Estimating Roth Conversions - How close to get?

[moustachebar]

Every year I mess up my Roth conversion amounts.
One year I was way off because I was converting in batches whenever the market would drop by a certain %, and forgot one batch, and overconverted. It pushed me into the next bracket and made my capital gains taxable, so I made an IRA contribution to get back under the bracket. Needless to say, defeats the purpose by making that money inaccessible again.
This year I got close but was over by $380, resulting in $35 in tax (15%) on qualified dividends. I had delayed Roth conversion until well into December to have good estimates of interest and dividends, but messed up somewhere, probably forgetting modest 'other income'.
I use the published brackets and the Go Curry Cracker calculator (the errors are mine, the calculator is great).
It's a small amount but here's my question:
How close do you folks get, how do you do it, and do you think it's better to be a little over or a little under?

[SeattleCPA]

I would be very surprised if you're saving much money at all if you're talking $35 here and there.

Roth IRAs and backdoor Roth IRAs have to be on the top ten list of overrated tax planning ideas.

You want to compare your marginal rate today to the marginal rate you'll pay later. You don't win with a Roth IRA unless you'll pay more later when you're retired than you do today.

If you're married, and assuming you draw 4%, you wouldn't even pay income taxes on the first $30K of income you realize due to the standard deduction. That suggests you could have about $750K in a regular IRA paying $30,000 a year before you'd even start paying federal income taxes. ($30K/4% equals $750K.)

The 10% bracket for a married filing joint taxpayer is $23,200. That suggests you have another $580K paying out $23,200 and only be paying that 10%. ($23,200/4% equals $580K.)

Finally, the 12% bracket for a married filing joint taxpayer is $71,100. You could have another $1,777,500 paying out that $71,100 and be in the 12% bracket. ($71,100/4% equals $1,777,500.)

So to get into the 22% bracket, you'd need in excess of $3M?

BTW keep in mind that's $3M in today's dollars. Not for example 2040 dollars. Because of the inflation.

[Sandi_k]

So to get into the 22% bracket, you'd need in excess of $3M?

BTW keep in mind that's $3M in today's dollars. Not for example 2040 dollars. Because of the inflation.

Or...have a pension. Or Social Security income. Or rental income.

There are many of us who will not be able to control our income in retirement. And IRMAA is a thing. And the tax rates for surviving spouses are also heinous.

To the OP: I have a spreadsheet that includes all gross income, minus pre-tax deductions and standard deductions, plus estimated interest income, to reach taxable income.

I then have a manual set of multipliers, for 10/12/22/24% brackets.

For this year and next, I am trying to get to $248k taxable income with the conversions, so that if the interest income is higher than planned, I don't fall into the NIIT trap.

For 2026, we have the potential reverting tax brackets (25/28/33%) *and* IRMAA calcs for my spouse. So yeah - 2024 and 2025 are where we're making some big moves in preparation for retirement.

[moustachebar]

@SeattleCPA  that's a great illustration.
I halfheartedly contributed to Roths when working just on the idea that flexibility would be good, but always frontloaded 401k.
Now we am RE except bringing in about what we spend, so we're right between accumulation and drawdown. During that time we're doing Roth conversions to ladder them, for spending up to 59.5 if needed, but also for estate reasons.

@Sandi_k Good points. We don't have the 3M described but will have SS and could end up with rental income - lot can happen in 30+ years. Medical costs in retirement are an even bigger unknown and I have a vague sense that it's easy to end up on the hook for potentially ruinous expenses with Medicare if we don't know what we are doing. Expiration of TCJA is one reason to do the Roth conversions now, getting started on a ladder is another, and expiration of me is a third!

Want to leave SO in a better spot; as you note, having to suddenly file as single makes a big difference. Also leaving Roth to a kid is a lot less of a pain for them than an IRA. Our family has never ever had anything to pass down. It made it quite a bit rougher than for friends who had backup in the form of family money even if never lent. If I happen to not be around I'd like my kid to be in that slightly less fraught position.

I haven't been up where NIIT exists but I see it applies at $250k, so you leave yourself $2k wiggle room. Obviously higher stakes than my 15% on minimal cap gains! I guess from all this I gather it's OK to overshoot if the consequences are minimal, and if they are significant, you are more cautious. Makes sense.

I like the spreadsheet idea, I will convert my scrawls to one. Not sure why I've been doing it the hard way, thanks for the nudge.

[secondcor521]

Every year I mess up my Roth conversion amounts.
One year I was way off because I was converting in batches whenever the market would drop by a certain %, and forgot one batch, and overconverted. It pushed me into the next bracket and made my capital gains taxable, so I made an IRA contribution to get back under the bracket. Needless to say, defeats the purpose by making that money inaccessible again.
This year I got close but was over by $380, resulting in $35 in tax (15%) on qualified dividends. I had delayed Roth conversion until well into December to have good estimates of interest and dividends, but messed up somewhere, probably forgetting modest 'other income'.
I use the published brackets and the Go Curry Cracker calculator (the errors are mine, the calculator is great).
It's a small amount but here's my question:
How close do you folks get, how do you do it, and do you think it's better to be a little over or a little under?

Messing up is always possible, of course.

I do several things to minimize the risk:

1.  I keep track of my income taxes throughout the year in a couple of Word documents and Excel spreadsheets.  "Don't forget the $200 you earned on <gig X>".  That way I am less likely to forget something tax-related.

2.  I use the Case Study Spreadsheet (can be found in a thread here on the forums), which handles all of my tax complexity including ACA subsidies and produces a nice graph to help me decide what AGI/TI to target.

3.  I convert about 1/3 of my tentative Roth target in January, 1/3 during the year if the market drops significantly, and then I do the final 1/3 at the end of December to "top up" once I use the CSS to figure it exactly.

