Roth conversion study surprised result

[Exchme]

Something that is generally not analyzed about Roth conversions is that the models are very unrealistic in that you have to use the same stock/bond allocation in all your accounts or the calculation mixes the higher returns of stocks with the benefits of Roth conversions and you learn nothing useful. However, no one is going to hold assets that way if they can help it, they want bonds in tax deferred to limit its growth and stocks in taxable and Roth.

The simple act of optimizing asset location looks like it drastically cuts the optimum amount of Roth conversions. In my own case, it cuts the optimum Roth conversion in half, from 70% of the initial tax deferred account if I used all accounts at equal asset allocations, to 35% when studied in more detail.

To walk through what happens, eventually, if you do enough Roth conversions and enough time passes, RMDs eat away at tax deferred and your tax deferred fills with bonds as your stocks grow and you rebalance to maintain your asset allocation. The overflow bonds have to go somewhere and for me Roth was less bad than taxable (yes I tested). But roughly speaking, once that overflow got serious, I wasn't getting additional much benefit from doing more conversions. This filling tax deferred with bonds happens in the No Roth case too, just slower.

The only consumer level tool I'm aware of that can tackle asset location optimization is Pralana Gold. It cost me ($99 1st year, I believe will be $49 renewal, needs Excel) and is very full featured (it's complex, but you don't write formulas and the layout, menu and manuals are good, but you really should be familiar with spreadsheets before tackling this). It reports the overall asset allocation and allows you to adjust the allocation of each type of account (taxable, tax deferred, Roth) up to four times during your lifetime. So I start with the overall stock allocation a couple percent below my target and over say six years, it grows to 2-3% above target, so I rebalance to a couple percent below target and repeat. This is manual and requires lots of fiddling between screens to get the amount above and below target to balance out each time to make the answer reasonably good.

Pralana Gold then gives both the NW and a rough estimate of the heir's taxes. If you are going to give your estate to charity, then NW at death is the right measure as the charity doesn't pay taxes on any bequests. I get only a very slight improvement in NW at death converting up to the 10% bracket and net losses after that.

If you are leaving your estate to heirs, then the heir's taxes are also part of the picture. I am using Roth conversions to help my heirs, so I include their taxes. Pralana Gold's estimate of heir's taxes is low in my case since conversion taxes came out of taxable and that leaves a smaller taxable account generating slightly lower taxes from dividends during the 10 year liquidation period for the heirs.

I'm sure your reaction is the incremental taxes during heir liquidation have to be very small - yes indeed, but the benefit for the Roth is pretty small too, so it is a noticeable part. (I had to do those estimates outside Pralana Gold). The results below include a present value calculation to bring the benefits back to current $, not after 40 years of compounding.

My final results show Roths net interesting, but not life changing money. (In my ready, fire, aim approach I've already converted past the optimum for this year)

The charts below show the benefit for the Roth with optimized asset location, with the values after heir liquidation corrected to present values. Each point on the graph was the max I could convert and stay within a given IRMAA tier or tax bracket.

The orange line is the benefit as a fraction of the conversion. For example, the third orange point (optimum) is about a 16% benefit on a conversion of 35.5% of the account = 5.7% of the tax deferred account balance can be gained with optimum Roth strategy (if the world unfolds exactly as assumed). The blue line is the incremental benefit between points, which goes negative when converting at higher tax brackets or IRMAA tiers.

I also ran a case positing the expiration of the TCJA, figuring that would show the much higher conversions I expected. Nope. It didn't change the optimum conversion amount.

A few more details - MFJ, 61 & 60, assuming I die at 80 and DW lives to 92, no pension, SS at 70 for me and spousal at 67 for DW, assumed 5% real on stocks, 1% real on bonds, 80/20 AA. Mid seven figures, 50/50 taxable/tax deferred. Seems like Roths are more of an "excess return insurance" or "I die early" insurance.

Like everything around Roth conversions, the results are very individual, so what is true for me may be ridiculous for others.

