Traditional versus Roth wiki update

[fyre4ce]

The only thing I can contribute is that I think the Savers Credit and other credits should be mentioned in the general guidelines section at the beginning in case that's the only thing people read. To me, it can make favoring traditional contributions for people that are within range of various credits especially beneficial.

For example, we're in a low enough bracket that usual advice would be to contribute to Roth, but thanks to the Savers credit our entire federal tax bill can be wiped out with enough Traditional contributions. We can then use those to put back into Roth if we haven't maximized IRA space.

Thanks for the comment. My original edits moved a set of typical cases where each of the two types of contributions are preferred to the top of the page. The second traditional case is:

Low-income investors who can realize a substantial tax savings through lowering Adjusted Gross Income, due to Earned Income Tax Credit, Saver's Credit, ACA subsides, lowering interest payments on an income-driven repayment plan for loans expected to be forgiven under Public Service Loan Forgiveness, and other benefits

Following some feedback, late last night I moved these slightly down the page, into their own section. Can you take a look and let me know whether this is OK? Or do you prefer those to be moved further up?

Side point: I'm not saying you are wrong to prefer traditional in your situation, but make sure you understand the math - the break-even withdrawal rate when the Saver's Credit is in-play shifts slightly toward Roth, compared to just your tax savings for traditional contributions. For example, if you save taxes at 30% for traditional contributions (including a 10% Saver's Credit), and you get just a 10% Saver's Credit for Roth contributions, traditional contributions are better if your withdrawal marginal tax rate is less than (30% - 10%) / (1 - 10%) = 22.2%, not the full 30% of traditional contributions alone. Feel free to check out the Saver's Credit section of either page for more details.

[FiveK]

That would actually be good programming for Google Maps if one told it "I don't know where I'm going."

In that case, turning either right or left may be correct, and there is not enough information for Google Maps to choose.

Remember that in the wiki one reaches the "either all traditional or 50/50" choice only "If one does not believe a reasonable estimate [of withdrawal marginal rates] is possible." That's pretty much the equivalent of telling Google Maps "I don't know where I'm going."

If this causes people to go do a back-of-the-envelope estimate of where they may end up, that will have been a useful outcome.

Nope, sorry, I don't agree. If you have advice that's targeted at the most novice readers who can't make any progress with a quantitative analysis, it needs to be super-simple and clear, not giving them choices to consider without a clear way to choose between them. Offering more nuance and putting more of the onus back on the reader may be appropriate for an advanced section where you can assume readers can figure stuff out for themselves, but not for what we're talking about.

As I said earlier, I'm not huge fan of a "if you can't make sense of anything, do X" instruction. If someone's gonna choose randomly, let them choose randomly, and invest that page space toward readers who could at least match their case to one of the typical ones, or convinced to try an analysis (it's not that hard, once you know get the bigger picture). But, I know you feel strongly so I'm trying to find something that's technically accurate, and also as helpful to readers as possible - including, as you mentioned, encouraging them to save anything in the first place. I just did an edit of my page with a couple sentences similar to what I proposed above. Take a look and let me know what you think. It can be edited further.

You're getting closer - a few more iterations and we may get back to what is in the wiki now.

The philosophy for "what should be in the Bogleheads wiki?" seems best expressed by this quote from the Bogleheads wiki main page:

If you find yourself writing a reply to a forum question that's been discussed a number of times before, consider creating a wiki page with the answer. Then you and others can reply to subsequent questions on that topic with a link and a quote of your text.

When people ask "what should we do?" on this issue, the "compare your marginal rate now vs. what you expect it to be later" answer is straightforward enough. When they say "but I don't know what my future marginal rate will be" we journey to the realm of pure speculation. On that journey we meet other topics such as "what should my asset allocation be?", "should I pay debt or invest in all stocks?", etc.

Lacking any knowledge of future rates at all, the choices of "all traditional" or "50/50" are both supportable, but rely more on the feelings of the individual (similar to the "hate debt" vs. "like leverage" camps) than "technical accuracy".

[fyre4ce]

When people ask "what should we do?" on this issue, the "compare your marginal rate now vs. what you expect it to be later" answer is straightforward enough. When they say "but I don't know what my future marginal rate will be" we journey to the realm of pure speculation. On that journey we meet other topics such as "what should my asset allocation be?", "should I pay debt or invest in all stocks?", etc.

