Why Roth conversions always pay off—if you can hold on long enough

Discuss all general (i.e. non-personal) investing questions and issues, investing news, and theory.
yog
Posts: 659
Joined: Wed Jan 15, 2020 11:57 am

Re: Why Roth conversions always pay off—if you can hold on long enough

Post by yog »

McQ wrote: Sat Oct 09, 2021 12:56 pm
FiveK wrote: Fri Oct 08, 2021 7:56 pm
curmudgeon wrote: Fri Oct 08, 2021 7:15 pm If we posit $80000 of combined SS benefits (slightly lower than I had used earlier), then the other income needed to meet the constraints tends to come in a range around $60000 to $77000. The mix of qualified dividends and RMD/other income is important (to meet the other constraints). If I plug in $30000 of qualified dividends, for RMDs in the range from about $30000 to $40000 this creates the appropriate scenario. In the current tax rates, this ought to give marginal rates of 1.85X 22% (for the extra RMD), plus .85X 15% (for the dividends moving to the 15% rate), for an overall marginal rate of 53.5%.
In that scenario, ordinary income is still in the 12% bracket.

Thus one gets 1.85 * 12% + 1.85 * 15% = 49.95% for the maximum marginal federal tax rate in that area.

Change the 12% bracket to 15% and it's a 55.5% marginal rate. Of course, state taxes could come into play....
Thank you for the wiki link, FiveK, you are teaching me again. I would have had the window as (1.85 X ordinary rate) + dividend rate, since dividends go into MAGI even when not taxed. But I set up your wiki scenario in TurboTax, and it is as you have it: 1.85X (ordinary rate + dividend rate). Married couples may enjoy this Kitces post: https://www.kitces.com/blog/long-term-c ... in-0-rate/.

Curmudgeon, if you'll accept 55.5% as the post-TCJA rate on this piece of the SS torpedo, I'll score the exchange for you. I did not think it possible on Federal alone.

Next, contra cas, the window is more than wide enough to matter within the Roth conversion space. TurboTax had the additional income window running from just over $16,000 to just under $24,000, per the wiki, an $8000 band. Age 72 RMD on a $100,000 conversion is only $3650. So it's easy to imagine a super savvy Roth conversion that reduced RMDs within that window: thus converting at 22% to save $55.5% post-TCJA--even better than the more typical SS tax torpedo examples.

But it will take a sharp pencil, extremely accurate income projections, and just the right personal situation to harvest this gain.
Here's another similar more recent Kitces article:
Navigating Income Harvesting Strategies: Harvesting (0%) Capital Gains Vs Partial Roth Conversions
cas
Posts: 2245
Joined: Wed Apr 26, 2017 8:41 am

Re: Why Roth conversions always pay off—if you can hold on long enough

Post by cas »

McQ wrote: Sat Oct 09, 2021 12:56 pm Next, contra cas, the window is more than wide enough to matter within the Roth conversion space. TurboTax had the additional income window running from just over $16,000 to just under $24,000, per the wiki, an $8000 band. Age 72 RMD on a $100,000 conversion is only $3650. So it's easy to imagine a super savvy Roth conversion that reduced RMDs within that window: thus converting at 22% to save $55.5% post-TCJA--even better than the more typical SS tax torpedo examples.

But it will take a sharp pencil, extremely accurate income projections, and just the right personal situation to harvest this gain.
True. I should have read that wiki example again.

But, switching to a slightly different (but related) subject, that is where the "Bob-and-Barb" scenario*** from your posts can get a bit more tricky than it first appears.

It is certainly possible, with common configurations of income types (ordinary income, SS, QD/LTCG), that Bob and Barb really did drop from a 22% marginal rate on their Roth conversion to a 12% marginal rate on their (avoided) RMDs.
(So your example, as written, is perfectly valid and plausible. No problem there.)

But it is also possible, with a different configuration of income types, that Bob and Barb "dropped" from 22% to 22.2% (due to SS tax hump effects in the nominal 12% bracket.) (In which case, their projections were wrong, but ... shrug ... no harm, no foul.)

It is also possible, with just the right configuration of income types, that Bob and Barb "dropped" from 22% to 49.95% rates. (In which case, their projections were wrong, but their Roth conversion at 22% ended up accidentally working out for them. (Lucky them.
I sure wouldn't want to count on their mistake ending up being a win.))

It is also possible that which of the above situations they find themselves in shifts around a bit as the years pass, their income mix changes, and inflation has odd effects on the shape of the SS tax hump. (Some of the inputs that determine the shape of the SS tax hump adjust for inflation and some do not.)

All of the above is mostly just a side comment. Public tools like the one you are building have to make simplifying assumptions and rely on the user to put in valid inputs (like expected marginal tax rates), so don't take it as a criticism of your tool.

*** "Bob-and-Barb" reference (plus more extensive description of the Bob-and-Barb scenario, with spreadsheet, on page 1 of this thread):
And the comparison case will be the “plausible worst case” outcome of converting at 22% to avert RMDs that were only going to be taxed at 12%, also known as a doing a Bob-and-Barb (i.e., failing to adjust future tax brackets for inflation).
User avatar
iceport
Posts: 6048
Joined: Sat Apr 07, 2007 4:29 pm

Re: Why Roth conversions always pay off—if you can hold on long enough

Post by iceport »

McQ wrote: Thu Oct 07, 2021 4:41 pm How good can it get?

<snip>

The most likely “win”

The statute says that the 22% rate will lapse and return to its pre-TCJA value of 25% in 2026. Here are the metrics:

Code: Select all

      Real      Ratio    Tax               Wealth
      Roth      to tax   debit    Wealth   ratio
Age   surplus   debit    ROI      ratio    ROI
------------------------------------------------

85    $14,868    0.68    3.76%    1.08     0.57%
95    $55,103    2.50    5.36%    1.17     0.66%
This small change in future rate has a noticeable impact: the real Roth surplus at age 85 doubles relative to constant rates. Interestingly, the incremental impact is comparatively less at age 95, and this “not that much better” result at 95 holds across metrics. I’ll interpret it later.

<snip>

Run the numbers.
Yes, we really need to run the numbers for each of our specific circumstances. When I run the numbers on this simplified spreadsheet, a Roth conversion is not worth much at 85, even assuming my marginal tax rate reverts back to 25% from the current 22%. (It shows virtual wash at 85 if the current tax rate is made permanent.)

The reason appears to be associated with the outsized effects the assumed rates of returns have on the analysis. For this exercise, I'm using 2.4% for both the TDA account and the Roth/taxable accounts.

Unfortunately, that's not realistic. My current TDAs are invested solely in fixed income, and overwhelmingly in a stable value fund, now kicking off ~2.4%. Because I'm holding the AA constant, any conversions would be invested in a similar asset in the Roth (total bond market or TIPS fund), currently yielding even less than the stable value fund. So the low ROR is appropriate for those two locations, the TDA and the Roth.

However, any RMDs reinvested in a taxable account would likely go into an equity fund (with a corresponding shift to fixed income in the Roth). So it's not valid to assume the same ROR for both the Roth and taxable accounts, in my circumstances.
"Discipline matters more than allocation.” |—| "In finance, if you’re certain of anything, you’re out of your mind." ─William Bernstein
curmudgeon
Posts: 2630
Joined: Thu Jun 20, 2013 11:00 pm

Re: Why Roth conversions always pay off—if you can hold on long enough

Post by curmudgeon »

McQ wrote: Sat Oct 09, 2021 12:56 pm
FiveK wrote: Fri Oct 08, 2021 7:56 pm
curmudgeon wrote: Fri Oct 08, 2021 7:15 pm If we posit $80000 of combined SS benefits (slightly lower than I had used earlier), then the other income needed to meet the constraints tends to come in a range around $60000 to $77000. The mix of qualified dividends and RMD/other income is important (to meet the other constraints). If I plug in $30000 of qualified dividends, for RMDs in the range from about $30000 to $40000 this creates the appropriate scenario. In the current tax rates, this ought to give marginal rates of 1.85X 22% (for the extra RMD), plus .85X 15% (for the dividends moving to the 15% rate), for an overall marginal rate of 53.5%.
In that scenario, ordinary income is still in the 12% bracket.

Thus one gets 1.85 * 12% + 1.85 * 15% = 49.95% for the maximum marginal federal tax rate in that area.

Change the 12% bracket to 15% and it's a 55.5% marginal rate. Of course, state taxes could come into play....
Curmudgeon, if you'll accept 55.5% as the post-TCJA rate on this piece of the SS torpedo, I'll score the exchange for you. I did not think it possible on Federal alone.

Next, contra cas, the window is more than wide enough to matter within the Roth conversion space. TurboTax had the additional income window running from just over $16,000 to just under $24,000, per the wiki, an $8000 band. Age 72 RMD on a $100,000 conversion is only $3650. So it's easy to imagine a super savvy Roth conversion that reduced RMDs within that window: thus converting at 22% to save $55.5% post-TCJA--even better than the more typical SS tax torpedo examples.

But it will take a sharp pencil, extremely accurate income projections, and just the right personal situation to harvest this gain.
I would agree that 55.5% is that inner window marginal rate. I'd also agree that it is pretty much a corner case in consideration of sustained scenarios for RMDs across retirement; it takes a substantial taxable account to throw off that much dividends and also needs a specific range of RMDs.

A more common reason to hit that window would be occasional bumps of LTCG. Perhaps someone has a taxable holding that they would like to reduce or rebalance. If this were going to generate $100,000 of LTCG, they might expect that splitting the sale across multiple years help mitigate the tax hit. If instead the split placed them in or over this window for several years, that might have a nasty tax cost (and one that might leave them scratching their heads come tax time).
Topic Author
McQ
Posts: 1414
Joined: Fri Jun 18, 2021 12:21 am
Location: California

Re: Why Roth conversions always pay off—if you can hold on long enough

Post by McQ »

cas wrote: Sat Oct 09, 2021 2:31 pm
McQ wrote: Sat Oct 09, 2021 12:56 pm Next, contra cas, the window is more than wide enough to matter within the Roth conversion space. TurboTax had the additional income window running from just over $16,000 to just under $24,000, per the wiki, an $8000 band. Age 72 RMD on a $100,000 conversion is only $3650. So it's easy to imagine a super savvy Roth conversion that reduced RMDs within that window: thus converting at 22% to save $55.5% post-TCJA--even better than the more typical SS tax torpedo examples.

But it will take a sharp pencil, extremely accurate income projections, and just the right personal situation to harvest this gain.
True. I should have read that wiki example again.

But, switching to a slightly different (but related) subject, that is where the "Bob-and-Barb" scenario*** from your posts can get a bit more tricky than it first appears.

It is certainly possible, with common configurations of income types (ordinary income, SS, QD/LTCG), that Bob and Barb really did drop from a 22% marginal rate on their Roth conversion to a 12% marginal rate on their (avoided) RMDs.
(So your example, as written, is perfectly valid and plausible. No problem there.)

But it is also possible, with a different configuration of income types, that Bob and Barb "dropped" from 22% to 22.2% (due to SS tax hump effects in the nominal 12% bracket.) (In which case, their projections were wrong, but ... shrug ... no harm, no foul.)

It is also possible, with just the right configuration of income types, that Bob and Barb "dropped" from 22% to 49.95% rates. (In which case, their projections were wrong, but their Roth conversion at 22% ended up accidentally working out for them. (Lucky them.
I sure wouldn't want to count on their mistake ending up being a win.))

It is also possible that which of the above situations they find themselves in shifts around a bit as the years pass, their income mix changes, and inflation has odd effects on the shape of the SS tax hump. (Some of the inputs that determine the shape of the SS tax hump adjust for inflation and some do not.)

All of the above is mostly just a side comment. Public tools like the one you are building have to make simplifying assumptions and rely on the user to put in valid inputs (like expected marginal tax rates), so don't take it as a criticism of your tool.

*** "Bob-and-Barb" reference (plus more extensive description of the Bob-and-Barb scenario, with spreadsheet, on page 1 of this thread):
And the comparison case will be the “plausible worst case” outcome of converting at 22% to avert RMDs that were only going to be taxed at 12%, also known as a doing a Bob-and-Barb (i.e., failing to adjust future tax brackets for inflation).
Excellent summary, cas. I never did much with decision trees, but entered my field back when expectancy-value models were widespread in the social sciences. So one can imagine the decision about whether to convert at 22% set up as follows:
-(probability weight X 22-->12 dollar outcome) +
-(probability weight X 22 --> 22 dollar outcome) +
-(probability weight X 22 --> 25 dollar outcome) +
-(probability weight X 22 --> 28 dollar outcome) +
... [all others having some probability > .01].
where the probability weights sum to 1.0. If the probabilities could be estimated rather than just guessed, the decision tree might be an aid. In any case, it helps address the reality, as your post sketched out, that multiple outcomes from a conversion are always possible.
You can take the academic out of the classroom by retirement, but you can't ever take the classroom out of his tone, style, and manner of approach.
Topic Author
McQ
Posts: 1414
Joined: Fri Jun 18, 2021 12:21 am
Location: California

Re: Why Roth conversions always pay off—if you can hold on long enough

Post by McQ »

iceport wrote: Sat Oct 09, 2021 3:47 pm
McQ wrote: Thu Oct 07, 2021 4:41 pm How good can it get?

<snip>

The most likely “win”

The statute says that the 22% rate will lapse and return to its pre-TCJA value of 25% in 2026. Here are the metrics:

Code: Select all

      Real      Ratio    Tax               Wealth
      Roth      to tax   debit    Wealth   ratio
Age   surplus   debit    ROI      ratio    ROI
------------------------------------------------

85    $14,868    0.68    3.76%    1.08     0.57%
95    $55,103    2.50    5.36%    1.17     0.66%
This small change in future rate has a noticeable impact: the real Roth surplus at age 85 doubles relative to constant rates. Interestingly, the incremental impact is comparatively less at age 95, and this “not that much better” result at 95 holds across metrics. I’ll interpret it later.

<snip>

Run the numbers.
Yes, we really need to run the numbers for each of our specific circumstances. When I run the numbers on this simplified spreadsheet, a Roth conversion is not worth much at 85, even assuming my marginal tax rate reverts back to 25% from the current 22%. (It shows virtual wash at 85 if the current tax rate is made permanent.)

The reason appears to be associated with the outsized effects the assumed rates of returns have on the analysis. For this exercise, I'm using 2.4% for both the TDA account and the Roth/taxable accounts.