4.  In December, when I'm doing that last 1/3, I used to convert 90% of my remaining 1/3 target, then wait and see what the actual conversion amount was, then repeat the 90% of the remaining.  Do that for two or three days and you can home in on the target pretty easily without going over.  The reason for this is that I Roth convert Vanguard MF shares in-kind, and the value is as of closing *after* you put the conversion order in.  So market fluctuations that day can change the amount a bit.

5.  I have an HSA / HDHP plan.  In January or February I sit down and do my final actual tax return.  If I'm over my target, I can contribute more to my HSA for the previous year, and thus I can dial in my AGI/TI to the exact dollar I want.  I find this optimization fun; so much so that I deliberately don't max out my HSA each year just so I can do this trick.  (I don't need to max out my HSA each year to get it to where I want it.)

To answer the other parts of the question, I think it is best to compare the reward of Roth converting those last few dollars below the cutoff, against the cost of converting a few dollars above.  And take into account your ability to hit targets.  In the case of cliffs (like ACA repayment or IRMAA cliffs), the risk can be asymmetric so it is better to be under by a bit, or a lot if your ability to hit targets is low.  If you're dealing with the difference between the 22% and 24% brackets, or even NIIT, a few percentage points on a few hundred or even a few thousand dollars over or under likely isn't going to matter in the big scheme of things, so just do whatever is most convenient.

[moustachebar]

Wow, @secondcor521 thank you for that amazing response. I am happy to see a good strategy and guiding principles! And you now have me thinking about cliffs that I haven't thought about yet, so thank you for that.

I like your process. I also used to split it into thirds but didn't follow your process #1 so forgot some capital gains, ouch. But that last third is aggravatingly inaccurate! So the elegant spiral of conversions at year end is a great idea. A year end countdown for mustachians!

We max the HSA now, but could potentially not for extra fun and precision during the late winter doldrums.

I had no idea there was this spreadsheet, thank you! There is some serious good stuff in there, looks really useful for all kinds of big decisions!

Thank you again!

[SeattleCPA]

Some comments to respond to folks.

First, @Sandi_k , if you're doing the math, you're doing it right. That's really my point. Generalized rules of thumb--and I see this dozens of times every tax season, result in suboptimal decisions. People need to do the math. BTW, in addition to things like IRMAA, you want to think about NIIT, tax loss harvesting of taxable account info, step-ups in basis of taxable accounts, gifting to heirs, qualified charitable distributions in some cases, opportunities to invest cleanly in alternative asset investments (which some investors will have access to at points in their life), etc. There are are lots of minor but significant factors in addition to marginal rates.

Second, a general comment: One wants to be careful to adjust for the effect of inflation. If you're using nominal rates of return to calculate your future values  you need to inflate the tax brackets so you're not applying 2024 tax brackets to 2034 or 2044 income. To make the math easy, if you're talking about a balance in 24 years and inflation runs 3%, you need to be looking at not $3M but $6M. BTW I agree you should use the "post-Trump-tax cut" rates for 2026 and beyond. Those are even more important to adjust for inflation because they're been so much in last four years.

Third, to add to something that @secondcor521 mentions, if your income is high or you're young, you will pay NIIT. The NIIT and Medicare surtax  thresholds aren't adjusted for inflation. BTW, I honestly thought the 2.9% Medicare tax would disappear probably before President Biden left office. Depending on inflation and when he leaves office, I'm guessing that's less likely now... but one wants to plan for that I think.

A fourth general comment: I know people want to plan to succeed with their investment plan and get a nice big best egg... and that we all hope returns are as good in the future as they've been in the past... but I think not just for Roth account analyses but for other retirement planning, we want to consider the possibilities your accumulation phase will get cut short and the possibility returns will be much lower. Those assumptions probably doom Roth style accounts for almost everybody.

[moustachebar]

@SeattleCPA Thanks for these points. I don't have a handle on most of the topics you bring up in your first paragraph but am vaguely aware of TLH and step-up in basis.

So when you say Roth accounts are doomed, you mean comparatively? I had come to that conclusion from reading GCC as well.

I stumbled along without any clear plan while working, and only accumulated a small Roth because of low contribution limits. Being quite afraid Social Security would be gone by my time, I maxed 401k contributions, even with very low salary (lentils, etc). So I never accumulated any in taxable - there wasn't any left over. My Roth conversions are partially trying to make up for that, building a ladder I can access in RE. And partially for estate purposes. So I think Roth is a decent option in this fringe case. But always open to other options! I am looking into SEPPs but would rather not be locked in for too long.

If I had it to do again I'd put more in taxable. Though it was likely helpful to have locked it away vs. being able to raid it.

[Sandi_k]

I disagree that Roths are doomed. Why would they be? All sorts of tax rules evolve, so it makes sense to plan for a multitude of future cases.

And they do make sense if you:

1) Already have a stash that is pre-tax.
2) Believe that tax rates will go higher in the future (currently scheduled for 2026)
3) Have a pension, so you cannot control your retirement income.
4) Have substantial differences in age or health.
5) Have looked at tax brackets for singletons.
6) Will have enough accumulated where IRMAA brackets are in play.

In short, Roths are a hedging bet. We've done the math, and they make sense for us.

[SeattleCPA]

@Sandi_k sorry "doom" was a poor word choice.

What I do think is for the great majority of people, Roth IRAs as an alternative to traditional IRAs and Roth 401(k)s as an alternative to traditional 401(k)s don't make sense.

First, most people never save enough to have to worry much about taxes in retirement as compared to taxes during their higher income working years. The median retirement savings for someone in their 60s or 70s is right about $200,000-ish? Less than 20% of people at that age have $500K or more in savings. One definitely doesn't need to worry about taxes with $200K (so the average person) and really not with $500K either (so someone at the 80th percentile).

Second, most people work because they make more money working than they'd "make" by retiring. That would suggest most people see both their taxable income and tax rates drop after they quit working. And the marginal tax rate thing is the big factor. BTW I'm happy to include IRMAA, change in filing status, etc. But we also want to include other factors like basis step-up, TLH, etc.