Lacking any knowledge of future rates at all, the choices of "all traditional" or "50/50" are both supportable, but rely more on the feelings of the individual (similar to the "hate debt" vs. "like leverage" camps) than "technical accuracy".

...but in the examples you give, you can at least describe the factors that affect the decision ("psychological or religious aversion to debt" vs "desire to maximize return"), even though they are hard to quantify. Here, it's left totally open - readers aren't given any tools or guidelines to choose between the two stated options within the context of that section. That's a problem.

As I said before, I think the biggest reason many people struggle with this decision is because they don't have a good way to estimate retirement income and tax rate. That's consistent with my experience with questions in the forums. Frankly, it's understandable, because a quantitative one like we have on the wiki isn't widely known. But, for those who are willing to spend a little time playing with numbers in Excel, they can come to a reasonable estimate, and I think we should encourage people to try, both in the wiki and on the forum. As I think about it more, a dedicated T vs R page in the PFT would be helpful. I don't want to hold up this review for that, but perhaps I'll have time over the next few months to take a crack. You would of course be welcome to participate.

We've spilled a ton of ink on a tiny piece of the edits I'm proposing. Can you provide feedback on any other changes? Maybe we can circle back on this one later.

[FiveK]

We've spilled a ton of ink on a tiny piece of the edits I'm proposing. Can you provide feedback on any other changes? Maybe we can circle back on this one later.

That I would be happy to do, and suspect we may be more aligned on those.

Not immediately, but in a day or so.

[bacon4retirement]

Traditional contributions and Roth conversions have the same net effect as a Roth contribution. So, Roth contributions can be simulated by, for example, making traditional contributions to a 401(k) and doing Roth conversions from a traditional IRA.

My thought was this: let's say you have a t401k and a tIRA, but you want to make Roth contributions. You could open a Roth IRA, but that's only a $6-7k limit. Instead or in addition, you could contribute $19.5k to the t401k and then convert $19.5k from the tIRA to a Roth IRA, getting the same net effect of a Roth 401k. The pre-tax contributions in one account give you tax-space to make Roth contributions from another. But you need a pre-tax tIRA, or a 401k that allows in-service withdrawals for this to work.

Maybe it's too confusing to justify the space - let me know what you think after this explanation. I'm open to just deleting it.

Oh, now it makes sense that this paragraph was pointing out that someone with no employer Roth option and tIRA could do Roth conversions instead of Roth contributions. I can see this being helpful for someone, but doubt it deserves space at the very top of the wiki article. I agree with your deletion.

[patrick013]

Side point: I'm not saying you are wrong to prefer traditional in your situation, but make sure you understand the math - the break-even withdrawal rate when the Saver's Credit is in-play shifts slightly toward Roth, compared to just your tax savings for traditional contributions. For example, if you save taxes at 30% for traditional contributions (including a 10% Saver's Credit), and you get just a 10% Saver's Credit for Roth contributions, traditional contributions are better if your withdrawal marginal tax rate is less than (30% - 10%) / (1 - 10%) = 22.2%, not the full 30% of traditional contributions alone. Feel free to check out the Saver's Credit section of either page for more details.

If I wanted to model this for an IRR calc which would really
crossfoot the whole thing, box it into a rigid format, I would
have these :

tIRA
$12,000 account contribution per year
$ 3,600 contribution tax reduction per year

Roth
$12,000 account contribution per year
$ 1,200 contribution tax reduction per year

20% accumulation marginal tax rate
22.2% withdrawal marginal tax rate
7% avg account return

I should be able to get a comparative IRR and cash aggregate
pro-forma using XIRR.

Correct ?

I haven't studied the new Savers Credit much so...

[fyre4ce]

Side point: I'm not saying you are wrong to prefer traditional in your situation, but make sure you understand the math - the break-even withdrawal rate when the Saver's Credit is in-play shifts slightly toward Roth, compared to just your tax savings for traditional contributions. For example, if you save taxes at 30% for traditional contributions (including a 10% Saver's Credit), and you get just a 10% Saver's Credit for Roth contributions, traditional contributions are better if your withdrawal marginal tax rate is less than (30% - 10%) / (1 - 10%) = 22.2%, not the full 30% of traditional contributions alone. Feel free to check out the Saver's Credit section of either page for more details.

