come. :beer
:beer indeed! Very happy we are able to make progress.

What about combining #2 and #3? They are both along the same lines, namely splitting contributions in some reasonable way. How about:
Splitting contribution in a reasonable way, such as 50/50, or traditional in an employer plan (eg. 401(k)) and Roth in an IRA, to get exposure to both account types
Seems to convey the same ideas more compactly, and removes the emotional appeals I was never a huge fan of. But, I'll leave you to make the call here however you see fit.
User avatar
FiveK
Posts: 11536
Joined: Sun Mar 16, 2014 2:43 pm

Re: Traditional versus Roth wiki update

Post by FiveK »

fyre4ce wrote: Thu Jun 10, 2021 12:25 am What about combining #2 and #3? They are both along the same lines, namely splitting contributions in some reasonable way. How about:
Splitting contribution in a reasonable way, such as 50/50, or traditional in an employer plan (eg. 401(k)) and Roth in an IRA, to get exposure to both account types
Seems to convey the same ideas more compactly, and removes the emotional appeals I was never a huge fan of. But, I'll leave you to make the call here however you see fit.
I wouldn't do the 50/50 personally, but "minimize maximum regret" does work for enough folks that if that's what it takes to help them over the "some tax-advantaged, any tax-advantaged..." hurdle then it's worth including.

Perhaps Lee_WSP, KlangFool, or other reader can provide a better reason why "traditional 401k and Roth IRA" would still be an acceptable rule of thumb even for those who have the choice between deductible and Roth IRA?
Lee_WSP
Posts: 4726
Joined: Fri Apr 19, 2019 5:15 pm
Location: Arizona

Re: Traditional versus Roth wiki update

Post by Lee_WSP »

FiveK wrote: Thu Jun 10, 2021 12:35 am Perhaps Lee_WSP, KlangFool, or other reader can provide a better reason why "traditional 401k and Roth IRA" would still be an acceptable rule of thumb even for those who have the choice between deductible and Roth IRA?
Deductible IRA is an odd and unusual circumstance to be in. You’d have to have no employer based retirement option or have a very small income. The best rule I can come up with is to just fill it up to the best of one’s ability, with a slight preference for Roth since a Roth dollar is worth more, but costs more.
User avatar
FiveK
Posts: 11536
Joined: Sun Mar 16, 2014 2:43 pm

Re: Traditional versus Roth wiki update

Post by FiveK »

Lee_WSP wrote: Thu Jun 10, 2021 1:01 am Deductible IRA is an odd and unusual circumstance to be in. You’d have to have no employer based retirement option or have a very small income.
Perhaps small is relative but a single filer putting $19.5K into a t401k could have a gross (i.e., before 401k deduction) income of $85.5K, and MFJ with two $19.5K t401k contributions a gross income of $144K, and still be eligible for full tIRA deductions, based on 2021 IRA Deduction Limits - Effect of Modified AGI on Deduction if You Are Covered by a Retirement Plan at Work.
Lee_WSP
Posts: 4726
Joined: Fri Apr 19, 2019 5:15 pm
Location: Arizona

Re: Traditional versus Roth wiki update

Post by Lee_WSP »

FiveK wrote: Thu Jun 10, 2021 1:35 am
Lee_WSP wrote: Thu Jun 10, 2021 1:01 am Deductible IRA is an odd and unusual circumstance to be in. You’d have to have no employer based retirement option or have a very small income.
Perhaps small is relative but a single filer putting $19.5K into a t401k could have a gross (i.e., before 401k deduction) income of $85.5K, and MFJ with two $19.5K t401k contributions a gross income of $144K, and still be eligible for full tIRA deductions, based on 2021 IRA Deduction Limits - Effect of Modified AGI on Deduction if You Are Covered by a Retirement Plan at Work.
I think such cases are edge cases as in so unlikely there's no broad answer we can give. Such a couple would be saving more than half their gross income. While possible, I don't think you'll find more than a handful of such couple's in the USA.