All that said, I think you have to run the numbers. There are situations where Roth accounts make sense. (Lots of situations in terms of absolute numbers in a country of 340 million people or whatever.)

But most people shouldn't do a Roth. Most people should save more and using a tax-deferred investment account.

[Sandi_k]

@SeattleCPA - again, I disagree.

Folks here on MMM are not "most people."

- They save more than average.

- They will still have the reality of IRMAA tiers, given Social Security income plus investment income.

- The tax brackets have a big impact once one becomes a singleton, through death or divorce. It's too late to plan once you're in your mid-60's.

[SeattleCPA]

@SeattleCPA - again, I disagree.

Folks here on MMM are not "most people."

- They save more than average.

- They will still have the reality of IRMAA tiers, given Social Security income plus investment income.

- The tax brackets have a big impact once one becomes a singleton, through death or divorce. It's too late to plan once you're in your mid-60's.

We obviously disagree. But to make this less contentious, I'd be curious where you see the breakpoint. I.e., I think at anything less than $1M in savings and in current day dollars, a Roth is a bust. In excess of $10M? Sure. It maybe works. Especially if you're just hedging.

However, I actually think the cutoff is higher than $1M... but where do you see the cutoff? Honestly curious. And again, we're talking 2024 dollars.

I might say if you're facing IRMAA you want to consider a Roth. But those limits are pretty high. The IRMAA affects single filers at roughly $100K married at roughly $200K. Those are 2024 dollars not 2034 or 2044 dollars. In effect, the IRMAA adds as much as $3K or $4K a year per person to your tax bill? Which effectively bumps up your retirement years tax rate a bit... but seems like you'd want to do the math. Paying an extra $8K in Medicare if you're talking (again in current day dollars) $400K of retirement income is a 2% bump in your retirement years federal rate.

[EliteZags]

so you're actually trying to say Roth is just useless for those that don't make enough to max their pretax contributions, and won't have enough in retirement to be affected by taxes

ok so most of this board checking out of the conversation

[moustachebar]

Thinking about the reasons I do Roth conversions:

1) Invested up to the limit of pretax when working. Didn't intend to RE, so trying to access that cash. Options include Roth conv or SEPP AFAIK. This leads me to convert at least as much as I spend, which would be under the top of the 10%.
2) Better to leave to my kid than a deductible IRA. I'm going to pay the tax (later) anyway, do it now and make the asset more valuable to them, also avoid the expense of life insurance. This leads me to convert to the top of 12%. I want to pile up a chunk that, if not needed for a major expense, can be passed on without the worry that it's upsetting their taxes.

I expect my marginal rate in retirement to remain around about the same, I guess, but my kid's to be higher.

@SeattleCPA Does either reason make sense to you?

[Sandi_k]

@SeattleCPA - again, I disagree.

Folks here on MMM are not "most people."

- They save more than average.

- They will still have the reality of IRMAA tiers, given Social Security income plus investment income.

- The tax brackets have a big impact once one becomes a singleton, through death or divorce. It's too late to plan once you're in your mid-60's.

We obviously disagree. But to make this less contentious, I'd be curious where you see the breakpoint. I.e., I think at anything less than $1M in savings and in current day dollars, a Roth is a bust. In excess of $10M? Sure. It maybe works. Especially if you're just hedging.

I don't think this is contentious - we just have different experiences.

Again, I have a pension, we have savings, we live in CA, and we will both have substantial SocSec. And even though we're pretty low-average income for the Bay Area, we will pay MORE in taxes once retired, because we won't have pre-tax options for savings, or for medical premiums. And with the pension and Social Security, withdrawing from investment retirement accounts means at a minimum, we'll be in the (currently 24%, will be 28% or higher) tax bracket.

So $1M = $40k in income, plus a six figure pension, plus Social Security. IRMAA tiers will have us at least in the second tier. And if we inherit an IRA from DH's mother? With a 10 year withdrawal mandate? We will be very happy to have Roth accounts while we draw that IRA down, and then revert to our pre-tax accounts in Year 11.

I'm not saying that all of our accounts should be converted. But I'd be thrilled if we can convert and contribute enough to Roth accounts, prior to DH's IRMAA snapshot, so that we're ~ 35% Roth and 65% pre-tax.

However, I actually think the cutoff is higher than $1M... but where do you see the cutoff? Honestly curious. And again, we're talking 2024 dollars.

I *am* talking in 2024 dollars. Seriously.

I might say if you're facing IRMAA you want to consider a Roth. But those limits are pretty high. The IRMAA affects single filers at roughly $100K married at roughly $200K. Those are 2024 dollars not 2034 or 2044 dollars.

Dude, I am aware of the IRMAA tiers, and where those thresholds are. Did you think I raised them as a concern WITHOUT knowing those thresholds?

You've proven my point. A singleton can EASILY have $100k in income, with Social Security and 4% withdrawals. Add in any kind of a pension, or an inherited IRA, and you're in those IRMAA tiers.

In effect, the IRMAA adds as much as $3K or $4K a year per person to your tax bill? Which effectively bumps up your retirement years tax rate a bit... but seems like you'd want to do the math. Paying an extra $8K in Medicare if you're talking (again in current day dollars) $400K of retirement income is a 2% bump in your retirement years federal rate.

Again, I *have done the math.* And yes, in 2024 dollars. Do you have any idea how condescending you are sounding here?

You're also not including NIIT of an additional 3.8% on incomes over $250k - which is NOT INDEXED for inflation.

For all of these reasons, Roths make sense for US. And for any other couple who might have $200k of income in retirement, through a variety of sources, including rental income, inherited IRAs (now that the stretch provisions have been eliminated), pension, Social Security, and pre-tax retirement accounts.

[Sandi_k]

Thinking about the reasons I do Roth conversions:

1) Invested up to the limit of pretax when working. Didn't intend to RE, so trying to access that cash. Options include Roth conv or SEPP AFAIK. This leads me to convert at least as much as I spend, which would be under the top of the 10%.