If I wanted to model this for an IRR calc which would really
crossfoot the whole thing, box it into a rigid format, I would
have these :

tIRA
$12,000 account contribution per year
$ 3,600 contribution tax reduction per year

Roth
$12,000 account contribution per year
$ 1,200 contribution tax reduction per year

20% accumulation marginal tax rate
22.2% withdrawal marginal tax rate
7% avg account return

I should be able to get a comparative IRR and cash aggregate
pro-forma using XIRR.

Correct ?

I haven't studied the new Savers Credit much so...

You're close, but you made one mistake. To compare apples-to-apples, you need to make the after-tax costs of the two options the same. (Otherwise, you'll need to track separately the difference in a taxable account, which will needlessly complicate it.)

Let's say you want to contribute $12,000 after-tax. At a 30% traditional tax rate, that's a contribution of $17,143 (= $12,000 / (1 - 30%)). If you want to check that it's correct, with this contribution you get a tax savings of $17,143 x 30% = $5,143, which is the difference between the $17,143 you contributed and the desired $12,000 out-of-pocket cost.

The equivalent Roth contribution is $13,333 (=$12,000 / (1 - 10%)). (That's $12,000 after-tax, because $13,333 minus $1,333 tax savings is $12,000.)

If investments grow by 10x between contribution and withdrawal, with the traditional option you'll have $171,429 future value, which after a 22.22% tax will be worth $133,333 (= $171,429 * (1 - 22.22%)). That's exactly what you'd have if the $13,333 in the Roth account grows tax-free by 10x.

So in this example, if you expect your future marginal rate to be more than 22.22%, Roth is better; less, traditional is better. The derivation of that formula is here. Disclaimer: I don't think it's possible to get a 20% or 30% marginal tax rate now; this was just for example purposes. You should plug in your own numbers.

[patrick013]

You're close, but you made one mistake. To compare apples-to-apples, you need to make the after-tax costs of the two options the same. (Otherwise, you'll need to track separately the difference in a taxable account, which will needlessly complicate it.)

Let's say you want to contribute $12,000 after-tax. At a 30% traditional tax rate, that's a contribution of $17,143 (= $12,000 / (1 - 30%)). If you want to check that it's correct, with this contribution you get a tax savings of $17,143 x 30% = $5,143, which is the difference between the $17,143 you contributed and the desired $12,000 out-of-pocket cost.

The equivalent Roth contribution is $13,333 (=$12,000 / (1 - 10%)). (That's $12,000 after-tax, because $13,333 minus $1,333 tax savings is $12,000.)

Not exacty the way to do an IRR calc.

Max contribution per year is 12,000 to either account.

tIRA net cost is 12,000 - 3,600

Roth net cost is 12,000 - 1,200

Those are the options available based on what will happen
if the transaction occurs. The Roth costs more and the
tIRA costs less.

From either net cost you compute the IRR to see which is
best, whether you spend the tax credit on rent on invest
it in taxable equites. A taxable account isn't really
needed.

Each account will start with 12,000 balance and grow to a
future balance. The return will be based on net cost,
the amount actually paid for the investment. That's the
textbook way to do an IRR calc. You can add a taxable
investment later but dollar for dollar that's the IRR.

I prefer an IRR calc for all my investments whether an IRA,
annuity or whatever against all other methods because it is
more theoretically accurate PV wise.

Try it sometime.

[fyre4ce]

You're close, but you made one mistake. To compare apples-to-apples, you need to make the after-tax costs of the two options the same. (Otherwise, you'll need to track separately the difference in a taxable account, which will needlessly complicate it.)

Let's say you want to contribute $12,000 after-tax. At a 30% traditional tax rate, that's a contribution of $17,143 (= $12,000 / (1 - 30%)). If you want to check that it's correct, with this contribution you get a tax savings of $17,143 x 30% = $5,143, which is the difference between the $17,143 you contributed and the desired $12,000 out-of-pocket cost.

The equivalent Roth contribution is $13,333 (=$12,000 / (1 - 10%)). (That's $12,000 after-tax, because $13,333 minus $1,333 tax savings is $12,000.)

Not exacty the way to do an IRR calc.

Max contribution per year is 12,000 to either account.

tIRA net cost is 12,000 - 3,600

Roth net cost is 12,000 - 1,200

Those are the options available based on what will happen
if the transaction occurs. The Roth costs more and the
tIRA costs less.