 

[Out-to-Lunch]

I will need to digest this, but thanks for sharing. This is valuable, coming from you!

 

[SevenUp]

My final results show Roths net interesting, but not life changing money.

That seems reasonable, and it's just a difference between marginal tax rates and not some 2X or other multiple.

Rather than "percent converted", do you have results by "marginal tax rate of conversion"? And compared with the marginal tax rates you are projecting when in your 80s?

 

[Scratchy]

Thanks for sharing this interesting analysis.

Given your assets today, to the extent Roth conversions are done, wouldn't there be reduction in estate tax due upon the death of the second spouse? And that's in the best case scenario where the estate tax exemption after portability goes down to around $11 million in 2026. Have you considered factoring this in?

 

[pb4uski]

As we have often discussed, if you're converting today at 22% to avoid 24% later then there isn't much punch to Roth conversions, but if you're converting and paying 12% today to avoid 22% later then it is a whole different story.

Over the last 9 years I've converted about $470k and paid $42k in federal tax (9%... a mix of 0%, 10% and 12%).... when I am RMD age and SS is online we'll definitely be in the 22% tax bracket.... so there is little doubt that I've saved $61k... more than "interesting" to me... though I'll concede that it isn't life changing.

 

[Exchme]

As we have often discussed, if you're converting today at 22% to avoid 24% later then there isn't much punch to Roth conversions, but if you're converting and paying 12% today to avoid 22% later then it is a whole different story.

Over the last 9 years I've converted about $470k and paid $42k in federal tax (9%... a mix of 0%, 10% and 12%).... when I am RMD age and SS is online we'll definitely be in the 22% tax bracket.... so there is little doubt that I've saved $61k... more than "interesting" to me... though I'll concede that it isn't life changing.

I guess I should have included it, the optimum conversion point was the 22% prior to IRMAA and 176K once IRMAA was in effect. There were a lot of years in the 24% bracket in the future, but as you say, it's not a big difference and the amount of money in the higher marginal bracket each year wasn't high.

 

[Markola]

Here’s another take from an early retiree who decided that Roth Conversions don’t result in enough benefit for his circumstance to be worth the effort.

https://www.caniretireyet.com/roth-iras-roth-conversions-needs/

 

[Exchme]

Thanks for sharing this interesting analysis.

Given your assets today, to the extent Roth conversions are done, wouldn't there be reduction in estate tax due upon the death of the second spouse? And that's in the best case scenario where the estate tax exemption after portability goes down to around $11 million in 2026. Have you considered factoring this in?

Ooh, that's been off my radar, but compounding is magical and the estate tax does seem likely not to be extended.

At a glance, it seems like it will make the differences between cases a little smaller. The case I found as optimum closes with $60K higher NW ($16K PV) than No Roth, so will look a little worse after estate taxes but the 60% conversion case ends with $130K less than No Roth ($36K PV) so will look a little bit better. Since the tax is 40% of those already small numbers, I don't think it changes the conclusion.

If the market continues its rocket to the moon phase, we will have to start looking at estate planning. We are slowly increasing gifts to the kids, but my wife doesn't want to do big gifts, she:
a) doesn't want them to slow their hard work and count on money from us.
b) saw my dad's money disappear due to a decade in the Alzheimer's unit for him while still paying the bills for my step mom and enduring the market crashes of 2000 and 2008.
or
c) really likes having more money

 

[Montecfo]

Exchme,

Thanks for that analysis. It does make sense that portfolio composition in each account is a part of the analysis.

And Markola, thanks for the link to the Kirkpatrick Darrow piece. I have made the similar point that knowing predicting market returns and heirs' tax rates far in the distant future means benefits of Roth conversions at tax rates similar to your current ones are highly uncertain.

For married folks, considering spouse mortality can make a difference given single tax rates but even there you are making assumptions about of course spouse (or your own) mortality and post actions such as remarriage, possibly many years or even decades down the line.