2) Better to leave to my kid than a deductible IRA. I'm going to pay the tax (later) anyway, do it now and make the asset more valuable to them, also avoid the expense of life insurance. This leads me to convert to the top of 12%. I want to pile up a chunk that, if not needed for a major expense, can be passed on without the worry that it's upsetting their taxes.

I expect my marginal rate in retirement to remain around about the same, I guess, but my kid's to be higher.

@SeattleCPA Does either reason make sense to you?

The legacy issue raised here is smart - being able to NOT have RMDs on the Roth account, and being able to pass it on to your heirs tax free is very strategic.

[Sandi_k]

And for those who want to be led through the assessment of whether a Roth conversion is worth it for you given Medicare, IRMAA thresholds and Social Security, The Finance Buff[/i] has an article that gets three thumbs up on Bogleheads:

https://thefinancebuff.com/roth-conversion-social-security-medicare-irmaa.html

[SeattleCPA]

Thinking about the reasons I do Roth conversions:

1) Invested up to the limit of pretax when working. Didn't intend to RE, so trying to access that cash. Options include Roth conv or SEPP AFAIK. This leads me to convert at least as much as I spend, which would be under the top of the 10%.
2) Better to leave to my kid than a deductible IRA. I'm going to pay the tax (later) anyway, do it now and make the asset more valuable to them, also avoid the expense of life insurance. This leads me to convert to the top of 12%. I want to pile up a chunk that, if not needed for a major expense, can be passed on without the worry that it's upsetting their taxes.

I expect my marginal rate in retirement to remain around about the same, I guess, but my kid's to be higher.

@SeattleCPA Does either reason make sense to you?

If your marginal rate is the same now as later, or that's your best guess, the tax burden of a Roth and a non-Roth account are equal. So given that equivalence? I think you'd look at other non-tax factors. E.g., i think it's fine to have some Roth money just to have some tax diversification.

[SeattleCPA]

Medical costs in retirement are an even bigger unknown and I have a vague sense that it's easy to end up on the hook for potentially ruinous expenses with Medicare if we don't know what we are doing. Expiration of TCJA is one reason to do the Roth conversions now, getting started on a ladder is another, and expiration of me is a third!

One thing about medical costs to consider is you may be able to draw dollars from a traditional IRA to pay for something like long-term care and then partially net that taxable income with the long-term care expenses deductions. This is obviously a worse case scenario.

I haven't been up where NIIT exists but I see it applies at $250k, so you leave yourself $2k wiggle room.

NIIT isn't indexed so lots of people will end up paying it in the future. Especially if they have a long retirement.

[secondcor521]

I've done the math for myself, using my home grown spreadsheet.  I admit it might have errors, but I've reviewed it several times over the years and believe it to be accurate.

My spreadsheet goes to age 95 and is in nominal dollars.  It accounts for the following:

1.  Investment growth on my trad IRA.
2.  RMDs on my trad IRA using RMD factors according to table 3.
3.  A likely inherited trad IRA and the 10 year drawdown rule.
4.  Roth conversions based on the different tiers (top of tax brackets and IRMAA thresholds).
5.  Social Security at age 70 with the possible haircut and the 85% taxation rule (my spreadsheet assumes 85% on all, which is on the high side, I know).
6.  The standard deduction, adjusted for inflation.
7.  Ordinary income tax brackets, IRMAA brackets, and IRMAA surcharges, all adjusted for either inflation or the historical growth rate.

This should get me in the neighborhood.  Mainly what I do each year is update the various constants in the spreadsheet once a year and then use that to figure out what my federal marginal plus IRMAA rate is at age 85, which for me is about my 50/50 life expectancy based on the Social Security Longevity Tool.

What this tells me is that I should be converting up into the mid-20%s now to avoid 32% and Tier 4 IRMAA later.  I usually don't go that far for a number of reasons.

The biggest uncertainties in my analysis are the growth rate of IRMAA premiums.  These are adjusted, as I understand it, based on the actual costs of the Medicare program so that it is funded at the appropriate levels.  Right now I use a factor of 7.7%, which seems plausible, but compounding that over time has my IRMAA surcharges alone (not base B+D) at over $100K at age 95, making it the equivalent of about a 10% additional marginal tax rate on top of the 32%.

And that's with doing Roth conversions every year between now and then.

[SeattleCPA]

I've done the math for myself, using my home grown spreadsheet.  I admit it might have errors, but I've reviewed it several times over the years and believe it to be accurate.

My spreadsheet goes to age 95 and is in nominal dollars.  It accounts for the following:

1.  Investment growth on my trad IRA.
2.  RMDs on my trad IRA using RMD factors according to table 3.
3.  A likely inherited trad IRA and the 10 year drawdown rule.
4.  Roth conversions based on the different tiers (top of tax brackets and IRMAA thresholds).
5.  Social Security at age 70 with the possible haircut and the 85% taxation rule (my spreadsheet assumes 85% on all, which is on the high side, I know).
6.  The standard deduction, adjusted for inflation.
7.  Ordinary income tax brackets, IRMAA brackets, and IRMAA surcharges, all adjusted for either inflation or the historical growth rate.

This should get me in the neighborhood.  Mainly what I do each year is update the various constants in the spreadsheet once a year and then use that to figure out what my federal marginal plus IRMAA rate is at age 85, which for me is about my 50/50 life expectancy based on the Social Security Longevity Tool.

What this tells me is that I should be converting up into the mid-20%s now to avoid 32% and Tier 4 IRMAA later.  I usually don't go that far for a number of reasons.

The biggest uncertainties in my analysis are the growth rate of IRMAA premiums.  These are adjusted, as I understand it, based on the actual costs of the Medicare program so that it is funded at the appropriate levels.  Right now I use a factor of 7.7%, which seems plausible, but compounding that over time has my IRMAA surcharges alone (not base B+D) at over $100K at age 95, making it the equivalent of about a 10% additional marginal tax rate on top of the 32%.