Unfortunately, that's not realistic. My current TDAs are invested solely in fixed income, and overwhelmingly in a stable value fund, now kicking off ~2.4%. Because I'm holding the AA constant, any conversions would be invested in a similar asset in the Roth (total bond market or TIPS fund), currently yielding even less than the stable value fund. So the low ROR is appropriate for those two locations, the TDA and the Roth.

However, any RMDs reinvested in a taxable account would likely go into an equity fund (with a corresponding shift to fixed income in the Roth). So it's not valid to assume the same ROR for both the Roth and taxable accounts, in my circumstances.
Fair point, iceport. I have a future post in preparation where I'll ratchet down the rate of return on the accounts to show the impact. However, I had not intended to go all the way down to stable value type returns; your post shows the need to do so.

If you downloaded the spreadsheet, easy enough to add an entry to the first row for an ROR specific to the taxable account; then change the taxable column first cell to reference that ROR instead of the current, Roth = taxable cell; use fill to complete the column. Not much more than a dozen keystrokes, I should think. The TDA return is already called out in a separate cell; now you can vary all three independently.
You can take the academic out of the classroom by retirement, but you can't ever take the classroom out of his tone, style, and manner of approach.
marcopolo
Posts: 8411
Joined: Sat Dec 03, 2016 9:22 am

Re: Why Roth conversions always pay off—if you can hold on long enough

Post by marcopolo »

McQ wrote: Sun Oct 10, 2021 12:46 pm
iceport wrote: Sat Oct 09, 2021 3:47 pm
McQ wrote: Thu Oct 07, 2021 4:41 pm How good can it get?

<snip>

The most likely “win”

The statute says that the 22% rate will lapse and return to its pre-TCJA value of 25% in 2026. Here are the metrics:

Code: Select all

      Real      Ratio    Tax               Wealth
      Roth      to tax   debit    Wealth   ratio
Age   surplus   debit    ROI      ratio    ROI
------------------------------------------------

85    $14,868    0.68    3.76%    1.08     0.57%
95    $55,103    2.50    5.36%    1.17     0.66%
This small change in future rate has a noticeable impact: the real Roth surplus at age 85 doubles relative to constant rates. Interestingly, the incremental impact is comparatively less at age 95, and this “not that much better” result at 95 holds across metrics. I’ll interpret it later.

<snip>

Run the numbers.
Yes, we really need to run the numbers for each of our specific circumstances. When I run the numbers on this simplified spreadsheet, a Roth conversion is not worth much at 85, even assuming my marginal tax rate reverts back to 25% from the current 22%. (It shows virtual wash at 85 if the current tax rate is made permanent.)

The reason appears to be associated with the outsized effects the assumed rates of returns have on the analysis. For this exercise, I'm using 2.4% for both the TDA account and the Roth/taxable accounts.

Unfortunately, that's not realistic. My current TDAs are invested solely in fixed income, and overwhelmingly in a stable value fund, now kicking off ~2.4%. Because I'm holding the AA constant, any conversions would be invested in a similar asset in the Roth (total bond market or TIPS fund), currently yielding even less than the stable value fund. So the low ROR is appropriate for those two locations, the TDA and the Roth.

However, any RMDs reinvested in a taxable account would likely go into an equity fund (with a corresponding shift to fixed income in the Roth). So it's not valid to assume the same ROR for both the Roth and taxable accounts, in my circumstances.
Fair point, iceport. I have a future post in preparation where I'll ratchet down the rate of return on the accounts to show the impact. However, I had not intended to go all the way down to stable value type returns; your post shows the need to do so.

If you downloaded the spreadsheet, easy enough to add an entry to the first row for an ROR specific to the taxable account; then change the taxable column first cell to reference that ROR instead of the current, Roth = taxable cell; use fill to complete the column. Not much more than a dozen keystrokes, I should think. The TDA return is already called out in a separate cell; now you can vary all three independently.
The problem with doing what you suggest (setting different return parameter per account type) is that it falls onto the same pitfall as some other Roth analysis tools.

In the scenario iceport is describing one keeps their overall asset allocation fixed (lets say 60/40). Then favors fixed income in TDA, equities in Roth and taxable accounts. But, those preferences are secondary to keeping the asset allocation consistent (don't add/reduce risk profile in the analysis). So, as money is spent, or moved between. accounts, the relative mix (and resulting return) per account changes as the asset mix changes in each account to maintain the same portfolio level asset allocation. This can get very complex to model, even more so if one considers tax-adjusted their asset allocation.

By assigning a static return parameter by account type, the analysis gets skewed by that more so than the tax impacts we are trying to analyze. For example, if we start with the condition that TDA holds fixed income (with low returns) and Roth hold equities (with higher return), then the simulation will show that doing large Roth Conversions will pay off in a big way. But, that has nothing to do with tax arbitrage, but rather is driven by moving assets to the higher return asset. So, one ends up with a 100% equity portfolio, probably not what was intended.

This has tripped up many people using a commonly reccommended tool. I believe this has contributed to an over zealous recommendation for Roth Conversions. Interestingly, advocates of such tools recommend setting the asset allocation (and returns) of all accounts to be the same as a kind of work around to reduce this effect, which brings us full circle back to assuming same return everywhere as you have done!
Once in a while you get shown the light, in the strangest of places if you look at it right.
User avatar
iceport
Posts: 6048
Joined: Sat Apr 07, 2007 4:29 pm

Re: Why Roth conversions always pay off—if you can hold on long enough

Post by iceport »

McQ wrote: Sun Oct 10, 2021 12:46 pm
iceport wrote: Sat Oct 09, 2021 3:47 pm Yes, we really need to run the numbers for each of our specific circumstances. When I run the numbers on this simplified spreadsheet, a Roth conversion is not worth much at 85, even assuming my marginal tax rate reverts back to 25% from the current 22%. (It shows virtual wash at 85 if the current tax rate is made permanent.)

The reason appears to be associated with the outsized effects the assumed rates of returns have on the analysis. For this exercise, I'm using 2.4% for both the TDA account and the Roth/taxable accounts.

Unfortunately, that's not realistic. My current TDAs are invested solely in fixed income, and overwhelmingly in a stable value fund, now kicking off ~2.4%. Because I'm holding the AA constant, any conversions would be invested in a similar asset in the Roth (total bond market or TIPS fund), currently yielding even less than the stable value fund. So the low ROR is appropriate for those two locations, the TDA and the Roth.

However, any RMDs reinvested in a taxable account would likely go into an equity fund (with a corresponding shift to fixed income in the Roth). So it's not valid to assume the same ROR for both the Roth and taxable accounts, in my circumstances.
Fair point, iceport. I have a future post in preparation where I'll ratchet down the rate of return on the accounts to show the impact. However, I had not intended to go all the way down to stable value type returns; your post shows the need to do so.
Well, I'm not sure how common my situation is. For example, TDAs comprise only about 28% of the portfolio — which is the only reason why most of it can be in a stable value fund — with 24% in Roth and 48% in taxable. I don't imagine that's a very common asset location scenario.

However, it's probably very common to selectively direct the lowest returning assets to TDAs, and the most tax-efficient (and typically highest returning) assets to taxable accounts. The Roth catches everything/anything else.

McQ wrote: Sun Oct 10, 2021 12:46 pm If you downloaded the spreadsheet, easy enough to add an entry to the first row for an ROR specific to the taxable account; then change the taxable column first cell to reference that ROR instead of the current, Roth = taxable cell; use fill to complete the column. Not much more than a dozen keystrokes, I should think. The TDA return is already called out in a separate cell; now you can vary all three independently.
Thanks for that. I made the simple modifications you suggested, and the numbers are absolutely grim! I set it up with taxable returns at 6% (I'm apparently not nearly as bullish as you), and TDA/Roth account returns of 2.0%.

Conversions loose money unless the future tax rate increases to almost 30%! Under a constant tax scenario, there's about a $7.5k loss at 85. What's more, the losses only keep compounding as time goes by!

Good thing I don't pay the taxes for my Roth conversions from the converted sum, but instead use my credit union savings account which is not counted in the portfolio balance. :|


marcopolo wrote: Sun Oct 10, 2021 2:19 pm
McQ wrote: Sun Oct 10, 2021 12:46 pm If you downloaded the spreadsheet, easy enough to add an entry to the first row for an ROR specific to the taxable account; then change the taxable column first cell to reference that ROR instead of the current, Roth = taxable cell; use fill to complete the column. Not much more than a dozen keystrokes, I should think. The TDA return is already called out in a separate cell; now you can vary all three independently.
The problem with doing what you suggest (setting different return parameter per account type) is that it falls onto the same pitfall as some other Roth analysis tools.

In the scenario iceport is describing one keeps their overall asset allocation fixed (lets say 60/40). Then favors fixed income in TDA, equities in Roth and taxable accounts. But, those preferences are secondary to keeping the asset allocation consistent (don't add/reduce risk profile in the analysis). So, as money is spent, or moved between. accounts, the relative mix (and resulting return) per account changes as the asset mix changes in each account to maintain the same portfolio level asset allocation. This can get very complex to model, even more so if one considers tax-adjusted their asset allocation.

By assigning a static return parameter by account type, the analysis gets skewed by that more so than the tax impacts we are trying to analyze. For example, if we start with the condition that TDA holds fixed income (with low returns) and Roth hold equities (with higher return), then the simulation will show that doing large Roth Conversions will pay off in a big way. But, that has nothing to do with tax arbitrage, but rather is driven by moving assets to the higher return asset. So, one ends up with a 100% equity portfolio, probably not what was intended.

This has tripped up many people using a commonly reccommended tool. I believe this has contributed to an over zealous recommendation for Roth Conversions. Interestingly, advocates of such tools recommend setting the asset allocation (and returns) of all accounts to be the same as a kind of work around to reduce this effect, which brings us full circle back to assuming same return everywhere as you have done!
Your analysis is spot-on, marcopolo. And the thing is, the effects of these ROR assumptions are not marginal — they are capable of completely upending the result!

However, even with this dilemma, the best way to approximate reality might be situational. In my circumstances, for example, it appears that using the same very low ROR in both the TDA and the Roth account is appropriate. The Roth balance being tracked in the spreadsheet is only from the initial conversion, which basically just moves a low-returning asset sum from the TDA to the Roth. Because the entire Roth balance isn't involved in the computations in this particular spreadsheet, isn't it appropriate just to use the same return in both the TDA and the Roth, assuming 100% low-returning fixed income in the TDA, as in my case?

In other circumstances, I agree with you, modeling the real world conditions could get messy. The typical strategy if the TDAs hold a wide variety of assets would be to move the highest-returning assets from the TDA first, thereby reducing the expected return in the TDA going forward.
"Discipline matters more than allocation.” |—| "In finance, if you’re certain of anything, you’re out of your mind." ─William Bernstein
Topic Author
McQ
Posts: 1414
Joined: Fri Jun 18, 2021 12:21 am
Location: California

Re: Why Roth conversions always pay off—if you can hold on long enough

Post by McQ »

Part III: Loose ends and special cases

Part I tried to support the point that Roth conversions always pay off if held long enough (which includes living long enough, per marcopolo). A good counterexample from cas pointed out that if, after the conversion, the retiree was in the 0% tax bracket, and stayed there their entire life + 10, then there could be no tax drag and the Roth conversion could never catch up / move ahead. But if 0% were the expected tax rate, why would the converter have undertaken a conversion which precipitated tax? It would seem that extreme ignorance of the tax law, or unexpected extremely adverse outcomes, are required to make this counterexample stick.

Part II switched gears and examined how slow and low most conversion payoffs must be, when evaluated during life, which will typically mean after a relatively short interval of two decades or so.

Part III now moves on to consider deferred and postponed questions of the form, What about …?

Today’s post answers the question, What if capital gains are not realized each year, but deferred? Marcopolo, who took the trouble to reconstruct the spreadsheet, has pressed this question.

Next post will address, What about a conversion before the age of 71? Does it make a difference if I convert at, say, 64? Woodspinner raised this question way up thread.

And the following post will examine, What if I paid the tax from outside the conversion, how much difference does that make?

I have a few more special cases on my handlist (including widow taxed as a single, and a sudden expense), but if there is something you asked about earlier, and you are not seeing it on this list, now would be a good time to post the query again.

Deferred capital gains

In the SS images previously posted, the taxable account builds up from reinvested RMDs, and it generates appreciation at the 10% rate applied to all the accounts pre-tax. To clarify the operation of tax drag, I marked to market each year, applying the 15% tax to both capital appreciation and dividends (assumed to be split 8/2). It seemed the only way to track results to the penny, and to make visible the exact effect of tax drag.

But as marcopolo argued, and as I argued in the SSRN paper, with today’s ETF structure, it is relatively easy to record only qualified dividends each year, with unrealized capital gains steadily accumulating until liquidation. And anyone savvy enough to scope out a Roth conversion would likely be savvy enough to take advantage of this tax deferral.

Result: annual tax drag will be reduced from 150 basis points (all the 10% gain X 15% rate) to 30 basis points (only the 2% dividends X 15%). Accordingly, the magnitude of the conversion pay off must grow more slowly. The question today is, How much more slowly? The initial tests will again assume evaluation-while-alive, at age 85 or 95. Things would go rather worse for the conversion analysis if the embedded capital gains were never taxed because the account was held to death and a step up received; I’ll look at that case later.

Tax drag is more important when conversion and RMD tax rates stay constant; so let’s take a look at the 22% -- 22% case to start. Likewise, tax drag is even more important if future rates drop, and that case will be examined later.

Revising the SS set up

Several columns have to be changed and new ones added. For instance, dividends have to be called out separately from capital gains, with the one taxed annually but not the other. And most important, cost basis on the taxable account has to be tracked. I make cost basis to be:
1) After-tax value of each RMD invested +
2) After-tax value of each dividend when reinvested.

Total after tax value of the no-conversion path is:
As before, [TDA account value X (1 – ordinary tax rate)] +
Taxable account value – [(taxable value – cost basis) X cap gains rate]
…with the second entry subtracting tax on all the built up gain.

Here is the revised ss image; down the road, I’ll add this and other special cases to the downloadable ss (once I decide how to address the security issue FiveK raised).

Image

Row 3, the column formula row, has a buff background in cases where new columns were added or column letter labels changed. Columns not important to this analysis have been hidden.