I continue to see the benefit of Roth conversions where there is a meaningful difference in tax rates e.g. not just 2 percent.

 

[ERD50]

A less obvious, and difficult to model, benefit of a Roth conversion is that you now have a larger source of funds that can be pulled w/o tax implications for some large expense.

If you suddenly decide to (or need to) "Blow That Dough!", pull it from the Roth with zero tax implications.

-ERD50

 

[RunningBum]

Here’s another take from an early retiree who decided that Roth Conversions don’t result in enough benefit for his circumstance to be worth the effort.

https://www.caniretireyet.com/roth-iras-roth-conversions-needs/

It's funny how someone who says he doesn't like complexity and paperwork does all that to write that blog entry anyway. I don't find it much effort at all. I'm sure some do, but if you don't want to go to the effort to figure out whether there's an advantage or not, I don't think you should be telling others it's not worthwhile. That's like telling someone to not eat at a restaurant that you've never even been to.

For the typical retiree who starts SS and tapping their retirement accounts right away, a conversion probably has little advantage, if any. It's also rare that someone still working gets a conversion advantage. There is probably no tax rate advantage. Many of the people here aren't typical though. The most likely time when conversions work well are your early retirement years until you start collecting SS and have to take RMDs.

I haven't really dug in to the OP's premise. Something about having to move bonds to your Roth once you've converted enough out of your tIRA that you need to put some bonds in your Roth, and that will cause drag. I've never done a conversion analysis that assumes a better ROI in the Roth post-conversion than what I was getting in the tIRA pre-conversion. I'm only looking at the tax rate arbitrage, plus the small extra benefit gained by paying the conversion taxes out of my taxable account. And yes, I do factor in the loss of return in funds I used in the taxable account. It still comes out better. A pretty simple spreadsheet proves this.

 

[Exchme]

That seems reasonable, and it's just a difference between marginal tax rates and not some 2X or other multiple.

Rather than "percent converted", do you have results by "marginal tax rate of conversion"? And compared with the marginal tax rates you are projecting when in your 80s?

The optimum was 22% before IRMAA and the 176K IRMAA tier.

In both the best case with no asset location optimization and the optimum found here, the marginal bracket was 22% while I was alive and 24% after I'm gone (assumed age 80). But location optimization cut the amount in the 24% bracket by ~80%, the remaining bit wasn't enough to make conversions into the 24% bracket now attractive.

I was absolutely convinced the right answer was top of the 24% bracket for several years, but that turns out to be value destruction in this case.

I get that the programming would be really hard to allow this analysis, but it's awful frustrating that it is so hard (and Pralana Gold is the only thing I've found where it's even marginally possible). Even here, I'm having to do approximations on the heirs' situation instead of the year by year simulation it deserves.

 

[Montecfo]

The optimum was 22% before IRMAA and the 176K IRMAA tier.

In both the best case with no asset location optimization and the optimum found here, the marginal bracket was 22% while I was alive and 24% after I'm gone (assumed age 80). But location optimization cut the amount in the 24% bracket by ~80%, the remaining bit wasn't enough to make conversions into the 24% bracket now attractive.

I was absolutely convinced the right answer was top of the 24% bracket for several years, but that turns out to be value destruction in this case.

I get that the programming would be really hard to allow this analysis, but it's awful frustrating that it is so hard (and Pralana Gold is the only thing I've found where it's even marginally possible). Even here, I'm having to do approximations on the heirs' situation instead of the year by year simulation it deserves.

I would think the benefit is least for those with high bond allocations and smaller tax deferred accounts.

And accordingly greatest for those with higher stock allocations and larger tax deferred balances.

Of course, statistically we can expect lower stock returns in the near term from current elevated levels. So maybe bonds outperform for a while. That could affect results as well.

The many variables.

 

[ERD50]

... I haven't really dug in to the OP's premise. Something about having to move bonds to your Roth once you've converted enough out of your tIRA that you need to put some bonds in your Roth, and that will cause drag. ...