And that's with doing Roth conversions every year between now and then.

Some comments about the above.

1. I think the assumption that 85% of your Social Security will be taxable is a good one for anyone who's done or plans to do a good job saving. Those $25K and $32K amounts (from 1984!?) don't get adjusted for inflation.

2. Agree that inherited IRA accounts create a planning challenge.

3. I can't imagine the top tier IRMAA adjustment going from roughly $400 a month to $8000 a month... or at least not unless there's just massive inflation that scrambles all of these numbers.

4. If you're married and looking at keeping out of the current 32% federal bracket (so more than $400K of AGI if married and more than $200K of AGI if single), that puts you into a pretty rare group. You know that. But for others reading this, understand there maybe a pretty big difference between @secondcor521 's situation and yours or mine.

5. A more general comment: Other ways exist to deal with high marginal rates in retirement other than Roth-style accounts: Changing state residency, home ownership, active real estate investments, strategically allocating taxable investments so a big chunk of your retirement income (the last bits) are qualified dividends or long-term capital gains, family income splitting, etc.

Also just to say this, I don't have a problem with Roth-style accounts per se. I have problem with people applying a rule of thumb that says, "Hey I don't want to pay taxes in retirement... so my financial planner has me doing Roth conversions." Clearly that's not the case with @Sandi_k or @secondcor521 and maybe two or three million other folks . But I see clients all the time who will retire with best case a $2M portfolio and, say, good Social Security benefits and are trying to load up their Roth-style accounts. And the problem is when they're making good six figure salaries as a household, the taxes they want to save or avoid are the ones they're paying during their higher income working years.

Note: My calculation of what you need to end up with a $2M-ish portfolio in real terms. Max 401(k) for three decades while making $200K a year and getting a safe harbor 4% match. Assuming a five percent real return. The Excel formula =FV(0.05,30,23000+.04*200000) returns roughly $2M.

With this 30-year earnings history, you'll get a $46K SS benefit if single, $69K if married with spousal benefits. If you draw $80K from your $2M IRA, your AGI is roughly $150K. That's $50K less than your earnings.

[secondcor521]

I've done the math for myself, using my home grown spreadsheet.  I admit it might have errors, but I've reviewed it several times over the years and believe it to be accurate.

My spreadsheet goes to age 95 and is in nominal dollars.  It accounts for the following:

1.  Investment growth on my trad IRA.
2.  RMDs on my trad IRA using RMD factors according to table 3.
3.  A likely inherited trad IRA and the 10 year drawdown rule.
4.  Roth conversions based on the different tiers (top of tax brackets and IRMAA thresholds).
5.  Social Security at age 70 with the possible haircut and the 85% taxation rule (my spreadsheet assumes 85% on all, which is on the high side, I know).
6.  The standard deduction, adjusted for inflation.
7.  Ordinary income tax brackets, IRMAA brackets, and IRMAA surcharges, all adjusted for either inflation or the historical growth rate.

This should get me in the neighborhood.  Mainly what I do each year is update the various constants in the spreadsheet once a year and then use that to figure out what my federal marginal plus IRMAA rate is at age 85, which for me is about my 50/50 life expectancy based on the Social Security Longevity Tool.

What this tells me is that I should be converting up into the mid-20%s now to avoid 32% and Tier 4 IRMAA later.  I usually don't go that far for a number of reasons.

The biggest uncertainties in my analysis are the growth rate of IRMAA premiums.  These are adjusted, as I understand it, based on the actual costs of the Medicare program so that it is funded at the appropriate levels.  Right now I use a factor of 7.7%, which seems plausible, but compounding that over time has my IRMAA surcharges alone (not base B+D) at over $100K at age 95, making it the equivalent of about a 10% additional marginal tax rate on top of the 32%.

And that's with doing Roth conversions every year between now and then.

Some comments about the above.

1. I think the assumption that 85% of your Social Security will be taxable is a good one for anyone who's done or plans to do a good job saving. Those $25K and $32K amounts (from 1984!?) don't get adjusted for inflation.

2. Agree that inherited IRA accounts create a planning challenge.

3. I can't imagine the top tier IRMAA adjustment going from roughly $400 a month to $8000 a month... or at least not unless there's just massive inflation that scrambles all of these numbers.

4. If you're married and looking at keeping out of the current 32% federal bracket (so more than $400K of AGI if married and more than $200K of AGI if single), that puts you into a pretty rare group. You know that. But for others reading this, understand there maybe a pretty big difference between @secondcor521 's situation and yours or mine.

5. A more general comment: Other ways exist to deal with high marginal rates in retirement other than Roth-style accounts: Changing state residency, home ownership, active real estate investments, strategically allocating taxable investments so a big chunk of your retirement income (the last bits) are qualified dividends or long-term capital gains, family income splitting, etc.

Also just to say this, I don't have a problem with Roth-style accounts per se. I have problem with people applying a rule of thumb that says, "Hey I don't want to pay taxes in retirement... so my financial planner has me doing Roth conversions." Clearly that's not the case with @Sandi_k or @secondcor521 and maybe two or three million other folks . But I see clients all the time who will retire with best case a $2M portfolio and, say, good Social Security benefits and are trying to load up their Roth-style accounts. And the problem is when they're making good six figure salaries as a household, the taxes they want to save or avoid are the ones they're paying during their higher income working years.

Note: My calculation of what you need to end up with a $2M-ish portfolio in real terms. Max 401(k) for three decades while making $200K a year and getting a safe harbor 4% match. Assuming a five percent real return. The Excel formula =FV(0.05,30,23000+.04*200000) returns roughly $2M.

With this 30-year earnings history, you'll get a $46K SS benefit if single, $69K if married with spousal benefits. If you draw $80K from your $2M IRA, your AGI is roughly $150K. That's $50K less than your earnings.