For a couple with such a high savings rate and low tax rate, I'm not sure Roth vs traditional will move the needle. If they keep it up for a lifetime, RMDs might push them from 12% to 22% though.
User avatar
FiveK
Posts: 11536
Joined: Sun Mar 16, 2014 2:43 pm

Re: Traditional versus Roth wiki update

Post by FiveK »

Lee_WSP wrote: Thu Jun 10, 2021 8:33 am I think such cases are edge cases as in so unlikely there's no broad answer we can give. Such a couple would be saving more than half their gross income. While possible, I don't think you'll find more than a handful of such couple's in the USA.
As percent of gross income, it would be 30% for the single, 35% for the couple. Coincidentally, the couple would hit the "22% marginal rate to traditional and 12% marginal to Roth" division almost exactly. The single would still be paying 22% for the privilege of Roth contributions.

But your point about "there's no broad answer we can give" is a good one, and consistent with fyre4ce's "big fan of encouraging readers to try to make a numerical estimate." If nobody has a good reason why "traditional 401k and Roth IRA" (when IRS rules don't force it) should be in the list then let's drop it.

That leads to the question of all the other rules of thumb that have their own exceptions. I'd also support dropping those.
For a couple with such a high savings rate and low tax rate, I'm not sure Roth vs traditional will move the needle. If they keep it up for a lifetime, RMDs might push them from 12% to 22% though.
Very true! For most the difference between choosing "correctly" vs. "incorrectly" will affect retirement income by less than +/-10%....
Lee_WSP
Posts: 4726
Joined: Fri Apr 19, 2019 5:15 pm
Location: Arizona

Re: Traditional versus Roth wiki update

Post by Lee_WSP »

FiveK wrote: Thu Jun 10, 2021 1:49 pm [Very true! For most the difference between choosing "correctly" vs. "incorrectly" will affect retirement income by less than +/-10%....
Exactly. I've been in some pretty heated arguments about Roth vs traditional in the past, but taking a step back, it really isn't anything more than massaging the numbers for the vast majority of BHers.

Now that I'm a little older and a lot wiser, it just doesn't matter nearly as much as the effort some people put into the calculations.

I'd posit that Roth is slightly better for super savers because of RMDs pushing them into higher brackets and they're unlikely to spend the whole RMD though.
Last edited by Lee_WSP on Thu Jun 10, 2021 2:05 pm, edited 2 times in total.
Lee_WSP
Posts: 4726
Joined: Fri Apr 19, 2019 5:15 pm
Location: Arizona

Re: Traditional versus Roth wiki update

Post by Lee_WSP »

FiveK wrote: Thu Jun 10, 2021 1:49 pm [As percent of gross income, it would be 30% for the single, 35% for the couple.
The limit is about 100k mfj, 2x 19,500 + 2x 6500 = roughly fifty thousand = half of gross by my napkin math.
User avatar
FiveK
Posts: 11536
Joined: Sun Mar 16, 2014 2:43 pm

Re: Traditional versus Roth wiki update

Post by FiveK »

Lee_WSP wrote: Thu Jun 10, 2021 1:59 pm
FiveK wrote: Thu Jun 10, 2021 1:49 pm [As percent of gross income, it would be 30% for the single, 35% for the couple.
The limit is about 100k mfj, 2x 19,500 + 2x 6500 = roughly fifty thousand = half of gross by my napkin math.
Yes, if gross income is $100K. If gross income is $144K the $39K 401k contributions make adjusted gross income (AGI) $105K and tIRA contributions remain deductible.
Topic Author
fyre4ce
Posts: 1386
Joined: Sun Aug 06, 2017 11:29 am

Re: Traditional versus Roth wiki update

Post by fyre4ce »

FiveK wrote: Thu Jun 10, 2021 1:49 pm
Lee_WSP wrote: Thu Jun 10, 2021 8:33 am I think such cases are edge cases as in so unlikely there's no broad answer we can give. Such a couple would be saving more than half their gross income. While possible, I don't think you'll find more than a handful of such couple's in the USA.
As percent of gross income, it would be 30% for the single, 35% for the couple. Coincidentally, the couple would hit the "22% marginal rate to traditional and 12% marginal to Roth" division almost exactly. The single would still be paying 22% for the privilege of Roth contributions.

But your point about "there's no broad answer we can give" is a good one, and consistent with fyre4ce's "big fan of encouraging readers to try to make a numerical estimate." If nobody has a good reason why "traditional 401k and Roth IRA" (when IRS rules don't force it) should be in the list then let's drop it.