Because the ss is getting more unwieldy and harder to peruse, I’ll excerpt results in the form of the key metrics introduced a few posts back. Recall that upthread, the 22% -- 22% conversion gave these results:

Code: Select all

      Real      Ratio    Tax               Wealth
      Roth      to tax   debit    Wealth   ratio
Age   surplus   debit    ROI      ratio    ROI
------------------------------------------------
85    $ 7,629    0.35    2.15%    1.04     0.28%
95    $42,190    1.92    4.56%    1.13     0.49%

Here is how the dollar outcome changes now that capital gains are not taxed each year:

Code: Select all

      Real      Ratio    Tax               Wealth
      Roth      to tax   debit    Wealth   ratio
Age   surplus   debit    ROI      ratio    ROI
------------------------------------------------
85    $6,128    0.27     1.77%    1.0323   0.23%
95    $29,490   1.34     3.61%    1.0846   0.34%
Result: at age 85 the payoff from the Roth conversion has dropped about $1500 in real terms, or 20%. The drop at age 95 is larger, closer to 30%. But the pay off at this later age is still almost $30,000 in real terms.

Glass half full or glass half empty? Conversion payoffs are reduced, but they are still ample even at a constant tax rate—the circumstance where, before my SSRN paper, conventional wisdom held that there could be no pay off to a Roth conversion.

Which raises the next question: many conversions are undertaken precisely because some increase in future tax rates is anticipated. If that increase occurs, how much difference does it make to leave capital gains untaxed until the end? Here are the 22% --25% results from before, with gains taxed each year:

Code: Select all

      Real      Ratio    Tax               Wealth
      Roth      to tax   debit    Wealth   ratio
Age   surplus   debit    ROI      ratio    ROI
------------------------------------------------
85    $14,868    0.68    3.76%    1.08     0.57%
95    $55,103    2.50    5.36%    1.17     0.66%
And here are the new results, with capital gains not taxed until liquidation:

Code: Select all

      Real      Ratio    Tax               Wealth
      Roth      to tax   debit    Wealth   ratio
Age   surplus   debit    ROI      ratio    ROI
------------------------------------------------
85    $13,424   0.61     3.46%    1.0736   0.51%
95    $42,892   1.95     4.61%    1.1280   0.50%

There is not much change at age 85, a drop of about 10% in payoff; the change is more noticeable at age 95, a drop of just over one-fifth, but still a smaller drop than in the constant rate case. The interpretation is straightforward: when conversion and RMD rates are constant, tax drag is the only source of Roth conversion payoff. Embedding capital gains reduces tax drag, hence makes conversion outcomes weaker. However, when future rates are higher, tax drag is only part of the payoff, and not realizing capital gains cannot have as much effect.

The logical next question is what happens when tax drag is reduced in this way and future rates drop. I’ll look at that next time. The focus will be the dread Bob-and-Barb: convert at 22%, to avoid RMDs taxed at 12%.

But first, alert readers might have begun to wonder whether I ceded too much ground to marcopolo’s criticism. I allowed him, as it were, to wave a hand and make the Net Investment Income Tax to disappear. Let me explain.

When a 22% tax rate on RMDs is assumed, and capital gains are taken annually, there is no need to add the NIIT rate of 3.8%. To be in the 22% bracket, a couple must have less than $200,450 AGI; and married joint NIIT only kicks in when AGI exceeds $250,000.

But when decades of embedded capital gains are liquidated, along with the remaining TDA balance, there must be NIIT. At age 85, over $45,000 of capital gain has built up (constant case); at 95, over $275,000. Corresponding TDA balances that would be added to ordinary income are themselves over $200,000. Therefore, when capital gains go unrealized until final liquidation, the NIIT rate must be assumed in evaluation. Unless, that is, we hew to that famous economist joke: “assume we have no NIIT.”

*I note in passing that the keep-almost-everything-off-camera approach used in the spreadsheets for this thread, alas, facilitates the inadvertent neglect of NIIT and probably more besides.

Here is how the with-NIIT results, total rate 18.8%, look for the 22% constant rate case.

Code: Select all

      Real      Ratio    Tax               Wealth
      Roth      to tax   debit    Wealth   ratio
Age   surplus   debit    ROI      ratio    ROI
------------------------------------------------
85    $7,277    0.33     2.06%    1.0386   0.27%
95   $34,649    1.57     4.02%    1.1009   0.40%
Oops: now leaving capital gains to embed has produced a slightly worse payoff at age 85, and almost a 20% decrement at age 95. These poor taxpayers would have been better off paying tax as they go. Unless they could hold unto death.

As the saying goes, the rate on tax paid today is fixed; but tax deferred is tax unknown.

I believe this post shows that embedding or not embedding capital gains does not make the huge impact that marcopolo seemed to assume, when cautioning BH to wait for a better analytic framework before accepting the metrics laid out upthread. It seems to me to have been a reasonable tradeoff to defer this analysis until much later in the thread, so that early posts could nail down the operations underlying any conversion payoff with the greater clarity that flows from the simpler set up where all gains are taxed each year.
You can take the academic out of the classroom by retirement, but you can't ever take the classroom out of his tone, style, and manner of approach.
User avatar
LilyFleur
Posts: 3497
Joined: Fri Mar 02, 2018 9:36 pm

Re: Why Roth conversions always pay off—if you can hold on long enough

Post by LilyFleur »

So all of this is based on 10% earnings each year?
What happens if earnings are 4%?
I am editing this to say: spreadsheets are only as good as their assumptions. Assuming 10% earnings each year is a big assumption, and depending on the size of the tax-deferred savings and the horizon to RMDs, a change of + or - a few percentage points can make a difference in the tax bracket during RMD years.
Many Bogleheads in their 50s and older have asset allocations that include a good percentage of bonds and cash.
Last edited by LilyFleur on Mon Oct 11, 2021 4:20 pm, edited 1 time in total.
User avatar
FiveK
Posts: 15693
Joined: Sun Mar 16, 2014 2:43 pm

Re: Why Roth conversions always pay off—if you can hold on long enough

Post by FiveK »

LilyFleur wrote: Mon Oct 11, 2021 4:11 pm What happens if earnings are 4%?
You could try the spreadsheet linked in How do RMDs affect Roth conversion choices? to see what that suggests.
User avatar
LilyFleur
Posts: 3497
Joined: Fri Mar 02, 2018 9:36 pm

Re: Why Roth conversions always pay off—if you can hold on long enough

Post by LilyFleur »

FiveK wrote: Mon Oct 11, 2021 4:16 pm
LilyFleur wrote: Mon Oct 11, 2021 4:11 pm What happens if earnings are 4%?
You could try the spreadsheet linked in How do RMDs affect Roth conversion choices? to see what that suggests.
Sorry, I edited my post to be more clear. I actually do know what happens if the earnings change. Believe me, I have run the Schwab RMD calculator many, many times.
2pedals
Posts: 1988
Joined: Wed Dec 31, 2014 11:31 am

Re: Why Roth conversions always pay off—if you can hold on long enough

Post by 2pedals »

The posts by McQ and others are quite fascinating. Since am we still are early retirement (ages 62 and 60) and have done 3 annual partial Roth conversions. I just don't know how I am going to take this information and change or modify my plans. I would be interested to know if there are any major flaws in the tools I have used such as Pralana Gold, Retiree Portfolio Model (RPM), and i-ORP that are being uncovered here. I am aware of the issues with trying to use different asset allocations in different account types (TDA, Roth, and taxable).
marcopolo
Posts: 8411
Joined: Sat Dec 03, 2016 9:22 am

Re: Why Roth conversions always pay off—if you can hold on long enough

Post by marcopolo »

McQ wrote: Mon Oct 11, 2021 4:04 pm Part III: Loose ends and special cases

Part I tried to support the point that Roth conversions always pay off if held long enough (which includes living long enough, per marcopolo). A good counterexample from cas pointed out that if, after the conversion, the retiree was in the 0% tax bracket, and stayed there their entire life + 10, then there could be no tax drag and the Roth conversion could never catch up / move ahead. But if 0% were the expected tax rate, why would the converter have undertaken a conversion which precipitated tax? It would seem that extreme ignorance of the tax law, or unexpected extremely adverse outcomes, are required to make this counterexample stick.

Part II switched gears and examined how slow and low most conversion payoffs must be, when evaluated during life, which will typically mean after a relatively short interval of two decades or so.

Part III now moves on to consider deferred and postponed questions of the form, What about …?

Today’s post answers the question, What if capital gains are not realized each year, but deferred? Marcopolo, who took the trouble to reconstruct the spreadsheet, has pressed this question.

Next post will address, What about a conversion before the age of 71? Does it make a difference if I convert at, say, 64? Woodspinner raised this question way up thread.

And the following post will examine, What if I paid the tax from outside the conversion, how much difference does that make?

I have a few more special cases on my handlist (including widow taxed as a single, and a sudden expense), but if there is something you asked about earlier, and you are not seeing it on this list, now would be a good time to post the query again.

Deferred capital gains

In the SS images previously posted, the taxable account builds up from reinvested RMDs, and it generates appreciation at the 10% rate applied to all the accounts pre-tax. To clarify the operation of tax drag, I marked to market each year, applying the 15% tax to both capital appreciation and dividends (assumed to be split 8/2). It seemed the only way to track results to the penny, and to make visible the exact effect of tax drag.

But as marcopolo argued, and as I argued in the SSRN paper, with today’s ETF structure, it is relatively easy to record only qualified dividends each year, with unrealized capital gains steadily accumulating until liquidation. And anyone savvy enough to scope out a Roth conversion would likely be savvy enough to take advantage of this tax deferral.

Result: annual tax drag will be reduced from 150 basis points (all the 10% gain X 15% rate) to 30 basis points (only the 2% dividends X 15%). Accordingly, the magnitude of the conversion pay off must grow more slowly. The question today is, How much more slowly? The initial tests will again assume evaluation-while-alive, at age 85 or 95. Things would go rather worse for the conversion analysis if the embedded capital gains were never taxed because the account was held to death and a step up received; I’ll look at that case later.

Tax drag is more important when conversion and RMD tax rates stay constant; so let’s take a look at the 22% -- 22% case to start. Likewise, tax drag is even more important if future rates drop, and that case will be examined later.

Revising the SS set up

Several columns have to be changed and new ones added. For instance, dividends have to be called out separately from capital gains, with the one taxed annually but not the other. And most important, cost basis on the taxable account has to be tracked. I make cost basis to be:
1) After-tax value of each RMD invested +
2) After-tax value of each dividend when reinvested.

Total after tax value of the no-conversion path is:
As before, [TDA account value X (1 – ordinary tax rate)] +
Taxable account value – [(taxable value – cost basis) X cap gains rate]
…with the second entry subtracting tax on all the built up gain.

Here is the revised ss image; down the road, I’ll add this and other special cases to the downloadable ss (once I decide how to address the security issue FiveK raised).

Image

Row 3, the column formula row, has a buff background in cases where new columns were added or column letter labels changed. Columns not important to this analysis have been hidden.

Because the ss is getting more unwieldy and harder to peruse, I’ll excerpt results in the form of the key metrics introduced a few posts back. Recall that upthread, the 22% -- 22% conversion gave these results:

Code: Select all

      Real      Ratio    Tax               Wealth
      Roth      to tax   debit    Wealth   ratio
Age   surplus   debit    ROI      ratio    ROI
------------------------------------------------
85    $ 7,629    0.35    2.15%    1.04     0.28%
95    $42,190    1.92    4.56%    1.13     0.49%

Here is how the dollar outcome changes now that capital gains are not taxed each year:

Code: Select all

      Real      Ratio    Tax               Wealth
      Roth      to tax   debit    Wealth   ratio
Age   surplus   debit    ROI      ratio    ROI
------------------------------------------------
85    $6,128    0.27     1.77%    1.0323   0.23%
95    $29,490   1.34     3.61%    1.0846   0.34%
Result: at age 85 the payoff from the Roth conversion has dropped about $1500 in real terms, or 20%. The drop at age 95 is larger, closer to 30%. But the pay off at this later age is still almost $30,000 in real terms.

Glass half full or glass half empty? Conversion payoffs are reduced, but they are still ample even at a constant tax rate—the circumstance where, before my SSRN paper, conventional wisdom held that there could be no pay off to a Roth conversion.

Which raises the next question: many conversions are undertaken precisely because some increase in future tax rates is anticipated. If that increase occurs, how much difference does it make to leave capital gains untaxed until the end? Here are the 22% --25% results from before, with gains taxed each year:

Code: Select all

      Real      Ratio    Tax               Wealth
      Roth      to tax   debit    Wealth   ratio
Age   surplus   debit    ROI      ratio    ROI
------------------------------------------------
85    $14,868    0.68    3.76%    1.08     0.57%
95    $55,103    2.50    5.36%    1.17     0.66%
And here are the new results, with capital gains not taxed until liquidation:

Code: Select all

      Real      Ratio    Tax               Wealth
      Roth      to tax   debit    Wealth   ratio
Age   surplus   debit    ROI      ratio    ROI
------------------------------------------------
85    $13,424   0.61     3.46%    1.0736   0.51%
95    $42,892   1.95     4.61%    1.1280   0.50%

There is not much change at age 85, a drop of about 10% in payoff; the change is more noticeable at age 95, a drop of just over one-fifth, but still a smaller drop than in the constant rate case. The interpretation is straightforward: when conversion and RMD rates are constant, tax drag is the only source of Roth conversion payoff. Embedding capital gains reduces tax drag, hence makes conversion outcomes weaker. However, when future rates are higher, tax drag is only part of the payoff, and not realizing capital gains cannot have as much effect.

The logical next question is what happens when tax drag is reduced in this way and future rates drop. I’ll look at that next time. The focus will be the dread Bob-and-Barb: convert at 22%, to avoid RMDs taxed at 12%.

But first, alert readers might have begun to wonder whether I ceded too much ground to marcopolo’s criticism. I allowed him, as it were, to wave a hand and make the Net Investment Income Tax to disappear. Let me explain.

When a 22% tax rate on RMDs is assumed, and capital gains are taken annually, there is no need to add the NIIT rate of 3.8%. To be in the 22% bracket, a couple must have less than $200,450 AGI; and married joint NIIT only kicks in when AGI exceeds $250,000.