Thanks for the comments / reply, @SeattleCPA.  I think you know this, but for everyone else playing along, my motivation in this conversation is to confirm my understanding of my own situation and that I've analyzed it approximately accurately and that the resulting conclusion is reliable.  In my case, in a nutshell, the conclusion is that I should be doing Roth conversions up to at least 20% (federal marginal) now so as to avoid ~35% (federal marginal) later.

Specifically on the top tier IRMAA adjustment number, since that's one that I specifically wonder about:  I am compounding $384.30 + $74.2 = $458.50 per month for IRMAA Part B and Part D at 7.7% annually for 40 years which gives me a result of $8,911.84 per month.  I'd appreciate anyone checking my math.  Again, my understanding is that the IRMAA brackets are adjusted by CPI or similar, but IRMAA premiums are adjusted by Medicare costs.

Even though I run my RMD spreadsheet out 40 years, I am 54 now and I might only end up in the 32% bracket in my mid-90s, and I only spill over into the bottom edge of the 32% bracket.  I might not live that long, or tax policy might change, or economics might change, or my investment performance may decline, so it might not be an issue.  What does seem to be clear to me is that compounding over 40 years means that my spreadsheet is really a long term planning guess, not a long term planning tool.

I also noticed a SS discrepancy:  My SS benefit at age 70 is listed now at $45K per year per the SS website.  My understanding is that means that is in 2024 dollars, so I inflate that in my spreadsheet between now and age 70, which I'm fairly certain is correct.  I didn't make anywhere near the earnings history you describe, so someone is off somewhere by perhaps a factor of 2 or 3.  Thoughts?

[SeattleCPA]

Specifically on the top tier IRMAA adjustment number, since that's one that I specifically wonder about:  I am compounding $384.30 + $74.2 = $458.50 per month for IRMAA Part B and Part D at 7.7% annually for 40 years which gives me a result of $8,911.84 per month.  I'd appreciate anyone checking my math.  Again, my understanding is that the IRMAA brackets are adjusted by CPI or similar, but IRMAA premiums are adjusted by Medicare costs.

So first know (or hear again) that I have a ton of respect for your tax skills and financial analytical skills. You've displayed those a ton over the years. Or longer. :-)

But to address the issue of adjusting the IRMAA bump for inflation--and I don't have your spreadsheet's formulas--but it seems like you'd want to check your inflation adjustments on the tax brackets too. I.e., at some point, if one assumes that medical costs just continue to increase at a rate two or three times the general inflation rate, it seems like they become such a big component of everybody's spending that the overall inflation rate should be adjusted by a way higher number too. I.e., that medicare inflation and general price level information should converge. Practically, not sure how you address that in a spreadsheet. But it seems like you need to model either a lower IRMAA "inflation rate" or a higher tax-bracket "inflation rate."

FWIW I do think (and this is another tricky issue) that it makes sense to assume the current tax laws apply. Mostly because it's so hard to predict what'll happen. And that means assuming the current rates expire in 2026 and return to the pre-2018 rates... but with inflation adjustments so all of those amounts move steadily higher.

I won't keep drawing out my commenting in this thread, but I think I think Roth-style accounts work more poorly than many here do because:
1. I'm assuming low-ish real rates of return going forward. E.g., like Vanguard's forecasts.
2. I think one has other good ways to avoid RMD troubles. (Tax efficient taxable accounts work really well for some people.)
3. I worry more about other things (sequence of returns risk) than I do about RMDs or paying IRMAAs.

[secondcor521]

Specifically on the top tier IRMAA adjustment number, since that's one that I specifically wonder about:  I am compounding $384.30 + $74.2 = $458.50 per month for IRMAA Part B and Part D at 7.7% annually for 40 years which gives me a result of $8,911.84 per month.  I'd appreciate anyone checking my math.  Again, my understanding is that the IRMAA brackets are adjusted by CPI or similar, but IRMAA premiums are adjusted by Medicare costs.

So first know (or hear again) that I have a ton of respect for your tax skills and financial analytical skills. You've displayed those a ton over the years. Or longer. :-)

But to address the issue of adjusting the IRMAA bump for inflation--and I don't have your spreadsheet's formulas--but it seems like you'd want to check your inflation adjustments on the tax brackets too. I.e., at some point, if one assumes that medical costs just continue to increase at a rate two or three times the general inflation rate, it seems like they become such a big component of everybody's spending that the overall inflation rate should be adjusted by a way higher number too. I.e., that medicare inflation and general price level information should converge. Practically, not sure how you address that in a spreadsheet. But it seems like you need to model either a lower IRMAA "inflation rate" or a higher tax-bracket "inflation rate."

FWIW I do think (and this is another tricky issue) that it makes sense to assume the current tax laws apply. Mostly because it's so hard to predict what'll happen. And that means assuming the current rates expire in 2026 and return to the pre-2018 rates... but with inflation adjustments so all of those amounts move steadily higher.

I won't keep drawing out my commenting in this thread, but I think I think Roth-style accounts work more poorly than many here do because:
1. I'm assuming low-ish real rates of return going forward. E.g., like Vanguard's forecasts.
2. I think one has other good ways to avoid RMD troubles. (Tax efficient taxable accounts work really well for some people.)
3. I worry more about other things (sequence of returns risk) than I do about RMDs or paying IRMAAs.

Thanks first to the OP for asking some related question so SeattleCPA and I can jabber on about perhaps tangential subjects! :)

Thanks for the kind words, but I still have little nightmares that maybe I did my math wrong.  It's a homegrown spreadsheet which has only been reviewed by my eyes, and I know for a fact that I've made spreadsheet errors before - some rather large.

I take your point - inflating IRMAA premiums at 7.7% and tax brackets at ~CPI-ish 3% does imply a future that probably doesn't exist for some reason or another.  Who knows what that will be?  I sort of shrug and figure I'll adapt as things change.  (Personally I suspect some sort of Medicare rework.  Maybe.)