From either net cost you compute the IRR to see which is
best, whether you spend the tax credit on rent on invest
it in taxable equites. A taxable account isn't really
needed.

Each account will start with 12,000 balance and grow to a
future balance. The return will be based on net cost,
the amount actually paid for the investment. That's the
textbook way to do an IRR calc. You can add a taxable
investment later but dollar for dollar that's the IRR.

I prefer an IRR calc for all my investments whether an IRA,
annuity or whatever against all other methods because it is
more theoretically accurate PV wise.

Try it sometime.

There's nothing wrong with calculating an IRR (provided that you include taxes on both ends, because that's what's at-issue here). The reason it's not more common in T vs R analyses is because much of the time, you assume the investment performance is the same in both cases, like when choosing between a tIRA and a Roth IRA at the same custodian. But, when investments and/or fees are different, it definitely makes sense to calculate an end-to-end IRR or future value.

It is important, though, to make sure the "initial conditions" are the same for both options. If you have an account that can accept a $12k investment either pre-tax or Roth (SIMPLE IRA?), the Roth starts off with a slight advantage because you can put more money in it. The nominal contribution is the same, but you're putting in after-tax dollars, which are worth more than pre-tax dollars. In your example, I'd suggest comparing the future value (or IRR) of $12k pre-tax plus $2,400 in a taxable account (the difference in taxes between the two choices) vs. $12k in a Roth account. If you don't account for the $2,400 difference in taxes, it won't be a fair comparison. There is a section in both pages (mine is here) that deals with this situation of contributing the maximum. The math isn't super-simple, but it's there for those who are interested.

[patrick013]

My spreadsheet from 2017 does all of that and I assume
a uniform long term return. Actually different calc's for
stocks and bonds in different Ira's leads to better decisions
for AA particularly bonds in tIRa and stocks in Roth.

No doubt Roth will have a higher net cash flow if one doesn't
need the tax credit than a stand alone tIRA while the tIRA
has a higher IRR with lower tax rate accounts but the option
with the higher IRR will suit most investors or the option
with net cash flow like on the spreadsheet.

Main point is the net cost is the contribution - tax credit
similar to the application of investment tax credit to fixed
asset purchase analyses commonly used in corp finance
to determine net cost there and for proper investment presentation.

You could use IRR exclusively for IRA's and get decisive
answers for different flowchart decision points. I use it
for everything and find it more inclusive than one line
equations. So it's preferable here and makes a good presentation.

That's all I know about it.

Have a good one.

[FiveK]

4. Tried to standardized calculations. Where possible, calculations are now in terms of a future tax rate on withdrawals, and use an inequality rather than an equals sign to make explicit when T or R is better. Tried to standardize variable names, and make consistent with formulas used in other parts of the wiki.

Looks fine.

7. Changed the terminology for the T vs R analysis to be a choice of "tax structure."

People go to Fidelity, Vanguard, etc. to open "accounts", not "structures". Leaving the current language would likely cause less confusion.

8. A couple analysis issues that are controversial (so far as I know), such as an asymmetric risk of traditional or roth, are now described as controversial and include links to relevant discussions in the forums. If these are actually settled questions (or if we settle them here) then these can be changed.

Having just reread the whole Asymmetric risk in the Roth vs Traditional question - Bogleheads.org thread, the main point seemed non-controversial. As long threads are wont to do, that one wandered into various side-disagreements but it doesn't seem necessary to throw gas on this fire.

9. Changed "Common misconceptions" header hierarchy. It makes much more sense for this section to be a child of "Calculations" than as the parent of current and future marginal rates. Actually, I assume this is an error in the current page.

Agreed that the current parent/child structure isn't great. Might be best to keep the "Common misconceptions" section by itself, neither a parent nor child.

It is very useful in forum threads simply to refer to "...the first (or second) of two Common misconceptions" when someone has one of them. That's a neutral observation, and doesn't affect the parent/child/standalone choice.

All for now, more later.

[Iorek]

I am suggesting deleting the examples completely. I think the level of detail on the current page is about right for a wiki overview of "general considerations".

If you really want to add more I think you could create a new section or possibly a new paragraph at the end of the current section which says something like "Some examples of situations where you might find yourself in a higher tax bracket in retirement than when saving are: [insert list of examples, such as military personnel, especially when receiving tax-free pay; young people who expect significant income growth later in their career; people with other significant sources of retirement income, etc]."