That's what I think OP is saying. Wouldn't the opposite also apply?

For many of us, even after conversions (especially if only going to the top of the 12% bracket), we will still have a lot in a tIRA (and/or t401/t403). So we can still be flexible with our AA allocation between tIRA and Roth.

So if I put my stocks in Roth, I don't pay any tax on the growth. If those stocks were in a tIRA, the growth portion that is withdrawn is taxed at regular income tax rates (to be clearer, it is *all* taxed as income, I'm just focusing on the growth for this comparison). So doesn't that mean it is advantageous to keep stocks (expected higher growth) in Roth, and bonds (expected lower growth) in Traditional IRA, to the extent possible?

-ERD50

 

[Out-to-Lunch]

Something that is generally not analyzed about Roth conversions is that the models are very unrealistic in that you have to use the same stock/bond allocation in all your accounts or the calculation mixes the higher returns of stocks with the benefits of Roth conversions and you learn nothing useful. However, no one is going to hold assets that way if they can help it, they want bonds in tax deferred to limit its growth and stocks in taxable and Roth.

I want to bring up for your consideration that "location optimization" may be too broad a brush when considering two tax-advantaged structures. (So I am not including taxable accounts here, just tIRA vs. Roth.)

Yes, many of us hold bonds in tax-deferred, and stocks in Roth. But the effect of this is to effectively increase your stock allocation, i.e., to take on more risk than your nominal AA would suggest. (See https://www.bogleheads.org/wiki/Tax-adjusted_asset_allocation .)

Also see: https://www.bogleheads.org/wiki/Tax-efficient_fund_placement

In considering asset location keep the following points in mind:

If your investments are all in tax-advantaged accounts, fund placement will not have a large impact on your returns. Tax-advantaged accounts include tax-deferred accounts, such as traditional 401(k), 403(b), and Traditional IRA, and the tax-free Roth versions of those accounts such as Roth IRA.

If you think about it, the same math (the commutative law of multiplication) that suggests that only the marginal tax rate on conversions vs. ordinary distributions matters, and the timing does not, also suggests that you should be indifferent to whether growth takes place inside of a tIRA. (EDIT: Unless the growth drives you into higher tax brackets upon distribution.)

So, I would submit that when you are optimizing your location, the main effect is that you are increasing your effective stock AA, and, therefore, your expected returns. But also increasing your risk.

(Note: I have learned in the past that tax-adjusted AA is not a well-accepted concept on this board, the way it is on BH. That is why I am merely suggesting that you consider this concomitant effect of asset location.)

 

[Exchme]

It's funny how someone who says he doesn't like complexity and paperwork does all that to write that blog entry anyway. I don't find it much effort at all. I'm sure some do, but if you don't want to go to the effort to figure out whether there's an advantage or not, I don't think you should be telling others it's not worthwhile. That's like telling someone to not eat at a restaurant that you've never even been to.

For the typical retiree who starts SS and tapping their retirement accounts right away, a conversion probably has little advantage, if any. It's also rare that someone still working gets a conversion advantage. There is probably no tax rate advantage. Many of the people here aren't typical though. The most likely time when conversions work well are your early retirement years until you start collecting SS and have to take RMDs.

I haven't really dug in to the OP's premise. Something about having to move bonds to your Roth once you've converted enough out of your tIRA that you need to put some bonds in your Roth, and that will cause drag. I've never done a conversion analysis that assumes a better ROI in the Roth post-conversion than what I was getting in the tIRA pre-conversion. I'm only looking at the tax rate arbitrage, plus the small extra benefit gained by paying the conversion taxes out of my taxable account. And yes, I do factor in the loss of return in funds I used in the taxable account. It still comes out better. A pretty simple spreadsheet proves this.