But when decades of embedded capital gains are liquidated, along with the remaining TDA balance, there must be NIIT. At age 85, over $45,000 of capital gain has built up (constant case); at 95, over $275,000. Corresponding TDA balances that would be added to ordinary income are themselves over $200,000. Therefore, when capital gains go unrealized until final liquidation, the NIIT rate must be assumed in evaluation. Unless, that is, we hew to that famous economist joke: “assume we have no NIIT.”

*I note in passing that the keep-almost-everything-off-camera approach used in the spreadsheets for this thread, alas, facilitates the inadvertent neglect of NIIT and probably more besides.

Here is how the with-NIIT results, total rate 18.8%, look for the 22% constant rate case.

Code: Select all

      Real      Ratio    Tax               Wealth
      Roth      to tax   debit    Wealth   ratio
Age   surplus   debit    ROI      ratio    ROI
------------------------------------------------
85    $7,277    0.33     2.06%    1.0386   0.27%
95   $34,649    1.57     4.02%    1.1009   0.40%
Oops: now leaving capital gains to embed has produced a slightly worse payoff at age 85, and almost a 20% decrement at age 95. These poor taxpayers would have been better off paying tax as they go. Unless they could hold unto death.

As the saying goes, the rate on tax paid today is fixed; but tax deferred is tax unknown.

I believe this post shows that embedding or not embedding capital gains does not make the huge impact that marcopolo seemed to assume, when cautioning BH to wait for a better analytic framework before accepting the metrics laid out upthread. It seems to me to have been a reasonable tradeoff to defer this analysis until much later in the thread, so that early posts could nail down the operations underlying any conversion payoff with the greater clarity that flows from the simpler set up where all gains are taxed each year.
You started with the premise that RMDs don't get spent and are passed on to heirs. You know, because everyone has 6 figure pensions and near max x2 Soc. Sec.. Now you have forced the retiree to liquidate all their RMDs, in some cases all at once, paying NIIT!

If I didn't know better, I would say you are putting your finger on the scale a bit :beer

What happens when we keep that same scenario and the accumulated gains get step-up basis to the heirs.
Once in a while you get shown the light, in the strangest of places if you look at it right.
cas
Posts: 2245
Joined: Wed Apr 26, 2017 8:41 am

Re: Why Roth conversions always pay off—if you can hold on long enough

Post by cas »

McQ wrote: Mon Oct 11, 2021 4:04 pm The logical next question is what happens when tax drag is reduced in this way and future rates drop. I’ll look at that next time. The focus will be the dread Bob-and-Barb: convert at 22%, to avoid RMDs taxed at 12%.
Careful. The Bob-and-Barb scenario has a significant flaw (assuming tax law since at least Bush 2). This flaw hasn't particularly mattered to the points you have been making thus far using that particular scenario, so I haven't mentioned it.

But if you want to use that scenario to make any points about tax drag on the taxable account, it will need to be addressed.

Here are specifics of the Bob-and-Barb scenario in your spreadsheet on Page 1 of this thread:

marginal tax rate on conversion (pre-RMD): 22%
marginal tax rate on ordinary income post-RMD: 12%
marginal tax rate on QD/LTCG post-RMD: 15%

Under tax law since at least Bush 2, this scenario is not possible***.

*If* the marginal tax rate on QD/LTCG really is 15%, *then* the marginal rate on the RMDs is 27% (rather than 12%).

The 27% is due to a "shadow bracket" or "bump zone" that results from the interaction of ordinary income with QD/LTCG income.
It overlays/replaces the upper portion of the nominal 12% bracket. (See the Kitces "bump zone" article you linked above, Example 3.)

If the marginal tax rate on the RMDs in "doing a Bob-and-Barb" ends up being in the 27% "shadow bracket"/"bump zone", then the Roth gains in a different way than taxable account tax drag.

***except for a couple hundred dollar space when an apparent error in the drafting of TCJA offset the top of the 0% QD/LTCG tax bracket from the top of the 12% bracket for ordinary income by a couple of hundred dollars.
Last edited by cas on Tue Oct 12, 2021 11:19 am, edited 1 time in total.
cas
Posts: 2245
Joined: Wed Apr 26, 2017 8:41 am

Re: Why Roth conversions always pay off—if you can hold on long enough

Post by cas »

Re: cas' counter-example to the "Why Roth conversions *always* pay off if you wait long enough" assertion...
McQ wrote: Mon Oct 11, 2021 4:04 pm But if 0% were the expected tax rate, why would the converter have undertaken a conversion which precipitated tax?
Because my counter-example doesn't require the marginal tax rate on ordinary income (the RMDs) to be 0%.

It requires only that the marginal tax rate on qualified dividends and long term capital gains be 0%.

In fact, if you assume compliance with current tax law, the Bob-and-Barb scenario that you keep bringing up, with its 12% marginal tax rate on the RMDs, is a counter-example:

Under tax law since at least Bush 2...

*If* Bob and Barb really have a 12% marginal tax rate on their RMDs, *then* the marginal tax rate on their QD/LTCG income must be 0%. (Not the 15% shown in the spreadsheet on Page 1 of this thread.)

Their Roth conversion will blow up on the spreadsheet rather more spectacularly than currently shown and never recover.

(However, in real life (as opposed to the spreadsheet), Bob-and-Barb would likely find it difficult to maintain 12% marginal for ordinary income + 0% marginal for QD/LTCG for the long term. Depending where in the nominal 12% bracket their SS hump ended, it is highly likely that, sooner or later, their increasing RMDs + increasing size of taxable account would push them back into 27% marginal for ordinary income + 15% marginal for QD/LTCG.)
cas
Posts: 2245
Joined: Wed Apr 26, 2017 8:41 am

Re: Why Roth conversions always pay off—if you can hold on long enough

Post by cas »

marcopolo wrote: Mon Oct 11, 2021 6:40 pm
McQ wrote: Mon Oct 11, 2021 4:04 pm

But when decades of embedded capital gains are liquidated, along with the remaining TDA balance, there must be NIIT.
You started with the premise that RMDs don't get spent and are passed on to heirs. You know, because everyone has 6 figure pensions and near max x2 Soc. Sec.. Now you have forced the retiree to liquidate all their RMDs, in some cases all at once, paying NIIT!

If I didn't know better, I would say you are putting your finger on the scale a bit :beer

What happens when we keep that same scenario and the accumulated gains get step-up basis to the heirs.

These sorts of spreadsheets are kind of stuck between a rock and a hard place.

Comparing the non-tax-adjusted value of TDA + taxable (with their embedded tax obligation) to the Roth isn't comparing apples to oranges. (It is kind of like comparing two people with houses worth $300,000 ... one with a $100,000 mortgage outstanding and one fully paid off ... and saying they have the same net worth.)

So, in order to get back to an apples to apples comparison, the spreadsheet takes the (simplified) approach, for each line/year, of pretending that the original owner ended up actually needing to spend those assets that he didn't think he would need during his lifetime ... forcing complete liquidation. What spendable money would be left in each scenario (conversion vs no conversion)?

But with that approach, you immediately run into the problem that "pretending" to liquidate everything likely blows the pretend AGI sky high and likely runs through multiple marginal tax rates.

So a common simplification/work-around seem to be to use the given marginal tax rates to make the tax adjustment, but wave one's hands and say something like "This isn't perfect, but it is a reasonable way to simplify the spreadsheet. If someone really needed this money in real life, they likely wouldn't really need it all in one year and would probably be able to come up with some way to stay near the usual marginal tax rates by spreading the liquidation over 2 or more years."

McQ seems to be using this simplified approach with the tax-adjustment of TDA.

So I'm not sure why he is now bothered with the idea that liquidating the entire taxable account would blow AGI sky high, incurring NIIT (not to mention IRMAA, etc. etc).

It isn't perfect, but the same simplification used with tax-adjusting the TDA can be used with tax adjusting the taxable account: just use the given marginal rate on QD/LTCG, wave your hands, and say "This isn't perfect, but it is a reasonable way to simplify the spreadsheet. If someone really needed this money in real life, they likely wouldn't really need it all in one year and would probably be able to come up with some way to stay near the usual marginal tax rates by spreading the liquidation over 2 or more years."
marcopolo
Posts: 8411
Joined: Sat Dec 03, 2016 9:22 am

Re: Why Roth conversions always pay off—if you can hold on long enough

Post by marcopolo »

cas wrote: Tue Oct 12, 2021 12:01 pm
marcopolo wrote: Mon Oct 11, 2021 6:40 pm
McQ wrote: Mon Oct 11, 2021 4:04 pm

But when decades of embedded capital gains are liquidated, along with the remaining TDA balance, there must be NIIT.
You started with the premise that RMDs don't get spent and are passed on to heirs. You know, because everyone has 6 figure pensions and near max x2 Soc. Sec.. Now you have forced the retiree to liquidate all their RMDs, in some cases all at once, paying NIIT!

If I didn't know better, I would say you are putting your finger on the scale a bit :beer

What happens when we keep that same scenario and the accumulated gains get step-up basis to the heirs.

These sorts of spreadsheets are kind of stuck between a rock and a hard place.

Comparing the non-tax-adjusted value of TDA + taxable (with their embedded tax obligation) to the Roth isn't comparing apples to oranges. (It is kind of like comparing two people with houses worth $300,000 ... one with a $100,000 mortgage outstanding and one fully paid off ... and saying they have the same net worth.)

So, in order to get back to an apples to apples comparison, the spreadsheet takes the (simplified) approach, for each line/year, of pretending that the original owner ended up actually needing to spend those assets that he didn't think he would need during his lifetime ... forcing complete liquidation. What spendable money would be left in each scenario (conversion vs no conversion)?

But with that approach, you immediately run into the problem that "pretending" to liquidate everything likely blows the pretend AGI sky high and likely runs through multiple marginal tax rates.

So a common simplification/work-around seem to be to use the given marginal tax rates to make the tax adjustment, but wave one's hands and say something like "This isn't perfect, but it is a reasonable way to simplify the spreadsheet. If someone really needed this money in real life, they likely wouldn't really need it all in one year and would probably be able to come up with some way to stay near the usual marginal tax rates by spreading the liquidation over 2 or more years."

McQ seems to be using this simplified approach with the tax-adjustment of TDA.

So I'm not sure why he is now bothered with the idea that liquidating the entire taxable account would blow AGI sky high, incurring NIIT (not to mention IRMAA, etc. etc).

It isn't perfect, but the same simplification used with tax-adjusting the TDA can be used with tax adjusting the taxable account: just use the given marginal rate on QD/LTCG, wave your hands, and say "This isn't perfect, but it is a reasonable way to simplify the spreadsheet. If someone really needed this money in real life, they likely wouldn't really need it all in one year and would probably be able to come up with some way to stay near the usual marginal tax rates by spreading the liquidation over 2 or more years."
It seems to me the most common scenarios for retirees are either

1) RMDs are spent each year to fund living expenses (probably true among the vast majority of the general public)

2) RMDs are largely saved and passed on to heirs (Evidently quite common among Bogleheads)

Case (1) has not been treated at all in these threads.
The analysis has mostly focused on Case (2). Now, we seemed to have switched to a 3rd scenario where the wealthy retiree who has been tax savy to date, for some inexplicable reason, suddenly decides to liquidate their entire portfolio all at once. I guess there are probably scenarios where this happens, but it seems mostly only useful as an academic excersize. This will clearly benefit the Roth Conversion scenario.

If assuming RMDs are accumulated (where Roth Conversion pays off due to tax drag on taxable account), then it seems to me the fair comparison would be assume that accumulation continues until the no-conversion scenario has a chance to "catch up" through step-up basis.

If wanting to look at terminal value, one could do that with liquidation the day after death.
Once in a while you get shown the light, in the strangest of places if you look at it right.
User avatar
FiveK
Posts: 15693
Joined: Sun Mar 16, 2014 2:43 pm

Re: Why Roth conversions always pay off—if you can hold on long enough

Post by FiveK »

marcopolo wrote: Tue Oct 12, 2021 1:55 pm It seems to me the most common scenarios for retirees are either

1) RMDs are spent each year to fund living expenses (probably true among the vast majority of the general public)

2) RMDs are largely saved and passed on to heirs (Evidently quite common among Bogleheads)

Case (1) has not been treated at all in these threads.
Case (1) is already covered by the "usual" Traditional versus Roth analysis.
marcopolo
Posts: 8411
Joined: Sat Dec 03, 2016 9:22 am

Re: Why Roth conversions always pay off—if you can hold on long enough

Post by marcopolo »

FiveK wrote: Tue Oct 12, 2021 2:42 pm
marcopolo wrote: Tue Oct 12, 2021 1:55 pm It seems to me the most common scenarios for retirees are either

1) RMDs are spent each year to fund living expenses (probably true among the vast majority of the general public)

2) RMDs are largely saved and passed on to heirs (Evidently quite common among Bogleheads)

Case (1) has not been treated at all in these threads.
Case (1) is already covered by the "usual" Traditional versus Roth analysis.
Agreed. I was just referring to the analysis Prof McQ is doing.
But, the point is taken that he is probably not looking at that since it has been well covered previously.
Once in a while you get shown the light, in the strangest of places if you look at it right.
cas
Posts: 2245
Joined: Wed Apr 26, 2017 8:41 am

Re: Why Roth conversions always pay off—if you can hold on long enough

Post by cas »

marcopolo wrote: Tue Oct 12, 2021 1:55 pm Case (1) has not been treated at all in these threads.
The analysis has mostly focused on Case (2). Now, we seemed to have switched to a 3rd scenario where the wealthy retiree who has been tax savy to date, for some inexplicable reason, suddenly decides to liquidate their entire portfolio all at once. I guess there are probably scenarios where this happens, but it seems mostly only useful as an academic excersize. This will clearly benefit the Roth Conversion scenario.
That is why the similar MDM spreadsheet (further along in development) has the 2 major sections:
  • The first section follows what is happening while the original owner is still alive, giving a hazy view of the horserace between the conversion and no-conversion scenarios. McQ's current spreadsheet is currently limited to just this section.
  • The second "after inheritance" section shows how things work out for the beneficiary. This is where step-up in basis can be assumed, plus tests can be run on various assumptions about beneficiary marginal tax rates and choices about how to handle the max 10 year payout.
I find the first "original owner" section useful because I can run my eye down the BETRs (break even tax rates) and get a hazy view of how resilient the Roth conversion is or is not to "what if" situations like these:
  • What if I ended up moving to a lower tax state? Would a Roth conversion be a disaster or is there some resiliency to that sort of thing?
  • So what if a change in state/federal tax law or an economic crisis produced an unexpected lower tax rate (say X% instead of Y%) for me? Would a Roth conversion be a disaster or is there some resiliency to that sort of thing?
  • What if I ran through all my primary funds for long term care and unexpectedly needed to tap these Roth conversion funds I expected never to touch? To about what age range would I need to survive in order for the Roth conversion to have not been a disaster?
  • What if I unexpectedly outlive my primary (human) beneficiary, so the Roth ends up going to my secondary beneficiary, who is a tax-exempt charity? To about what age would I need to survive for the Roth conversion to still work out for the charity? (This one would then need more refinement using the second "beneficiary" section of the spreadsheet, but running one's eye down the BETRs gives one a good hint on what death-age to start doing the refining.)
  • How do changes in investment returns and asset allocation affect all the above?
The second "beneficiary" section is useful for "what if"s like these:
  • So how does a Roth conversion work out for my beneficiary once step-up is considered?
  • What if life happens to my beneficiary and s(he) ends up with a lower marginal tax rate than expected? How resilient is the Roth conversion to that type of scenario?
  • What happens with various choices my beneficiary might make about how to disburse the inherited funds over the 10 years?
Obviously my interests currently lie in issues of conversion-vs-no-conversion resiliency to possible shocks.