I think that would do better to get people thinking about the issue in a useful way and it would avoid us having to debate how likely it really is that social security and/or a pension (and/or RMDs) will really bump people to a higher marginal tax rate in retirement.

Several reviewers in this thread liked my idea of moving the cases toward the top (and of course I did too). But I want to consider all opinions here. I made some edits to the draft page based on what you suggest. I separated out the cases and also tweaked the language of the retirement income case. Let me know if you think this is better, or at least a compromise you think you can live with. Thanks!

I do think separating out the cases helps, thanks.

I could probably nitpick about how to write the cases -- for example I'd probably include somewhere military receiving tax-free pay and I'd think about whether people might read the retirement income case as suggesting that anyone receiving social security would be better off in Roth-- but those are pretty minor and reasonable minds can disagree.

[fyre4ce]

Also, there are many inconsistencies with the capitalization of "traditional" throughout the page. I believe that all of the instances should be lowercase unless (1) it is the beginning of a sentence, (2) it is the title/link of another page, or (3) it is used inside a table.

Thanks. I’ll take a look and clean this and your other comment up tomorrow.

Updates made, thanks!

[fyre4ce]

People go to Fidelity, Vanguard, etc. to open "accounts", not "structures". Leaving the current language would likely cause less confusion.

I get what you're saying. I made that change to emphasize that the different tax options can be had independently of the different account. I'd be interested to hear other opinions on this one, and will keep an open mind.

Regardless of where we end up on this, I do think it's important to say that this is a tax planning question, which will help contextualize it for readers.

Having just reread the whole Asymmetric risk in the Roth vs Traditional question - Bogleheads.org thread, the main point seemed non-controversial. As long threads are wont to do, that one wandered into various side-disagreements but it doesn't seem necessary to throw gas on this fire.

It's not controversial that lower-than-expected future pre-tax balances will (usually - tax spikes excepted) result in higher-than-expected after-tax availability of funds. What is controversial is how significant this effect is, and whether it should be accounted for in planning. I think not; I remember doing an analysis that said that for every $1 short you were on pre-tax funds, you'd get back $0.10 in tax savings to compensate. That doesn't seem like much. And the theoretical maximum for this is somewhere around $0.25, if you go from a ~20% average tax rate (huge) to 0. I think a better idea is to plan for your nominal case (adjusted every year of course, as circumstances change) and save 10% more than you would otherwise (eg 15% to 16.5% of income). That way, you'll keep get to keep your money in the nominal case, rather than lose it to paying a higher tax rate through some kind of biased tax plan. But, I can see how there could be other opinions. My edits on this topic (I think) capture the question fairly. Are there authorities that recommend a traditional bias for this reason?

Edit: I was wrong about a couple things in this post. First, for a small shortfall in pre-tax funds, the tax savings you get to compensate for this shortfall is simply your marginal tax rate; it has nothing to do with an average tax rate. So if you're in the 24% tax bracket in retirement, and you have a shortfall compared to your plan of $100 (due to lower contributions or investment performance than planned) then the government will compensate you $24. The average tax rate would only come into play if you're talking about a large shortfall where you drop to near $0 pre-tax dollars; then, the higher brackets get averaged in with lower brackets and you get an average tax savings of less than your marginal.

Second, the theoretical maximum average tax rate, in retirement or otherwise, is simply the rate on the top bracket. That's the average (and marginal) rate you'd pay with infinite income. Now, to be fair, it takes huge amounts of income to get close to this value. $400k of income (which could be generated from a $10M pre-tax account, a reasonable upper limit unless you're ultra-wealthy) incurs only a 21.8% average tax rate, which is what I was trying to say. But it's not correct to say this is a theoretical maximum is ~20%.

I still like the wording in my update as it's written, but I'll take a closer look at this when I have time over the next few days.

Agreed that the current parent/child structure isn't great. Might be best to keep the "Common misconceptions" section by itself, neither a parent nor child.

It is very useful in forum threads simply to refer to "...the first (or second) of two Common misconceptions" when someone has one of them. That's a neutral observation, and doesn't affect the parent/child/standalone choice.

Sounds good to me.

[fyre4ce]

Bumping this up for more comments.