Well I obviously wasn't clear, I think it's a simple premise - you would not hold the same allocation in all accounts, you would put bonds in tax deferred. That one act completely changes the optimum Roth conversion strategy. That slows the growth in tax deferred so much that projections about how much to convert go out the window. So doing the analysis the normal way over-estimated the optimum conversions in my case by a factor of 2.

I was surprised to see that tax deferred got full of bonds in this analysis and was speculating that that may be a secondary factor in explaining things. Considering I hold a low percentage bonds, if that is a factor, the situation might be worse for others. But it's way too much work to do test cases to see if this is a bend point that rapidly makes things worse or just an expected result. So you can ignore that paragraph and still get the idea.

 

[pb4uski]

Here’s another take from an early retiree who decided that Roth Conversions don’t result in enough benefit for his circumstance to be worth the effort.

https://www.caniretireyet.com/roth-iras-roth-conversions-needs/

WADR to the author who I have read before and like, he is a bit all over the place on this article. First, early in the article he correctly observes that Roth conversions are principally a tax rate arbitrage play.

So, there is no difference in the ending values of the two accounts, assuming your tax rate is unchanged between the initial contribution, and your withdrawal. (Thanks to Mike Piper at Oblivious Investor for making this commutative property of multiplication crystal clear.)

If your tax rate changes, though, the story is different: If your tax rate goes down, a Traditional IRA does better. And if your tax rate goes up, then the Roth does better.

Then he goes on to say that his tax rate is dramatically lower in retirement so Roth conversions are not particularly beneficial.

A tax rate lower in retirement than when we were working is true for us too, even once RMDs begin.

But what we are talking about when we talk Roth conversions is a "sweet spot" between early retirement and RMDs. During first 12 years of this 14 year period for us, before I start my SS, our marginal tax rate with no Roth conversons is 0% as our income is less than the standard deduction. For the last 2 years, after I am collecting SS but before RMDs begin (I'm 70-72), we are just a little into the 12% tax bracket. After that, once RMDs begin, we are in the 22% tax bracket for life.

He goes on to say that his modeling with the Prelana calculator showed a 5-10% increase in net worth for doing Roth conversions... that is very consistent with my Excel model for Roth conversions... my model suggests that Roth conversions increase our net worth by 7% measured without any deferred income tax liability and 9% if I include a deferred income tax liability. ... and these calculations are using conservative investment return assumptions. He then goes on to say:

Given the variation in results from the commercial calculators, I created my own spreadsheets to further analyze and understand the factors related to RMDs and the Roth conversion decision. My spreadsheets show us maintaining our current 15% tax bracket, with headroom to spare, and project only a 2-3% benefit to net worth from doing Roth conversions.

It seems to me that even a 2-3% improvement in net worth makes it worth the risk of being wrong and the negligible effort of doing Roth conversions.... so it is definitely worthwhile for us but I can see that for others... especially those at the bottom of the 22% tax bracket without Roth conversions... that it might not be worth the effort.

 

[target2019]

Well I obviously wasn't clear, I think it's a simple premise - you would not hold the same allocation in all accounts, you would put bonds in tax deferred. That one act completely changes the optimum Roth conversion strategy. That slows the growth in tax deferred so much that projections about how much to convert go out the window. So doing the analysis the normal way over-estimated the optimum conversions in my case by a factor of 2.

I was surprised to see that tax deferred got full of bonds in this analysis and was speculating that that may be a secondary factor in explaining things. Considering I hold a low percentage bonds, if that is a factor, the situation might be worse for others. But it's way too much work to do test cases to see if this is a bend point that rapidly makes things worse or just an expected result. So you can ignore that paragraph and still get the idea.

Sure, your circumstances and placement in an account are unique. And it's all speculation until you do conversions and have measurable results.

If you have all bonds in tax-deferred and nothing else, it is very different than our situation (we have 6-7 accounts by necessity, for a while longer). For us it was very simple - convert VTSAX total market to VTSAX in the Roth.