McQ's current interests seem to lie more in some referenced promises somewhere (financial media I don't read?) that Roth conversions produce "acres of diamonds."
jhawktx
Posts: 441
Joined: Thu May 19, 2011 8:15 am

Re: Why Roth conversions always pay off—if you can hold on long enough

Post by jhawktx »

FiveK wrote: Tue Oct 12, 2021 2:42 pm
marcopolo wrote: Tue Oct 12, 2021 1:55 pm It seems to me the most common scenarios for retirees are either

1) RMDs are spent each year to fund living expenses (probably true among the vast majority of the general public)

2) RMDs are largely saved and passed on to heirs (Evidently quite common among Bogleheads)

Case (1) has not been treated at all in these threads.
Case (1) is already covered by the "usual" Traditional versus Roth analysis.
I agree that the vast majority of people use RMDs to fund living expenses. As such, that same vast majority would do well to ignore this thread entirely and instead use the mentioned wiki rule of thumb. For that vast majority, reading a thread title that says "Roth conversions always pay off" borders on financial malfeasance...IMHO.
Topic Author
McQ
Posts: 1414
Joined: Fri Jun 18, 2021 12:21 am
Location: California

Re: Why Roth conversions always pay off—if you can hold on long enough

Post by McQ »

Thank you to all: many insightful comments today (Internet outage most of my work day, so I only just had a chance to see). I am going to concentrate on replies to cas today. In a day or two a post will address marcopolo's request to "show me the step up."

But first a brief reply to LilyFleur: when I alter the return assumptions in my spreadsheet (i.e., reducing 10% to 8%, or 5% real, close to the bottom of most multi-decade stock returns in the US), the dollar amounts go down on constant / favorable tax rate changes, but the pattern doesn't change. In unfavorable cases, the breakeven is pushed out a bit.

If you have a dollar threshold for do / don't do the conversion(is the game worth the candle?), then the return assumption is crucial, especially in out years after compounding. But that's not the metric I choose to use.

Likewise, if the conversion is total, then the return can't really be the all-stock return, because the AA will be balanced. But I assume throughout that the conversion occurs on the margin, so that this portion of the total portfolio can be all in stocks. If Roth conversion payoffs appear de minimis even with a 10% return, that's important to know.

That's why all the cases thus far assume an all-stock portfolio with a return of 10% nominal, 7% real. For just this small portion that is on-camera.
You can take the academic out of the classroom by retirement, but you can't ever take the classroom out of his tone, style, and manner of approach.
Topic Author
McQ
Posts: 1414
Joined: Fri Jun 18, 2021 12:21 am
Location: California

Re: Why Roth conversions always pay off—if you can hold on long enough

Post by McQ »

cas wrote: Tue Oct 12, 2021 5:58 am
McQ wrote: Mon Oct 11, 2021 4:04 pm The logical next question is what happens when tax drag is reduced in this way and future rates drop. I’ll look at that next time. The focus will be the dread Bob-and-Barb: convert at 22%, to avoid RMDs taxed at 12%.
Careful. The Bob-and-Barb scenario has a significant flaw (assuming tax law since at least Bush 2). This flaw hasn't particularly mattered to the points you have been making thus far using that particular scenario, so I haven't mentioned it.

But if you want to use that scenario to make any points about tax drag on the taxable account, it will need to be addressed.

Here are specifics of the Bob-and-Barb scenario in your spreadsheet on Page 1 of this thread:

marginal tax rate on conversion (pre-RMD): 22%
marginal tax rate on ordinary income post-RMD: 12%
marginal tax rate on QD/LTCG post-RMD: 15%

Under tax law since at least Bush 2, this scenario is not possible***.
Thank you cas, you are correct. The Bob and Barb scenario (22% --> 12% ordinary, but still 15% on the capital gains and dividends) cannot occur under current tax law. Let me explain why I will continue to use it (with the occasional nod to your correction).

It's there to make a point about how bad it can get if the future tax rate drops 10%. It could equally well be made by using a couple who converted at 32% and end up paying only 22%. But, and this was one of several important takeaways from the first thread on the SSRN paper: I have been beaten out of using high income couples as my examples. BH tend to decry such examples as unrealistic and unrepresentative. Marcopolo did it just upthread; dodecahedron did it on another thread. Conversion in the 22% bracket gives me a wider reach.
You can take the academic out of the classroom by retirement, but you can't ever take the classroom out of his tone, style, and manner of approach.
Topic Author
McQ
Posts: 1414
Joined: Fri Jun 18, 2021 12:21 am
Location: California

Re: Why Roth conversions always pay off—if you can hold on long enough

Post by McQ »

cas wrote: Tue Oct 12, 2021 12:01 pm
marcopolo wrote: Mon Oct 11, 2021 6:40 pm
McQ wrote: Mon Oct 11, 2021 4:04 pm

But when decades of embedded capital gains are liquidated, along with the remaining TDA balance, there must be NIIT.
You started with the premise that RMDs don't get spent and are passed on to heirs. You know, because everyone has 6 figure pensions and near max x2 Soc. Sec.. Now you have forced the retiree to liquidate all their RMDs, in some cases all at once, paying NIIT!

If I didn't know better, I would say you are putting your finger on the scale a bit :beer

What happens when we keep that same scenario and the accumulated gains get step-up basis to the heirs.

These sorts of spreadsheets are kind of stuck between a rock and a hard place.

Comparing the non-tax-adjusted value of TDA + taxable (with their embedded tax obligation) to the Roth isn't comparing apples to oranges. (It is kind of like comparing two people with houses worth $300,000 ... one with a $100,000 mortgage outstanding and one fully paid off ... and saying they have the same net worth.)

So, in order to get back to an apples to apples comparison, the spreadsheet takes the (simplified) approach, for each line/year, of pretending that the original owner ended up actually needing to spend those assets that he didn't think he would need during his lifetime ... forcing complete liquidation. What spendable money would be left in each scenario (conversion vs no conversion)?

But with that approach, you immediately run into the problem that "pretending" to liquidate everything likely blows the pretend AGI sky high and likely runs through multiple marginal tax rates.

So a common simplification/work-around seem to be to use the given marginal tax rates to make the tax adjustment, but wave one's hands and say something like "This isn't perfect, but it is a reasonable way to simplify the spreadsheet. If someone really needed this money in real life, they likely wouldn't really need it all in one year and would probably be able to come up with some way to stay near the usual marginal tax rates by spreading the liquidation over 2 or more years."

McQ seems to be using this simplified approach with the tax-adjustment of TDA.

So I'm not sure why he is now bothered with the idea that liquidating the entire taxable account would blow AGI sky high, incurring NIIT (not to mention IRMAA, etc. etc).

It isn't perfect, but the same simplification used with tax-adjusting the TDA can be used with tax adjusting the taxable account: just use the given marginal rate on QD/LTCG, wave your hands, and say "This isn't perfect, but it is a reasonable way to simplify the spreadsheet. If someone really needed this money in real life, they likely wouldn't really need it all in one year and would probably be able to come up with some way to stay near the usual marginal tax rates by spreading the liquidation over 2 or more years."
ahh, but here I want to go back to your very helpful metaphor of on-camera vs off-camera. The spreadsheets in this thread look at a tiny portion of the total retirement portfolio; that is all that is "on-camera." Assuming that the initial RMD of $3650 will all be taxed at 22% is not a heavy lift; and likewise, that subsequent RMDs, which will grow to $10,000 and then $15,000, will also be taxed within a bracket.

I like your reminder that in analyses like these, one mentally liquidates everything each year to get a proper apples-to-apples comparison. Not that one would, but simply to avoid comparing pretax to post tax dollars. But liquidating the $100,000 TDA is also not a problem, if we place the taxpayer just at the bottom of the 22% bracket. It does not blow AGI sky high.

But marcopolo's quite reasonable suggestion to keep capital gains unrealized blows up that comfortable certainty. With tens and then hundreds of thousands of dollars on realized gains, the rate on liquidation is going to move into the NIIT range.

And in any case, I thought it useful to remind everyone that tax deferred--capital gains or ordinary--is tax unknown, and to calibrate the effect of even the small bump of having to pay NIIT.
You can take the academic out of the classroom by retirement, but you can't ever take the classroom out of his tone, style, and manner of approach.
Topic Author
McQ
Posts: 1414
Joined: Fri Jun 18, 2021 12:21 am
Location: California

Re: Why Roth conversions always pay off—if you can hold on long enough

Post by McQ »

cas wrote: Tue Oct 12, 2021 3:24 pm ...

Obviously my interests currently lie in issues of conversion-vs-no-conversion resiliency to possible shocks.

McQ's current interests seem to lie more in some referenced promises somewhere (financial media I don't read?) that Roth conversions produce "acres of diamonds."
Actually, I am trying to thread a finer needle than that. It's the F. Scott Fitzgerald needle. I am trying to hold two opposed ideas in mind while continuing to provide insight into how Roth conversions play out.
1. That tax drag (if present, I take your point about 0% rates) will heal all conversion wounds in time.
2. But that however resilient, conversions are typically not that lucrative, certainly not as much as some people think ("acres of diamonds," which I believe I have observed here at BH no less than in the media).
It's a tough needle to thread. That's where BH has been so valuable in helping me refine my arguments. Which is also why I continue to work through these (soon to be dozens of) posts rather than proceeding directly to a revision of the paper. Too much still to learn, and too fine a needle to be all that confident as yet.
You can take the academic out of the classroom by retirement, but you can't ever take the classroom out of his tone, style, and manner of approach.
Topic Author
McQ
Posts: 1414
Joined: Fri Jun 18, 2021 12:21 am
Location: California

Re: Why Roth conversions always pay off—if you can hold on long enough

Post by McQ »

jhawktx wrote: Tue Oct 12, 2021 4:48 pm
FiveK wrote: Tue Oct 12, 2021 2:42 pm
marcopolo wrote: Tue Oct 12, 2021 1:55 pm It seems to me the most common scenarios for retirees are either

1) RMDs are spent each year to fund living expenses (probably true among the vast majority of the general public)

2) RMDs are largely saved and passed on to heirs (Evidently quite common among Bogleheads)

Case (1) has not been treated at all in these threads.
Case (1) is already covered by the "usual" Traditional versus Roth analysis.
I agree that the vast majority of people use RMDs to fund living expenses. As such, that same vast majority would do well to ignore this thread entirely and instead use the mentioned wiki rule of thumb. For that vast majority, reading a thread title that says "Roth conversions always pay off" borders on financial malfeasance...IMHO.
One liability of multipage threads is that nobody but the OP and one or two others reads every post. I'm guessing that you missed cas' insightful distinction upthread between the portion of the converter's assets that are "on camera,' i.e., visible in the spreadsheet, and the (much larger) portion that are off-camera.

Of course, most of us spend most of our RMDs. That's certainly what I plan to do; that's why I saved for decades in my TDAs, boosted by the government subsidy: so that I would have more money to spend and could live that much better in my retirement. And that is what I assume is occurring off-camera in every spreadsheet. (How else did the retiree get up into the 22% bracket?)

The spreadsheets simply ask, If you made what for many would be a large conversion of $100,000, you could reduce your taxable RMDs in the first year by $3650. If you made three large conversions you could reduce your taxable RMDs by $10,950. Presumably you are contemplating these conversions because just this much of your RMDs represents money you don't need right now, and you don't like being forced to take the RMD and pay tax on it. The spreadsheets look at the wealth outcomes of paying tax now with a conversion versus having to pay tax year after year because there was no conversion. Of just that one piece of total wealth.

If instead the plan is to spend every year from the converted funds, same as the RMDs would have been spent, but with greater after-tax spending power because they are Roth funds, then the conversion is not left undisturbed, and, I hope to show, may not work out. Those spreadsheets will be forthcoming in a week or two.
You can take the academic out of the classroom by retirement, but you can't ever take the classroom out of his tone, style, and manner of approach.
marcopolo
Posts: 8411
Joined: Sat Dec 03, 2016 9:22 am

Re: Why Roth conversions always pay off—if you can hold on long enough

Post by marcopolo »

McQ wrote: Tue Oct 12, 2021 5:49 pm
cas wrote: Tue Oct 12, 2021 3:24 pm ...

Obviously my interests currently lie in issues of conversion-vs-no-conversion resiliency to possible shocks.

McQ's current interests seem to lie more in some referenced promises somewhere (financial media I don't read?) that Roth conversions produce "acres of diamonds."
Actually, I am trying to thread a finer needle than that. It's the F. Scott Fitzgerald needle. I am trying to hold two opposed ideas in mind while continuing to provide insight into how Roth conversions play out.
1. That tax drag (if present, I take your point about 0% rates) will heal all conversion wounds in time.
2. But that however resilient, conversions are typically not that lucrative, certainly not as much as some people think ("acres of diamonds," which I believe I have observed here at BH no less than in the media).
It's a tough needle to thread. That's where BH has been so valuable in helping me refine my arguments. Which is also why I continue to work through these (soon to be dozens of) posts rather than proceeding directly to a revision of the paper. Too much still to learn, and too fine a needle to be all that confident as yet.
I am pretty sure (1) is not true.
I agree with (2), both the media and some on this forum oversell the benefits of Roth Conversions.
There are many scenarios where they are quite beneficial, but also some where they do more harm than good. Those seem to often be glossed over.