All I've done with TCJA expiration is be aware it exists and consider this year and next year a chance to maybe do some bigger conversions at lower rates because probably maybe the rates will go up 3%-ish and the brackets will shrink.  If they expire I'll rework my spreadsheet and go from there.  I don't look back so I won't evaluate woulda coulda on my conversions from 2016 through 2024...I'll just be glad (probably) that I at least did some.

1.  I do project higher-ish real rates than most people - 10% investment growth and 3% inflation.  My measured rate of return over the past decade is 12.9% and my measured personal inflation rate is approximately 0.5% over the past decade or so.)  That's obviously a whole other debate/discussion, but if you accept the premise I'm sure you can see why Roth conversions start to make more sense.

2.  I have tax efficient taxable accounts that already work really well for me.  QCDs might be in my future and are a good way to whack down RMD trouble for people who are already charitably inclined.

3.  At a net WR% bouncing around but solidly under 1% and age 55, I don't worry about SORR.  I also don't necessarily worry about RMDs or IRMAA; I just figure I might as well make some obvious (?) optimizations since I have the time, interest, and probably the ability to do so.

[SeattleCPA]

I do project higher-ish real rates than most people - 10% investment growth and 3% inflation.  My measured rate of return over the past decade is 12.9% and my measured personal inflation rate is approximately 0.5% over the past decade or so.)  That's obviously a whole other debate/discussion, but if you accept the premise I'm sure you can see why Roth conversions start to make more sense.

So for contrast, I'm using a 3.5% real rate of return. Half of yours obviously. And totally agree, that one change to the inputs in our spreadsheets makes a huge difference. This isn't really the spreadsheet I use, but I did a blog post a few weeks ago about how to use the expected low rates going forward and a Monte Carlo simulation to think about impact of lower rates: https://evergreensmallbusiness.com/monte-carlo-safe-withdrawal-rates-for-low-expected-returns/ And lower rates paint a very different picture.

I have tax efficient taxable accounts that already work really well for me.  QCDs might be in my future and are a good way to whack down RMD trouble for people who are already charitably inclined.

To go off on a tangent with this, what I find is taxable, tax-efficient investments can make a really big difference in how the RMD issue plays out. You just have to live long enough. But I think you need maybe 20% of your nest egg in taxable.

Example using numbers that make math easy even if they're unrealistic...

Someone has $3,000,000 at retirement with $1,000,000 in taxable invested in total stock market portfolio and $2,000,000 in IRA invested in say Vanguard's Life Strategy Fund with 60% in bonds.

Overall asset allocation is 60/40. Assume investor draws 4% or $120K a year to start.

But say $18K of that comes from the $1,000,000 taxable and represents qualified dividends. Very lightly taxed at a federal level obviously. All of the other return on the taxable is unrealized so untaxed.

The other $102K gets drawn from the $2,000,000. That is taxed. But the WR is slightly over 5%. Also with a tax-deferred account that's 60% bonds, that 5%+ WR will at least in my modeling effectively dodge the RMD issue. Both for taxpayer and heirs.

BTW if one combined qualified charitable distributions with the above tactic, some of that $102K distribution isn't even taxed obviously.

I also don't necessarily worry about RMDs or IRMAA; I just figure I might as well make some obvious (?) optimizations since I have the time, interest, and probably the ability to do so.

That makes sense. I often tell people worrying about paying high taxes that in general high taxes are not really a big problem. It signals a high income! The big problem obviously is finding one's income too low to comfortably get through the day or the month.

At a net WR% bouncing around but solidly under 1% and age 55, I don't worry about SORR.

I wonder if we haven't both in effect addressed the possibility of low expected returns. But in different ways. But to comment on above comment, yeah, I wouldn't worry about SORR if the WR is less than the real return on TIPS.

[Omy]

I very much appreciate the RMD discussion/insights in this thread.

We fat-FIREd 4 years ago and it's clear from our spreadsheets that we will have RMD/IRMAA issues 18 years from now. Because we don't have kids and aren't terribly concerned about leaving a legacy, we aren't getting ourselves too worked up about it.

We've been advised that we really need to get going on Roth conversions, but right now we're paying very little tax and getting $1000/mo in ACA subsidies. If we started seriously converting to Roths, our current tax bill would increase substantially and our ACA subsidies would disappear. Since we're relatively early in retirement, it feels good to minimize taxes right now.

There are so many things to optimize that it can feel overwhelming. For now our main focus is maintaining our ACA subsidies. QCDs might help keep us out of IRMAA range two decades from now. If one of us dies, the tax situation gets significantly worse...but I've always joked that having a large tax bill is one of the best "problems" you can have.

[Ron Scott]

Roth IRAs and backdoor Roth IRAs have to be on the top ten list of overrated tax planning ideas.

Amen

[EvenSteven]

I've done the math for myself, using my home grown spreadsheet.  I admit it might have errors, but I've reviewed it several times over the years and believe it to be accurate.

My spreadsheet goes to age 95 and is in nominal dollars.  It accounts for the following:

1.  Investment growth on my trad IRA.
2.  RMDs on my trad IRA using RMD factors according to table 3.
3.  A likely inherited trad IRA and the 10 year drawdown rule.
4.  Roth conversions based on the different tiers (top of tax brackets and IRMAA thresholds).
5.  Social Security at age 70 with the possible haircut and the 85% taxation rule (my spreadsheet assumes 85% on all, which is on the high side, I know).
6.  The standard deduction, adjusted for inflation.
7.  Ordinary income tax brackets, IRMAA brackets, and IRMAA surcharges, all adjusted for either inflation or the historical growth rate.

This should get me in the neighborhood.  Mainly what I do each year is update the various constants in the spreadsheet once a year and then use that to figure out what my federal marginal plus IRMAA rate is at age 85, which for me is about my 50/50 life expectancy based on the Social Security Longevity Tool.