Our 50% fixed (bonds, etc.) will probably float down to something less, and come out as RMDs at some point. So our future decision is whether to maintain the same AA, year-by-year.

I have spreadsheets and read the mega-threads here about Roth conversion. Eventually I let go of the idea of maximizing every investment dollar into a future I'll never experience. I put an annual Roth conversion into Flexible Retirement Planner and see some benefit way out there in time. Others have a very different take on things. It's all good!

 

[Exchme]

I want to bring up for your consideration that "location optimization" may be too broad a brush when considering two tax-advantaged structures. (So I am not including taxable accounts here, just tIRA vs. Roth.)

Yes, many of us hold bonds in tax-deferred, and stocks in Roth. But the effect of this is to effectively increase your stock allocation, i.e., to take on more risk than your nominal AA would suggest. (See https://www.bogleheads.org/wiki/Tax-adjusted_asset_allocation .)

Also see: https://www.bogleheads.org/wiki/Tax-efficient_fund_placement



If you think about it, the same math (the commutative law of multiplication) that suggests that only the marginal tax rate on conversions vs. ordinary distributions matters, and the timing does not, also suggests that you should be indifferent to whether growth takes place inside of a tIRA. (EDIT: Unless the growth drives you into higher tax brackets upon distribution.)

So, I would submit that when you are optimizing your location, the main effect is that you are increasing your effective stock AA, and, therefore, your expected returns. But also increasing your risk.

(Note: I have learned in the past that tax-adjusted AA is not a well-accepted concept on this board, the way it is on BH. That is why I am merely suggesting that you consider this concomitant effect of asset location.)

I was worried about this, but hadn't checked until your post.

Since most people won't read the links, I'll summarize. When you look at the overall asset allocation and hold different things in different places, you should ideally correct for the fact that the government owns a share of whatever is in tax deferred. In the schema I looked at, tax deferred gets filled with bonds, but the government owns a share of them, so I don't own as many bonds as I might think. Since I used the overall allocation as if all accounts were equal when they're not, there can be a bias.

This is true, but between the optimum case and the next point with increasing conversions, the difference strikes me as too small.

The bonds belonging to the government hiding in tax deferred can be found by:
% of total savings that is tax deferred
x % bonds in tax deferred
x %marginal tax bracket

When comparing my optimum with the next higher conversion, I find the asset allocation bias was actually in favor of the larger Roth conversion case up to about age 80 and then trended towards biasing in favor of less Roth conversion, but was always within tenths of a percent of the correct target asset allocation.

So this bias is real but maybe 0.25%-0.30% average increase in asset allocation in the case with less Roth Conversions. With the rate of returns I'm using that is 0.01% per year in total value. That seems too small to affect the results.

As for asset allocation bias in the taxable account holdings, I'm holding all stocks in both taxable and Roth until near the end when some bonds need to go in Roth. Since the tax drag in taxable is maybe 0.2-0.3%/year, the change in returns due to a hidden bias in asset allocation caused by such a small drag that only matters late in the analysis when bonds appear in Roth seems safe to ignore and I have not tried to calculate it.

 

[GTFan]

One important thing to consider about conversions is how it impacts your ACA subsidies, assuming you are using this insurance. Makes the calc even more complicated.

 

[DawgMan]

I have to admit, as I start to play with some different Roth conversion strategies, my head spins. You guys are all much smarter than me. A tool I have used to help with this is New Retirement Planner Plus. I would be curious to get everyone's feedback on the results I have gotten so far...

Macro Modeling Assumptions
- Start withdrawals 2022 (no earned income), age 57 (both me and DW)
- Using current tax brackets (married, file jointly), which stay "as is" growing 2% annually
- Higher than typical annual planned withdrawal amount (FatFire- very discretionary spend), growing 2% annually
- 2% annual inflation
- 60/40 AA averaged across all retirement accounts, ideally positioned to minimize taxes, underwriting 5% annual returns
- Current retirement accounts: 55% after tax/45% Tax deferred, 0% Roth
- Planning on 75% of current SS benefit. Current WR in low 2% range before SS.