You are correct that i raised questions about how realistic some of the scenarios are. But, my bigger concern is the inconsistency in the assumptions.
They have so far been geared to make Roth Conversions look more favorable than they are by making assumption for the Conversion case that favor it (high returns, saving RMDs), and making assumptions for the non-conversion case that handicap it (realize gains, no step-up, TDA holding high return assets).

I know you are working towards incorporating some of these issues. But, as you have astutely pointed out, most people don't have that kind of attention span. By starting your thread with a title like "Roth Conversions always payoff" and using these simple assumptions as a starting point to support that assertion, i fear you may be perpetuating the myth you seem to be wanting to address in item (2) above.

I look forward to seeing this analysis take more of those nuances into account, but did notice a few posts already where people were making takeaway statements for this work in progress.
Once in a while you get shown the light, in the strangest of places if you look at it right.
User avatar
FiveK
Posts: 15693
Joined: Sun Mar 16, 2014 2:43 pm

Re: Why Roth conversions always pay off—if you can hold on long enough

Post by FiveK »

Timely advice on how to make this decision: Comparing Unknowns - Dilbert
curmudgeon
Posts: 2630
Joined: Thu Jun 20, 2013 11:00 pm

Re: Why Roth conversions always pay off—if you can hold on long enough

Post by curmudgeon »

McQ wrote: Tue Oct 12, 2021 6:06 pm
Of course, most of us spend most of our RMDs. That's certainly what I plan to do; that's why I saved for decades in my TDAs, boosted by the government subsidy: so that I would have more money to spend and could live that much better in my retirement. And that is what I assume is occurring off-camera in every spreadsheet. (How else did the retiree get up into the 22% bracket?)

The spreadsheets simply ask, If you made what for many would be a large conversion of $100,000, you could reduce your taxable RMDs in the first year by $3650. If you made three large conversions you could reduce your taxable RMDs by $10,950. Presumably you are contemplating these conversions because just this much of your RMDs represents money you don't need right now, and you don't like being forced to take the RMD and pay tax on it. The spreadsheets look at the wealth outcomes of paying tax now with a conversion versus having to pay tax year after year because there was no conversion. Of just that one piece of total wealth.

If instead the plan is to spend every year from the converted funds, same as the RMDs would have been spent, but with greater after-tax spending power because they are Roth funds, then the conversion is not left undisturbed, and, I hope to show, may not work out. Those spreadsheets will be forthcoming in a week or two.
For many of us, the RMD is just a number, devised by congress to ensure that those deferred taxes start being paid. Our spending is determined by our own calculations of our resources and desires, and may have limited relationship to the RMD amount. I can spend from our TDA, taxable investments, Roth, or maybe even tap real estate equity. I definitely don't plan to set my year-by-year budget based on the vagaries of the RMD percentages. That said, the common "4% withdrawal rate" guideline common around here is somewhat similar to the RMD numbers.

From my perspective, the "tax drag" on RMDs retained in a taxable account is just not all that significant to the moderately wealthy retiree. If I'm really that worried about it, I'd just dump the RMDs into Berkshire Hathaway. I have little interest in converting at 22% to avoid future tax on RMDs at 22% (I might do some, just to have a reserve of resources that I could draw at once without tax impact). Anything I can convert at 12%, however, is automatic.

Where it gets somewhat interesting is the "SS torpedo". If your SS benefit and other tax situation is such that you will get hit by the torpedo, and, there is potential for avoiding it by conversion, then the decision gets more complex, but the possible tax savings may be interesting. One thing that can be useful is to put some bounds on the potential, to get a feel for the most that could be possible. Lets say a couple has SS at age 70 of $80K, and has a $2M TDA. Suppose they converted $325K per year to Roth for six years in early retirement, paying the $390K in taxes out of their taxable savings (and living off those savings as well). They could then take an RMD equivalent from the Roth, combined with SS, and pay essentially no income tax for the rest of retirement. In reality, they still might have some other income spinning off of their taxable investments, but they still have to potential to avoid the taxation of the 85% of their $80K of SS.

If I look at it this way, the maximum benefit from avoiding the SS torpedo, if everything lines up right, is about $15K per year, and might optimistically last from age 70 to age 90, so about $300K overall. It's significant, but in the scope of such large accounts over the course of 20 years, it's probably of less effect than many other things that might happen. On the other hand, the cost is not necessarily all that large; paying a tax on a chunk of the conversions at 24% instead of 22%, six years of IRMAA (but possibly avoiding more IRMAA later).
Topic Author
McQ
Posts: 1414
Joined: Fri Jun 18, 2021 12:21 am
Location: California

Re: Why Roth conversions always pay off—if you can hold on long enough

Post by McQ »

yog wrote: Sat Oct 09, 2021 1:49 pm
Here's another similar more recent Kitces article:
Navigating Income Harvesting Strategies: Harvesting (0%) Capital Gains Vs Partial Roth Conversions
Yog, I’ve been remiss in not thanking you for the updated kitces link on tax humps and the earlier link to an article by my colleagues Dan Ostrov and Sanjiv Das—two of the best financial mathematicians I know.
You can take the academic out of the classroom by retirement, but you can't ever take the classroom out of his tone, style, and manner of approach.
Topic Author
McQ
Posts: 1414
Joined: Fri Jun 18, 2021 12:21 am
Location: California

Re: Why Roth conversions always pay off—if you can hold on long enough

Post by McQ »

marcopolo wrote: Tue Oct 12, 2021 6:47 pm
McQ wrote: Tue Oct 12, 2021 5:49 pm
cas wrote: Tue Oct 12, 2021 3:24 pm ...

Obviously my interests currently lie in issues of conversion-vs-no-conversion resiliency to possible shocks.

McQ's current interests seem to lie more in some referenced promises somewhere (financial media I don't read?) that Roth conversions produce "acres of diamonds."
Actually, I am trying to thread a finer needle than that. It's the F. Scott Fitzgerald needle. I am trying to hold two opposed ideas in mind while continuing to provide insight into how Roth conversions play out.
1. That tax drag (if present, I take your point about 0% rates) will heal all conversion wounds in time.
2. But that however resilient, conversions are typically not that lucrative, certainly not as much as some people think ("acres of diamonds," which I believe I have observed here at BH no less than in the media).
It's a tough needle to thread. That's where BH has been so valuable in helping me refine my arguments. Which is also why I continue to work through these (soon to be dozens of) posts rather than proceeding directly to a revision of the paper. Too much still to learn, and too fine a needle to be all that confident as yet.
I am pretty sure (1) is not true.
I agree with (2), both the media and some on this forum oversell the benefits of Roth Conversions.
There are many scenarios where they are quite beneficial, but also some where they do more harm than good. Those seem to often be glossed over.

You are correct that i raised questions about how realistic some of the scenarios are. But, my bigger concern is the inconsistency in the assumptions.
They have so far been geared to make Roth Conversions look more favorable than they are by making assumption for the Conversion case that favor it (high returns, saving RMDs), and making assumptions for the non-conversion case that handicap it (realize gains, no step-up, TDA holding high return assets).

I know you are working towards incorporating some of these issues. But, as you have astutely pointed out, most people don't have that kind of attention span. By starting your thread with a title like "Roth Conversions always payoff" and using these simple assumptions as a starting point to support that assertion, i fear you may be perpetuating the myth you seem to be wanting to address in item (2) above.

I look forward to seeing this analysis take more of those nuances into account, but did notice a few posts already where people were making takeaway statements for this work in progress.
That’s useful feedback for me marcopolo; and let me take this chance to thank you for your ongoing participation in the thread (I’m still impressed that you took the time to reconstruct the ss from the image way upthread).

I will at some point edit the opening post to focus more tightly on the “two opposed notions.” In the meantime, I choose to expect the best from my readers; I'm writing for the patient ones who have grappled with some of these conversion issues for a long time.
You can take the academic out of the classroom by retirement, but you can't ever take the classroom out of his tone, style, and manner of approach.
Topic Author
McQ
Posts: 1414
Joined: Fri Jun 18, 2021 12:21 am
Location: California

Re: Why Roth conversions always pay off—if you can hold on long enough

Post by McQ »

curmudgeon wrote: Wed Oct 13, 2021 1:42 am
McQ wrote: Tue Oct 12, 2021 6:06 pm
Of course, most of us spend most of our RMDs. That's certainly what I plan to do; that's why I saved for decades in my TDAs, boosted by the government subsidy: so that I would have more money to spend and could live that much better in my retirement. And that is what I assume is occurring off-camera in every spreadsheet. (How else did the retiree get up into the 22% bracket?)

The spreadsheets simply ask, If you made what for many would be a large conversion of $100,000, you could reduce your taxable RMDs in the first year by $3650. If you made three large conversions you could reduce your taxable RMDs by $10,950. Presumably you are contemplating these conversions because just this much of your RMDs represents money you don't need right now, and you don't like being forced to take the RMD and pay tax on it. The spreadsheets look at the wealth outcomes of paying tax now with a conversion versus having to pay tax year after year because there was no conversion. Of just that one piece of total wealth.

If instead the plan is to spend every year from the converted funds, same as the RMDs would have been spent, but with greater after-tax spending power because they are Roth funds, then the conversion is not left undisturbed, and, I hope to show, may not work out. Those spreadsheets will be forthcoming in a week or two.
For many of us, the RMD is just a number, devised by congress to ensure that those deferred taxes start being paid. Our spending is determined by our own calculations of our resources and desires, and may have limited relationship to the RMD amount. I can spend from our TDA, taxable investments, Roth, or maybe even tap real estate equity. I definitely don't plan to set my year-by-year budget based on the vagaries of the RMD percentages. That said, the common "4% withdrawal rate" guideline common around here is somewhat similar to the RMD numbers.

From my perspective, the "tax drag" on RMDs retained in a taxable account is just not all that significant to the moderately wealthy retiree. If I'm really that worried about it, I'd just dump the RMDs into Berkshire Hathaway. I have little interest in converting at 22% to avoid future tax on RMDs at 22% (I might do some, just to have a reserve of resources that I could draw at once without tax impact). Anything I can convert at 12%, however, is automatic.

Where it gets somewhat interesting is the "SS torpedo". If your SS benefit and other tax situation is such that you will get hit by the torpedo, and, there is potential for avoiding it by conversion, then the decision gets more complex, but the possible tax savings may be interesting. One thing that can be useful is to put some bounds on the potential, to get a feel for the most that could be possible. Lets say a couple has SS at age 70 of $80K, and has a $2M TDA. Suppose they converted $325K per year to Roth for six years in early retirement, paying the $390K in taxes out of their taxable savings (and living off those savings as well). They could then take an RMD equivalent from the Roth, combined with SS, and pay essentially no income tax for the rest of retirement. In reality, they still might have some other income spinning off of their taxable investments, but they still have to potential to avoid the taxation of the 85% of their $80K of SS.

If I look at it this way, the maximum benefit from avoiding the SS torpedo, if everything lines up right, is about $15K per year, and might optimistically last from age 70 to age 90, so about $300K overall. It's significant, but in the scope of such large accounts over the course of 20 years, it's probably of less effect than many other things that might happen. On the other hand, the cost is not necessarily all that large; paying a tax on a chunk of the conversions at 24% instead of 22%, six years of IRMAA (but possibly avoiding more IRMAA later).
Curmudgeon, I like the judiciousness of the approach you outlined. I am fine with a retiree who looks at the metrics for conversion at a constant tax rate, and … shrugs. The more so, if the return assumption is ratcheted down, as iceport and LilyFleur.

And I do think that avoidance of the social security tax torpedo may be where the very highest conversion payoffs lie.

Now whether it makes sense to unnecessarily incur n years of IRMAA at the outset of retirement, where n is a single digit number, to avoid nn years of IRMAA later, where nn is low double-digit number, is an analysis I hope to do but have not done yet. I am not positive that the payoff is sufficient in cases where n is 6 and nn is, say, 15; probably not, if the pay off from a 22% --> 22% conversion, over those same 21 years, was unexciting to you. And given that IRMAA brackets have already been changed once, I doubt the payoff from prepaying IRMAA based on assumptions about future bracket boundaries would satisfy cas’ criterion of resiliency.

When one is only a little bit into an IRMAA bracket. it seems to me that the thing to do is a QCD--the reduced tax and IRMAA could be 50% or more of the donation.
You can take the academic out of the classroom by retirement, but you can't ever take the classroom out of his tone, style, and manner of approach.
Topic Author
McQ
Posts: 1414
Joined: Fri Jun 18, 2021 12:21 am
Location: California

Re: Why Roth conversions always pay off—if you can hold on long enough

Post by McQ »

Poor Bob and Barb. You’ll recall from upthread that they were a little too enamored of Roth conversions and not well enough informed about how tax brackets adjust for inflation. As a consequence, they converted at 22% to avert RMDs that were only going to be taxed at 12%. Sad.

Fortunately, tax drag bailed them out in the end, allowing them to breakeven on the conversion at age 96. But that was under the mark-to-market formulation where for simplicity, all tax on appreciation in the taxable account holding reinvested RMDs was assessed annually. Time to look at it again the marcopolo way, with its more realistic formulation in which only dividends are taxed annually, while capital gains accumulate untaxed until liquidation at the end.

• I note and acknowledge cas’ point that under current tax law, one cannot pay an ordinary income rate of 12% and a qualified dividend rate of 15% (it would be 0%). So this example assumes that Congress passed a special tax rate of 12% on RMDs for, say, singles with income under $400,000 / couples with income under $450,000—just to pull some numbers out of the hat. (The affluent seniors lobby got to Congress.) The example could instead be reworked for a couple converting at 32% to save 22%; but it seemed better to stick to the 22% base. The point is to examine outcomes for a semi-disastrous conversion where the estimated future tax rate was off by 10 percentage points but there was still tax drag on dividends.

Here is the deferred cap gain spreadsheet set up from the last post, now with conversion at 22% and RMDs at 12%.

Image

There’s bad news and good news for Bob and Barb. The bad news: breakeven is pushed out six years, to an age of 102. The good news: they still do break even (eventually), despite the disastrous tax planning error. Time, or rather, tax drag, heals all conversion mistakes.

But wait—what about NIIT? With a breakeven pushed out to a late age, unrealized capital gains will have had a chance to compound for decades. There’s over $360,000 nominal in untaxed gains at the original breakeven age, call it $175,000 real; and almost $850,000 nominal at the new breakeven age, call it $350,000 real.