That leads to the question of all the other rules of thumb that have their own exceptions. I'd also support dropping those.
For a couple with such a high savings rate and low tax rate, I'm not sure Roth vs traditional will move the needle. If they keep it up for a lifetime, RMDs might push them from 12% to 22% though.
Very true! For most the difference between choosing "correctly" vs. "incorrectly" will affect retirement income by less than +/-10%....
I would lean toward not including the "traditional in 401k and Roth in IRA". It seems like it was proposed more because that's the only choice that a good percentage of readers have, not necessarily because there is anything special about it from a numbers perspective. But I don't think it's terrible to include either. It might make asset allocation and rebalancing easier if readers are tax-adjusting their accounts. The page does discuss eligibility elsewhere and makes it clear that not everyone will even be able to make this choice because of accounts and income limitations.

And yes, you're absolutely right, T vs R is probably a third-order impact to one's planning. First-order would be savings rate (saving 15-20%+ of one's income, instead of 5%, 0%, or negative savings by taking on debt and declaring bankruptcy which unfortunately many Americans do). Second-order would be having some kind of reasonable asset allocation and keeping investment fees minimized (1% fee is about 20% of account value over 40 years). The numbers here really bear out that tax rate is a smaller impact than investment quality. T vs R is probably <5% swing if we're excluding clearly foolish decisions, like deferring at 12% when you predict a future 40%, or vice versa. This is why I haven't been a big proponent of the "risk asymmetry" and "insurance" aspects of this decision, because it seems like the total impact just isn't big enough to make these factors significant. That said, 2-4% on $100k of retirement income is still $2-4k/year after taxes, which is not tiny. I still feel like we have a duty to readers who are looking for that last few percent optimization in their portfolios to give them the tools to make the best decision for themselves. But I also like that we've included caveats in some places to suggest that the impact will probably be small, especially when the numbers look close.
Lee_WSP
Posts: 4726
Joined: Fri Apr 19, 2019 5:15 pm
Location: Arizona

Re: Traditional versus Roth wiki update

Post by Lee_WSP »

FiveK wrote: Thu Jun 10, 2021 2:17 pm
Lee_WSP wrote: Thu Jun 10, 2021 1:59 pm
FiveK wrote: Thu Jun 10, 2021 1:49 pm [As percent of gross income, it would be 30% for the single, 35% for the couple.
The limit is about 100k mfj, 2x 19,500 + 2x 6500 = roughly fifty thousand = half of gross by my napkin math.
Yes, if gross income is $100K. If gross income is $144K the $39K 401k contributions make adjusted gross income (AGI) $105K and tIRA contributions remain deductible.
:oops: I always forget to add back in the deductions.

For such a couple I think the versatility of having a Roth just in case is valuable.
User avatar
FiveK
Posts: 11536
Joined: Sun Mar 16, 2014 2:43 pm

Re: Traditional versus Roth wiki update

Post by FiveK »

fyre4ce wrote: Thu Jun 10, 2021 2:27 pmI still feel like we have a duty to readers who are looking for that last few percent optimization in their portfolios to give them the tools to make the best decision for themselves.
We can do that, but
a) that's probably a smaller fraction than those who just want to know the basics, and
b) trying to optimize over 30-40 years into the future is "ambitious" to say the least.

Consequently, any such advice and tool discussion would be better placed in a secondary page.
Lee_WSP
Posts: 4726
Joined: Fri Apr 19, 2019 5:15 pm
Location: Arizona

Re: Traditional versus Roth wiki update

Post by Lee_WSP »

FiveK wrote: Thu Jun 10, 2021 5:17 pm
fyre4ce wrote: Thu Jun 10, 2021 2:27 pmI still feel like we have a duty to readers who are looking for that last few percent optimization in their portfolios to give them the tools to make the best decision for themselves.
We can do that, but
a) that's probably a smaller fraction than those who just want to know the basics, and
b) trying to optimize over 30-40 years into the future is "ambitious" to say the least.

Consequently, any such advice and tool discussion would be better placed in a secondary page.
While I think they'd be better served if we were able to convince them that the future is essentially unknowable and they're only pushing a few cents around the margins until they're actually in a position where the picture is pretty clear; yes I agree. Either send them to the forums for a discussion or point them towards a tool for them to plug in their own numbers.
Post Reply