What this tells me is that I should be converting up into the mid-20%s now to avoid 32% and Tier 4 IRMAA later.  I usually don't go that far for a number of reasons.

The biggest uncertainties in my analysis are the growth rate of IRMAA premiums.  These are adjusted, as I understand it, based on the actual costs of the Medicare program so that it is funded at the appropriate levels.  Right now I use a factor of 7.7%, which seems plausible, but compounding that over time has my IRMAA surcharges alone (not base B+D) at over $100K at age 95, making it the equivalent of about a 10% additional marginal tax rate on top of the 32%.

And that's with doing Roth conversions every year between now and then.

Some comments about the above.

1. I think the assumption that 85% of your Social Security will be taxable is a good one for anyone who's done or plans to do a good job saving. Those $25K and $32K amounts (from 1984!?) don't get adjusted for inflation.

2. Agree that inherited IRA accounts create a planning challenge.

3. I can't imagine the top tier IRMAA adjustment going from roughly $400 a month to $8000 a month... or at least not unless there's just massive inflation that scrambles all of these numbers.

4. If you're married and looking at keeping out of the current 32% federal bracket (so more than $400K of AGI if married and more than $200K of AGI if single), that puts you into a pretty rare group. You know that. But for others reading this, understand there maybe a pretty big difference between @secondcor521 's situation and yours or mine.

5. A more general comment: Other ways exist to deal with high marginal rates in retirement other than Roth-style accounts: Changing state residency, home ownership, active real estate investments, strategically allocating taxable investments so a big chunk of your retirement income (the last bits) are qualified dividends or long-term capital gains, family income splitting, etc.

Also just to say this, I don't have a problem with Roth-style accounts per se. I have problem with people applying a rule of thumb that says, "Hey I don't want to pay taxes in retirement... so my financial planner has me doing Roth conversions." Clearly that's not the case with @Sandi_k or @secondcor521 and maybe two or three million other folks . But I see clients all the time who will retire with best case a $2M portfolio and, say, good Social Security benefits and are trying to load up their Roth-style accounts. And the problem is when they're making good six figure salaries as a household, the taxes they want to save or avoid are the ones they're paying during their higher income working years.

Note: My calculation of what you need to end up with a $2M-ish portfolio in real terms. Max 401(k) for three decades while making $200K a year and getting a safe harbor 4% match. Assuming a five percent real return. The Excel formula =FV(0.05,30,23000+.04*200000) returns roughly $2M.

With this 30-year earnings history, you'll get a $46K SS benefit if single, $69K if married with spousal benefits. If you draw $80K from your $2M IRA, your AGI is roughly $150K. That's $50K less than your earnings.

The bolded has been the biggest monkey wrench in my planning. From one set of parents, anywhere from $0 to $2M, all in traditional IRA, inherited anywhere from 1 year to 25 years from now. Same with the other set of parents. Try planning around that.

[EscapeVelocity2020]

I know this'll be a controversial take, but I got burned back when the Roth first became a thing in 1998 and I converted all of my traditionals to Roth (back when my income was low).  Paid higher taxes at the time, but felt awfully smart about all the tax free growth.... until the market collapsed in 2000 and many of my conversions were then in the red.  So yeah, I'm going to err on the side of caution when it comes to voluntarily paying more tax in the present.  Paying high taxes is generally a sign that you have done well, annoying as it can be, and there are options and strategies to offset overly burdensome taxes if your FI is solid.  It was a great idea when I was in my 20's and those Roths are way up now, probably not as clear when I'm in my 60's.  There are also assumptions around estate planning that are awfully hard to figure out... 

[freeree]

"What I do think is for the great majority of people, Roth IRAs as an alternative to traditional IRAs and Roth 401(k)s as an alternative to traditional 401(k)s don't make sense."

I understand that a Roth conversion probably doesn't make financial sense for many people (present/future tax brackets, size of stash etc) BUT what I'm wondering is --- does anyone convert to avoid RMDs? Senility might be in my future, and the thought of calculating RMDs (however small) from the age of 72 (or 73) onwards is yet one more thing to worry about. [Mind you, if I do become senile what will I care???]

[secondcor521]

"What I do think is for the great majority of people, Roth IRAs as an alternative to traditional IRAs and Roth 401(k)s as an alternative to traditional 401(k)s don't make sense."

I understand that a Roth conversion probably doesn't make financial sense for many people (present/future tax brackets, size of stash etc) BUT what I'm wondering is --- does anyone convert to avoid RMDs? Senility might be in my future, and the thought of calculating RMDs (however small) from the age of 72 (or 73) onwards is yet one more thing to worry about. [Mind you, if I do become senile what will I care???]

Most custodians are able to set up automatic RMDs for you.  They'll do the math, the tax withholding, and the distributions on whatever schedule you select.  My Dad has been doing this with Vanguard for years and it works like a charm.

Of course, the RMD owner (you) are still responsible for ensuring that the RMD is distributed each year.  So far, though, Vanguard has gotten it correct every year, even when the IRS changed the divisor table a few years ago.

[Catbert]

"What I do think is for the great majority of people, Roth IRAs as an alternative to traditional IRAs and Roth 401(k)s as an alternative to traditional 401(k)s don't make sense."

I understand that a Roth conversion probably doesn't make financial sense for many people (present/future tax brackets, size of stash etc) BUT what I'm wondering is --- does anyone convert to avoid RMDs? Senility might be in my future, and the thought of calculating RMDs (however small) from the age of 72 (or 73) onwards is yet one more thing to worry about. [Mind you, if I do become senile what will I care???]

As secondcor said your custodian will happily figure your RMDs and send them to you on whatever schedule you want.  Monthly?  Yearly in a given month? Whatever.

I was apparently an over saver and consequently have been madly making Roth conversions for the past 12 years or so.  I'm still facing a tsunami of RMDs starting next year.  That's why I have made Roth conversion.