If the NIIT still exists then, Bob and Barb will have to pay it. Let’s see how applying a rate of 18.8% affects the outcome. Whoops—now breakeven moves back to age 99 (ss not shown). Comforting, in a strange sort of way, to know that even a disaster like converting at 22% to save 12% can work out … in less than thirty years.

“Oh really?” (here I am putting words in marcopolo’s mouth). “Have you forgotten the goal in managing the annual tax liability on the taxable account? The specific hope is to get a step up at death, and never pay tax on those capital gains. In which case, Bob and Barb aren’t ever going to recover from their horrendous conversion mistake.”

Hmmm … is that correct? Let’s see. Here is a spreadsheet version where capital gains are not taxed each year and not taxed at the end. Many rows collapsed and columns hidden. Column Y shows the value of the taxable account with a step up at death.
Image

Again, bad news and good news for Bob and Barb. The bad news: now they don’t break even until an age of 132. That’s cold comfort for a human couple in 2021; few expect life extension technologies to have developed to that point for at least a generation or two. In fact, from the original breakeven age of 96 through about age 117, their conversion outcomes with a step up will actually get worse and worse each year. Someone who did not extend the spreadsheet beyond that point would assume that breakeven can never occur.

Good news: But then things turn around. The annual dividend tax burden climbs above $10,000 by age 111 and over $20,000 by age 120; it is over $60,000 by the time breakeven is reached. That’s tax drag in action: the Roth is tax free, and must eventually catch up and surpass any account that is not tax free—even when “tax’ means a 30 basis point haircut off a return of 10%. Again, tax drag heals all conversion wounds.

Of course, this is a rather arid demonstration, amounting to not much more than a display of the arithmetic of compounding in action. But it does help to clarify the process, especially for greatly delayed breakeven points consequent to no good, terrible, very bad conversion decisions.

In terms of process, up to an age of 100 or so, my spreadsheet set up can be thought of as capturing the returns on a dual account, consisting of the remaining TDA counterfactual and the accumulating taxable account full of after-tax surplus RMDs. But later, at an age of 107 the RMD divisor drops below 4.0 and at 114 it hits 3.0. At an age of 120 the divisor plateaus at 2.0.

No ordinary portfolio, subject to being cut by a third or by a half, every year, can retain a meaningful value for long. That means that after age 110 or so, there really aren’t two accounts on the counterfactual side; there’s just the taxable account holding the reinvested after-tax RMDs. As it is subject to tax drag relative to the Roth, the Roth account must prevail over it in time.

Deferring capital gains weakens conversion outcomes. Never paying tax on capital gains weakens them further. But as long as there is tax drag, the conversion must work out. Eventually.

But hey Bob and Barb—maybe don’t convert at 22% if there is any meaningful likelihood that your RMDs would be taxed at a significantly lower rate? For a $100,000 conversion, remember, there will only be a $3650 swing in RMD payments in the first year if the conversion is done/not done. If subtracting that amount from projected income would put you close to the bottom of the 22% bracket as inflated in the future—well, I would hesitate to do that conversion.
You can take the academic out of the classroom by retirement, but you can't ever take the classroom out of his tone, style, and manner of approach.
yog
Posts: 659
Joined: Wed Jan 15, 2020 11:57 am

Re: Why Roth conversions always pay off—if you can hold on long enough

Post by yog »

McQ wrote: Wed Oct 13, 2021 5:25 pm
yog wrote: Sat Oct 09, 2021 1:49 pm
Here's another similar more recent Kitces article:
Navigating Income Harvesting Strategies: Harvesting (0%) Capital Gains Vs Partial Roth Conversions
Yog, I’ve been remiss in not thanking you for the updated kitces link on tax humps and the earlier link to an article by my colleagues Dan Ostrov and Sanjiv Das—two of the best financial mathematicians I know.

:sharebeer

In case you haven't caught these on other threads, here are two additional key Kitces Roth articles from 2019 & 2021. The first is on the importance of understanding your tax equilibrium in retirement, and Sanjiv's article reminds me of certain aspects of this. The second is about tax alpha, and a few folks in the comments have read your work.
https://www.kitces.com/blog/tax-rate-eq ... nversions/
https://www.kitces.com/blog/tax-diversi ... qus_thread
Topic Author
McQ
Posts: 1414
Joined: Fri Jun 18, 2021 12:21 am
Location: California

Re: Why Roth conversions always pay off—if you can hold on long enough

Post by McQ »

But what if I convert before age 71?

Back to more pragmatic, actionable concerns in this and the next post.

Claim: that converting before age 71 won’t make any difference to the level of payoff from a Roth conversion, after proper adjustment.

Way upthread I made that claim in response to a query from Woodspinner. I reasoned as follows: prior to taking RMDs, the performance of an asset in a TDA account must be exactly the same as performance of that asset in a Roth account—if it is the identical asset, here taken to be stocks returning 10% when inflation is 3%. Both accounts are equally tax deferred prior to RMDs; each account’s value appreciates by exactly the asset’s appreciation.

For example, if, a few years earlier at age 65, one had converted $100,000 at 22%, and paid tax out of the conversion, then there will be $100,000 to start in the counterfactual no-conversion TDA, and $78,000 in the Roth account. Six years later, after compounding at 10%, there will be about $177,000 in the TDA (pretax) and about $138,000 in the Roth. The ratio is the same: 177/138 ~ 100/78. My point: the ratio 100/78 will be maintained at all points until RMDs begin, with that ratio understood to have the general form of [conversion / (conversion – tax)].

Only when RMDs begin can the Roth conversion, which was intended to reduce RMDs and the tax burden they incur, start to pay off.

That was the claim; now here is the spreadsheet. The first iteration looks at a Roth conversion aged 65, six years earlier than the base case. The second iteration will look at a conversion at age 59.51, twelve years early. The ss shows the constant tax rate case 22%-->22%.

Image

To focus discussion, here is a table with a few key values from the ss.

(age 85)

Code: Select all

                 Real Roth   Age 71	 Inflation   Wealth
                 surplus     TDA         divisor     ratio
--------------------------------------------------------                                       
Convert at 71:   $ 7,629     $100,000    1.51      1.0405
Convert at 65:   $11,319     $177,156    1.81      1.0405

The real dollar payoff for early conversion is greater, now up to $11,319. Natch—the age 71 TDA and Roth values are $177,156 and $138,182, not $100,000 and $78,000. Converting early is the close equivalent of converting a larger dollar amount at age 71. But not quite. The new Roth surplus is not 1.77X the old surplus recorded for conversion of $100,000 at age 71.

Why not?? Answer: inflation. Early conversion means an early start to the inflation divisor. In this early scenario, for six years there is no difference between TDA and Roth appreciation, because no RMDs. But the inflation divisor, water running underground, is already over 1.2 by the time the first RMD is taken. So even though we have no catch up for the converted funds in the six years after age 65, we do have inflation exposure. When results are evaluated at age 85, now after 20 years rather than 14, the inflation divisor cuts a little deeper and the real Roth surplus is not quite as much as if we had just converted $177,000 at age 71.

In terms of a formula, the real Roth surplus with early conversion, relative to the surplus for conversion at age 71, looks like this:
New surplus =
Old surplus X
[New TDA $ at 71 / old TDA $ at 71] X
inflation divisor at 71 / inflation divisor (early convert).

Using values from the table and ss, you can confirm that:
$11,319 =
$7,629 X
$177,156 / $100,000 X
1.51 / 1.81
In other words: converting early can’t magically produce an additional or larger Roth payoff, nor does it change the underlying operation that produces a payoff from conversion. This can be seen in the wealth ratio, which doesn’t budge.

On the other hand, you can’t convert $177,000 in one year and stay in a single tax bracket (other than the top one). That’s one proper role for a pre-71 conversion: to convert more within a particular bracket by spreading the conversion over multiple years. Likewise, to get a tax rate as low as 22%, conversions may have to be done before Social Security payments begin; and for that matter, before IRMAA begin. In fact, for IRMAA avoidance, 65 may be too late for the optimal conversion, unless one spouse is still working.

Accordingly, here is the spreadsheet for converting 12 years early.

Image

And here is a summary table:

Code: Select all

                 Real Roth   Age 71	 Inflation   Wealth
                 surplus     TDA         divisor     ratio
--------------------------------------------------------                                       
Convert at 71:   $ 7,629     $100,000    1.51      1.0405
Convert at 59:   $16,794     $313,843    2.16      1.0405
A conversion this early might appear to be the approximate equivalent of converting three times as much at age 71. But the real Roth surplus is not 3X as great as the base case; it is only about 2.2X. Explanation: there are now twelve extra years of inflation. The inflation ratio is about 1.51/2.16, or only 70%. And 3.13 X 0.7 is about 2.2.

Converting early is no magic bullet able to enhance the real dollar wealth surplus generated by a conversion. But it will still be a good idea for many BH to convert before 71, as discussed next.

(Bad) Good reasons to convert early

Expecting a bigger payoff, or a different mechanism underlying the conversion payoff, is the wrong reason to convert early.

Good reasons to convert early include:
1. To get access to a favorably low tax bracket by converting before Social Security, IRMAA, or other impediments arise;
2. To convert larger amounts within a favorable tax bracket than could be done in any single year;
3. To reduce TDA balances by a larger amount and thereby drive down taxable RMDs by a larger amount.

With respect to #3, I commented up thread that it took a big conversion to materially reduce RMDs: in the running example, it took $100,000 to lower the RMD flow by a mere $3650 at age 72. But if you convert at age 59.51, you convert before the TDA records another twelve years of appreciation. That same $100,000 removes a larger proportion of the TDA, reducing now over $10,000 of RMDs at age 72, because, performed earlier, the conversion removed three times the age 72 TDA value (under the assumed rate of return).

The conjunction of all three good reasons will drive many Roth conversions into the early 60s, after one / both salaries have stopped, and likely before SS has begun, and to avoid IRMAA penalties on the conversion. The same conjunction will lead to spreading conversion activity across multiple years in the early 60s, to the extent affordable.

More fatefully: if one or both of you work until age 65 or beyond, or had earlier arranged for other income in the gap years before social security (i.e., 409A distributions, which are difficult to reschedule), then there may never be a good opportunity to convert at low tax rates, because you never will be in a low tax bracket. Ever. In which case, you will have to decide whether a high-rate, with-IRMAA conversion makes sense for you.

One good reason not to convert way in advance

Upthread, FiveK and curmudgeon educated me on the nose cone of the social security tax torpedo—an income range of about $8000, where, with the exact right combination of social security, qualified dividends, and unwanted RMDs, your marginal rate would be 49.95%, or 55.5% post-TCJA, on each additional dollar of RMD. A Roth conversion that moved you down in that range could be the most lucrative ever.

But it’s a narrow range. How certain could you be, at age 59.5, that all the stars would align, so that twelve years later the RMDs reduced by conversion would fall into that narrow window and save you 49.95% / 55.5%? Not very. If IRMAA won’t be a problem, better to wait until your late 60s, when forecasts will have less error.

The social security torpedo more generally covers a wider range; but the high explosive compartment occupies a fairly narrow range at the bottom portion of the 22% bracket, where the marginal tax rate will be 40.7% or 46.25% post-TCJA, if you hit that bullseye. Again, a later conversion is more likely to harvest these juicy results because more certain the torpedo will be launched in the expected direction.

These comments are also a way of acknowledging LilyFleur’s concern about rate of return and its uncertainties. Because RMDs are determined by accumulated wealth, which is an exponential function of rate of return, errors in estimating that rate, which become more likely a dozen years out, can play hob with the forecasted return from a conversion. The later the conversion is made, ceteris paribus, the less the risk that forecasted wealth / RMDs are significantly off.

*ceteris paribus is Latin for “I won’t have to worry about IRMAA or other burdensome increases in my tax rate on later conversions.”

In sum: I would expect most real people to convert before age 71. But for purposes of the abstract demonstrations pursued in this thread, age 71 keeps things simple. It also avoids the complications that come with the Everything-on-camera style of spreadsheet used in the SSRN paper. Respondents in the earlier thread on that paper were wont to ask, On what funds did Rob and Sue live, between the conversion at age 65 and when RMDs commence at 72? That was easy enough to answer (TDA withdrawals, SS, savings) but that whole line of thought distracted from the fundamental question: what determines Roth conversion payoffs? What are the underlying operations?
You can take the academic out of the classroom by retirement, but you can't ever take the classroom out of his tone, style, and manner of approach.
Running Bum
Posts: 356
Joined: Sun Feb 10, 2013 1:00 pm

Re: Why Roth conversions always pay off—if you can hold on long enough

Post by Running Bum »

McQ wrote: Fri Oct 15, 2021 2:14 pm But what if I convert before age 71?

Back to more pragmatic, actionable concerns in this and the next post.

Claim: that converting before age 71 won’t make any difference to the level of payoff from a Roth conversion, after proper adjustment.
Of course if the numbers are otherwise the same, there's no difference (other than inflation) by converting earlier. The reality is that if you defer collecting SS to age 70, and are retired, you may have a opportunity to convert at a better tax rate than you will be paying later.

You hint at this later:
On the other hand, you can’t convert $177,000 in one year and stay in a single tax bracket (other than the top one). That’s one proper role for a pre-71 conversion: to convert more within a particular bracket by spreading the conversion over multiple years. Likewise, to get a tax rate as low as 22%, conversions may have to be done before Social Security payments begin; and for that matter, before IRMAA begin. In fact, for IRMAA avoidance, 65 may be too late for the optimal conversion, unless one spouse is still working.
Some of us can convert at lower than 22% in this time frame. I've been able convert a fair amount at 12%, some at 10%, and even a little at 0% federal tax rate. However, ACA subsidy reduction or loss (until this year) is an issue, probably bigger than IRMAA if one is getting subsidies. And btw, IRMAA concerns start at age 63 since IRMAA is based on your income two years earlier.
In sum: I would expect most real people to convert before age 71. But for purposes of the abstract demonstrations pursued in this thread, age 71 keeps things simple.
Well, that explains why my eyes get so bleary in this thread. IMO it's more useful to look at real life factors and help people determine how much, if any, Roth conversion they should be doing. That's probably been covered in other threads. It certainly has been at early-retirement.org, where I've been more active. There were a couple people over there who have used your research to "prove" that Roth conversions have very little benefit and usually only after age 90 at the so-called breakeven point. I've followed this thread trying to understand that rationale, but given that this is in the abstract, I'll stop trying.