Test 1: No Roth Conversions, live off after tax $$ until RMDs (no tax deferred acct withdrawals), upon age 72 (RMDs) results in hitting the 32% tax bracket until age 78, then goes to 35%+.

Test 2: Do larger Roth Conversions starting in 2022 (age 57) into the 24% tax bracket through age 71 and then stop (no withdrawals other than conversions from tax deferred accts), then RMDs kick in resulting in me staying in the 24% tax bracket until one of us passes. Lifetime tax savings save me over 31% from Test 1, which relates to real $$ in my case.

So, aside from legacy concerns (ideally transferring Roth vs tax deferred accounts to kids) and future single spouse tax issues, should that not be a strong enough argument alone to do Roth conversions into a tax bracket that in theory is the same as it would be in RMDs? The only counter argument I would make to my 31% tax savings is that it is in real $$, and probably should be somewhat discounted since I can effectively pay no taxes until RMDs hit (a dollar today is worth more than a dollar tomorrow)? What discount rate should I use and what does it do to the 31% (is it really 20%?), I'm not sure?

So, based on my analysis, it seems Roth conversions are worth it, even into the 24% tax bracket. Throw in the legacy and single spouse benefits, even more beneficial. Assume high probability of even higher taxes in the future, they become even more more beneficial.

What say you?

 

[pb4uski]

I think it is worthwhile in your case because with no conversions you would be in the 32% or 35% tax bracket vs paying 24% today.... that 8-11% tax savings adds up. And 31% tax savgs seems credible given the disparity in rates that you quoted.

Similarly, in my case trading off 12% for 22% is also worthwhile.

But if your're in the 22% for 24% range then not so much.

 

[Montecfo]

The legacy benefits are only there if there are tax savings. Or are you identifying other legacy benefits?

 

[DawgMan]

The legacy benefits are only there if there are tax savings. Or are you identifying other legacy benefits?

Legacy, as in the kids will not have taxable RMDs with inherited Roth accounts.

Frankly, I look at anything my kids get (which will be plenty) as of little concern as they should look at it as found money. My main focus is the benefits it provides for my DW & me.

 

[RunningBum]

Test 2: Do larger Roth Conversions starting in 2022 (age 57) into the 24% tax bracket through age 71 and then stop (no withdrawals other than conversions from tax deferred accts), then RMDs kick in resulting in me staying in the 24% tax bracket until one of us passes. Lifetime tax savings save me over 31% from Test 1, which relates to real $$ in my case.

So, aside from legacy concerns (ideally transferring Roth vs tax deferred accounts to kids) and future single spouse tax issues, should that not be a strong enough argument alone to do Roth conversions into a tax bracket that in theory is the same as it would be in RMDs? The only counter argument I would make to my 31% tax savings is that it is in real $$, and probably should be somewhat discounted since I can effectively pay no taxes until RMDs hit (a dollar today is worth more than a dollar tomorrow)? What discount rate should I use and what does it do to the 31% (is it really 20%?), I'm not sure?

So, based on my analysis, it seems Roth conversions are worth it, even into the 24% tax bracket. Throw in the legacy and single spouse benefits, even more beneficial. Assume high probability of even higher taxes in the future, they become even more more beneficial.

Yes, in your situation I would be probably convert to the top of the 24% bracket.

There is the see-saw effect to watch out for. As you convert and pay some taxes now, does it eliminate or reduce your tIRA enough that you won't be in that higher tax bracket later? You don't want to convert at 24% only to find out that you're down to 22% because you have little or no RMD. OTOH, that 22% bracket may revert back to 25% in 2026. And a 2% difference in either direction isn't going to amount to that much.

You talk about the inflation effect and a dollar now being worth more. That is offset by the fact that your Roth grows tax free, but your tIRA doesn't. Run that through a spreadsheet to see for yourself.