In reality, the breakeven point is nonsense. You can't spend money in a tIRA until you pay the deferred taxes, so it's meaningless to equally value money in a tIRA vs. money in a Roth and try to claim there's a breakeven point.
curmudgeon
Posts: 2630
Joined: Thu Jun 20, 2013 11:00 pm

Re: Why Roth conversions always pay off—if you can hold on long enough

Post by curmudgeon »

McQ wrote: Fri Oct 15, 2021 2:14 pm
Converting early is no magic bullet able to enhance the real dollar wealth surplus generated by a conversion. But it will still be a good idea for many BH to convert before 71, as discussed next.

(Bad) Good reasons to convert early

Expecting a bigger payoff, or a different mechanism underlying the conversion payoff, is the wrong reason to convert early.

Good reasons to convert early include:
1. To get access to a favorably low tax bracket by converting before Social Security, IRMAA, or other impediments arise;
2. To convert larger amounts within a favorable tax bracket than could be done in any single year;
3. To reduce TDA balances by a larger amount and thereby drive down taxable RMDs by a larger amount.

With respect to #3, I commented up thread that it took a big conversion to materially reduce RMDs: in the running example, it took $100,000 to lower the RMD flow by a mere $3650 at age 72. But if you convert at age 59.51, you convert before the TDA records another twelve years of appreciation. That same $100,000 removes a larger proportion of the TDA, reducing now over $10,000 of RMDs at age 72, because, performed earlier, the conversion removed three times the age 72 TDA value (under the assumed rate of return).

The conjunction of all three good reasons will drive many Roth conversions into the early 60s, after one / both salaries have stopped, and likely before SS has begun, and to avoid IRMAA penalties on the conversion. The same conjunction will lead to spreading conversion activity across multiple years in the early 60s, to the extent affordable.

More fatefully: if one or both of you work until age 65 or beyond, or had earlier arranged for other income in the gap years before social security (i.e., 409A distributions, which are difficult to reschedule), then there may never be a good opportunity to convert at low tax rates, because you never will be in a low tax bracket. Ever. In which case, you will have to decide whether a high-rate, with-IRMAA conversion makes sense for you.
I think this summarizes quite well; not everyone will have a lower-income phase of life in which Roth conversions will have particular advantage. On the flip side, there may be value in earlier conversions to the extent that your retirement investments grow faster than inflation moves the tax brackets.

Personally, I look at tax and Roth conversion decisions in two ways: from a strategic position (long term choices of when and how much to convert, what accounts to draw when), and from a tactical position (given my income and tax brackets this year, what can I do to optimize right now. An example of a tactical choice would be to use a QCD to reduce gross income if it looked like you would be just a little over one of the IRMAA boundaries. But I also look at this in the context of the SS torpedo; if I'm near or above the top side of the torpedo, I will Roth convert as much additional as I can while staying within the "clean" 22% bracket. This is in hope that in future years I may be able to move down into the torpedo zone, or even below it someday.
McQ wrote: Fri Oct 15, 2021 2:14 pm One good reason not to convert way in advance

Upthread, FiveK and curmudgeon educated me on the nose cone of the social security tax torpedo—an income range of about $8000, where, with the exact right combination of social security, qualified dividends, and unwanted RMDs, your marginal rate would be 49.95%, or 55.5% post-TCJA, on each additional dollar of RMD. A Roth conversion that moved you down in that range could be the most lucrative ever.

But it’s a narrow range. How certain could you be, at age 59.5, that all the stars would align, so that twelve years later the RMDs reduced by conversion would fall into that narrow window and save you 49.95% / 55.5%? Not very. If IRMAA won’t be a problem, better to wait until your late 60s, when forecasts will have less error.

The social security torpedo more generally covers a wider range; but the high explosive compartment occupies a fairly narrow range at the bottom portion of the 22% bracket, where the marginal tax rate will be 40.7% or 46.25% post-TCJA, if you hit that bullseye. Again, a later conversion is more likely to harvest these juicy results because more certain the torpedo will be launched in the expected direction.
Making a dent in the SS torpedo requires long-term strategic planning. Any given $100,000 conversion may or may not move RMDs out of that marginal rate, but a sustained program of conversion has much more likelihood of clearing some or all of that zone (though the largest potential happens only IF you are in the smallish set of people with the right combination of minimal pension and largish SS benefit size and TDA size).

This discussion overall has helped me to recognize that there is a concept I hadn't been fully grasping, which is "how to I value retirement dollars at different points during retirement?" Beyond the question of real vs nominal dollars lies the question of the utility of extra funds at different stages of life. If I'm spending money today, Roth dollars are greater than TDA dollars by my applied tax rate. Alternately if you are leaving your estate to charity, Roth dollars would be equal to TDA dollars and spending on Roth conversions might be a loss (but if you live long enough the TDA may be largely emptied by RMDs anyway). I consider that it's quite possible (though by no means certain) that by our mid-80's we will have lost interest in spending as much; it's a pattern I've commonly seen among older relatives. There's also the potential for higher spending needs for extra care in the last few years of life. For me, dollars available to spend over the next twenty years are more valuable than dollars later, and I need to be aware of those implications. We plan to leave our estate largely to family, so prepaid taxes in the form of Roth conversions are not a loss, but our kids are already in good position for retirement, so optimizing their inheritances is lower priority.
curmudgeon
Posts: 2630
Joined: Thu Jun 20, 2013 11:00 pm

Re: Why Roth conversions always pay off—if you can hold on long enough

Post by curmudgeon »

McQ wrote: Wed Oct 13, 2021 5:58 pm Poor Bob and Barb. You’ll recall from upthread that they were a little too enamored of Roth conversions and not well enough informed about how tax brackets adjust for inflation. As a consequence, they converted at 22% to avert RMDs that were only going to be taxed at 12%. Sad.
......................

But hey Bob and Barb—maybe don’t convert at 22% if there is any meaningful likelihood that your RMDs would be taxed at a significantly lower rate? For a $100,000 conversion, remember, there will only be a $3650 swing in RMD payments in the first year if the conversion is done/not done. If subtracting that amount from projected income would put you close to the bottom of the 22% bracket as inflated in the future—well, I would hesitate to do that conversion.
An interesting point relative to this, however, is that the "SS torpedo" turns the 12% bracket into an effective 22.2% bracket for any couple with significant SS benefit. So that lost ground may well be imaginary. Even the 10% bracket turns into 18.5%, for a fairly minor effect.

As in many of these cases, I expect there are exceptions - a small SS benefit (WEP/GPO) with mid-size government pension comes to mind as a case where the torpedo vanishes pretty early.
single2019
Posts: 262
Joined: Thu May 27, 2021 12:13 pm

Re: Why Roth conversions always pay off—if you can hold on long enough

Post by single2019 »

Is there a layman summary of this thread?
sc9182
Posts: 2165
Joined: Wed Aug 17, 2016 7:43 pm

Re: Why Roth conversions always pay off—if you can hold on long enough

Post by sc9182 »

It will be addressed in the next post - as in: “Next post will address” :-)
User avatar
FiveK
Posts: 15693
Joined: Sun Mar 16, 2014 2:43 pm

Re: Why Roth conversions always pay off—if you can hold on long enough

Post by FiveK »

single2019 wrote: Sat Oct 16, 2021 9:00 pm Is there a layman summary of this thread?
Perhaps this: "A Roth conversion is always beneficial if the tax rate on the conversion now is lower than it will be in the future. It might be beneficial even if the future tax rate is lower than the current rate. The evaluation of "might be" depends on individual circumstances."
single2019
Posts: 262
Joined: Thu May 27, 2021 12:13 pm

Re: Why Roth conversions always pay off—if you can hold on long enough

Post by single2019 »

FiveK wrote: Sat Oct 16, 2021 9:26 pm
single2019 wrote: Sat Oct 16, 2021 9:00 pm Is there a layman summary of this thread?
Perhaps this: "A Roth conversion is always beneficial if the tax rate on the conversion now is lower than it will be in the future. It might be beneficial even if the future tax rate is lower than the current rate. The evaluation of "might be" depends on individual circumstances."
Thanks for the reply. That’s all you guys discussed in 290 posts?
User avatar
FiveK
Posts: 15693
Joined: Sun Mar 16, 2014 2:43 pm

Re: Why Roth conversions always pay off—if you can hold on long enough

Post by FiveK »

single2019 wrote: Sat Oct 16, 2021 9:56 pm
FiveK wrote: Sat Oct 16, 2021 9:26 pm
single2019 wrote: Sat Oct 16, 2021 9:00 pm Is there a layman summary of this thread?
Perhaps this: "A Roth conversion is always beneficial if the tax rate on the conversion now is lower than it will be in the future. It might be beneficial even if the future tax rate is lower than the current rate. The evaluation of "might be" depends on individual circumstances."
Thanks for the reply. That’s all you guys discussed in 290 posts?
Well, there are at least two areas that engender much discussion:
1) prediction of future tax rates that any given individual will encounter
2) the math behind how the inclusion of taxable accounts (either by choice or when forced) affects the traditional vs. Roth choice
Topic Author
McQ
Posts: 1414
Joined: Fri Jun 18, 2021 12:21 am
Location: California

Re: Why Roth conversions always pay off—if you can hold on long enough

Post by McQ »

Running Bum wrote: Fri Oct 15, 2021 2:41 pm ....

Well, that explains why my eyes get so bleary in this thread. IMO it's more useful to look at real life factors and help people determine how much, if any, Roth conversion they should be doing. That's probably been covered in other threads. It certainly has been at early-retirement.org, where I've been more active. There were a couple people over there who have used your research to "prove" that Roth conversions have very little benefit and usually only after age 90 at the so-called breakeven point. I've followed this thread trying to understand that rationale, but given that this is in the abstract, I'll stop trying.

...
Sorry this thread hasn’t worked for you. It is placed on the “Investments: Theory ... General” board for good reason. You might get more value from the many Roth conversion threads on the Personal Finance board, where all threads are expected to satisfy the criteria, “Is it personal?” and “Is it actionable?”

I believe the results laid out here are general, but I leave the generalization to individuals.

For my part, I remain grateful to the Bogleheads.org for permitting me to start a thread to work out the implications of and further refine the analyses launched in the SSRN paper. That means thinking out loud, which can seem messy and ... verbose.

In some ways I still feel like an invited guest here, and I would like state for the record how much I appreciate the privilege of being able to publish post after lengthy post covering some fairly abstract territory.

Next planned post: What if I pay the tax from outside the conversion?
You can take the academic out of the classroom by retirement, but you can't ever take the classroom out of his tone, style, and manner of approach.
Topic Author
McQ
Posts: 1414
Joined: Fri Jun 18, 2021 12:21 am
Location: California

Re: Why Roth conversions always pay off—if you can hold on long enough

Post by McQ »

single2019 wrote: Sat Oct 16, 2021 9:00 pm Is there a layman summary of this thread?
Short answer: no. It's designed for folks who want to dig deeper, especially those willing to look at a spreadsheet, or rather, multiple permutations of a base spreadsheet. That gets fractal after a while.

For your purposes, I recommend the fine wiki here at BH.
You can take the academic out of the classroom by retirement, but you can't ever take the classroom out of his tone, style, and manner of approach.
single2019
Posts: 262
Joined: Thu May 27, 2021 12:13 pm

Re: Why Roth conversions always pay off—if you can hold on long enough

Post by single2019 »

McQ wrote: Sun Oct 17, 2021 4:04 pm
single2019 wrote: Sat Oct 16, 2021 9:00 pm Is there a layman summary of this thread?
Short answer: no. It's designed for folks who want to dig deeper, especially those willing to look at a spreadsheet, or rather, multiple permutations of a base spreadsheet. That gets fractal after a while.

For your purposes, I recommend the fine wiki here at BH.
Is this discussion proposing a edit to the “fine wiki for my purpose”
User avatar
FiveK
Posts: 15693
Joined: Sun Mar 16, 2014 2:43 pm

Re: Why Roth conversions always pay off—if you can hold on long enough

Post by FiveK »

single2019 wrote: Sun Oct 17, 2021 9:00 pm
McQ wrote: Sun Oct 17, 2021 4:04 pm
single2019 wrote: Sat Oct 16, 2021 9:00 pm Is there a layman summary of this thread?
Short answer: no. It's designed for folks who want to dig deeper, especially those willing to look at a spreadsheet, or rather, multiple permutations of a base spreadsheet. That gets fractal after a while.

For your purposes, I recommend the fine wiki here at BH.
Is this discussion proposing a edit to the “fine wiki for my purpose”
Once McQ has a link to his spreadsheet that doesn't cause a security warning, the strategy outlined in this post ("enough to mention the possibility in those articles and refer people to threads like this and tools mentioned herein for those who want to evaluate the applicability to their own situations") will likely be implemented. Having two spreadsheet references (McQ's and MDM's) seems worthwhile.
wrongfunds
Posts: 3187
Joined: Tue Dec 21, 2010 2:55 pm

Re: Why Roth conversions always pay off—if you can hold on long enough

Post by wrongfunds »

McQ wrote: Sun Oct 17, 2021 3:59 pm Next planned post: What if I pay the tax from outside the conversion?
I have fundamental problem with this issue. Why is there "outside" money in the first place? Did it have a specific reason to be there? As somebody else aptly put that money did NOT fall from the sky! What is individual giving up by depleting the "outside" money and how would you account for that?

Frankly, analysis based upon the above assumption is close to basing it on "assume you have unlimited money ..."
Chip
Posts: 3994
Joined: Wed Feb 21, 2007 3:57 am

Re: Why Roth conversions always pay off—if you can hold on long enough

Post by Chip »

wrongfunds wrote: Mon Oct 18, 2021 8:45 am
McQ wrote: Sun Oct 17, 2021 3:59 pm Next planned post: What if I pay the tax from outside the conversion?
I have fundamental problem with this issue. Why is there "outside" money in the first place? Did it have a specific reason to be there? As somebody else aptly put that money did NOT fall from the sky! What is individual giving up by depleting the "outside" money and how would you account for that?

Frankly, analysis based upon the above assumption is close to basing it on "assume you have unlimited money ..."
I don't understand the concern. Many people max out their tax-deferred options and save additional money in taxable accounts. That's where all the tax payments came from for our own Roth conversions.
Post Reply