[**Update] - Metrics to Compare Roth Conversion Scenarios ??

Non-investing personal finance issues including insurance, credit, real estate, taxes, employment and legal issues such as trusts and wills.
User avatar
Lee_WSP
Posts: 10403
Joined: Fri Apr 19, 2019 5:15 pm
Location: Arizona

Re: [**Update] - Metrics to Compare Roth Conversion Scenarios ??

Post by Lee_WSP »

WoodSpinner wrote: Sat Jul 24, 2021 8:42 pm Metrics Considered:
  • Marginal Tax Rates vs. Future Marginal Rates (Fed + CA State)
  • Total Portfolio Value
  • Total Tax Adjusted Portfolio Value
  • Total Taxes (Fed +IRMAA + State) paid
  • Return on Investment (Tax Adjusted Portfolio Delta/ Total Taxes Paid Delta (Scenario compared to No Conversions)
  • Internal Rate of Return ( Tax Adjusted Portfolio Delta/ Total Taxes Paid Delta (Scenario compared to No Conversions)

I'm not seeing the update, but answering the question again, the two options I haven't crossed off are the most relevant, but neither are perfect for the reason I stated in my post above. You could die before it breaks even.

I think of the two remaining, marginal rates is still the best, but let's set that aside. Let's look at the other one.

Prima facie, it makes a lot of sense. Except that the calculation itself is flawed. Unless you run it through an actual lifetime simulator, you're not going to get an accurate result. But if you do run it through an actual simulator and calculate the total amount of after tax withdrawals for a given lifetime, that is the correct metric and should get you similar results to the marginal rate analysis, but with shorter lifespans preferring no or less conversions.
User avatar
FiveK
Posts: 15742
Joined: Sun Mar 16, 2014 2:43 pm

Re: Metrics to Compare Roth Conversion Scenarios ??

Post by FiveK »

Lee_WSP wrote: Sat Jul 31, 2021 1:51 pm
FiveK wrote: Sat Jul 31, 2021 1:24 pm Why do you think the Total After Tax Portfolio Value (TATPV) is not immediately breakeven? Given equal marginal tax rates for current and future conversions, I would expect "no change" in TATPV when the conversion occurs.
Because you could die before it actually breaks even.
Because it is immediately break even, how can one die before it breaks even? :confused
User avatar
Lee_WSP
Posts: 10403
Joined: Fri Apr 19, 2019 5:15 pm
Location: Arizona

Re: Metrics to Compare Roth Conversion Scenarios ??

Post by Lee_WSP »

FiveK wrote: Sat Jul 31, 2021 2:26 pm
Lee_WSP wrote: Sat Jul 31, 2021 1:51 pm
FiveK wrote: Sat Jul 31, 2021 1:24 pm Why do you think the Total After Tax Portfolio Value (TATPV) is not immediately breakeven? Given equal marginal tax rates for current and future conversions, I would expect "no change" in TATPV when the conversion occurs.
Because you could die before it actually breaks even.
Because it is immediately break even, how can one die before it breaks even? :confused
If you pay taxes today with the plan to pay less taxes tomorrow, but tomorrow never comes, you paid more taxes in life than necessary.

That's the theory, obviously one's heirs still benefit, which complicates the calculation, but not the underlying theory.
JoinToday
Posts: 1110
Joined: Sat Mar 10, 2007 8:59 pm

Re: Metrics to Compare Roth Conversion Scenarios ??

Post by JoinToday »

WoodSpinner wrote: Fri Jul 30, 2021 6:55 pm
Image
Observations:
1. I can't follow this thread very well, in part because I think your metrics are not correct, as others have noted. This has complicated the thread above, making it difficult for me to follow.

2. In the plot above, why isn't the tax adjusted portfolio value the same for all scenarios at age 62? (I assume you are 62, this is t=0; I may be in error). Regardless of what I do in the future, my portfolio has a certain tax adjusted portfolio value today, determined primarily by the tax rate I assume for trad-IRA and LTCG).

3. I may have missed this, but are you considering your heir's tax brackets in your analysis?

4. I think I am largely aligned with FiveK here. If I can do Roth conversions today that minimize taxes (in real dollars) or maximize spendable money (= after tax money) for me, my wife, and my heirs, that is what I want to do. I am not sure why you don't think the Total After Tax Portfolio Value (TATPV) changes when a Roth conversion is performed. Break even point is immediate, as FiveK states. If I die in a year of two, my wife & heirs benefit from reduced taxes. But the benefits mostly boil down to (a) tax rates now vs the future (tax arbitrage), and (b) reducing tax drag if taxes are paid from after tax account.
FiveK wrote: Sat Jul 31, 2021 1:24 pm
WoodSpinner wrote: Sat Jul 31, 2021 8:58 am
FiveK wrote: Fri Jul 30, 2021 7:54 pm
WoodSpinner wrote: Fri Jul 30, 2021 6:55 pm 1. Roth Conversions as a Tax-Arbitrage scheme won't break even until Spouse1 is about 84-88 years old.
For equal current and future tax rates, we know Roth conversions are immediately "break even" based on the commutative property of multiplication.

How are you defining "break even" such that it becomes time dependent? It certainly can be - it just isn't clear what assumptions are being used.
FiveK,

Using 2 metrics:
- From a tax Arbitrage perspective the Cumulative Taxes Paid (this occurs in the 84-88 year old range), see 1st graph. My observation is that tax arbitrage is not a strong driver for conversions.
- From a Total After Tax Portfolio Value (this occurs around 72) see second bar chart, all scenarios are now more valuable than the No Conversion.

WoodSpinner
Others have also commented on why "taxes paid" is not a useful metric.

Why do you think the Total After Tax Portfolio Value (TATPV) is not immediately breakeven? Given equal marginal tax rates for current and future conversions, I would expect "no change" in TATPV when the conversion occurs. But maybe I'm expecting something incorrectly...?
I wish I had learned about index funds 25 years ago
marcopolo
Posts: 8446
Joined: Sat Dec 03, 2016 9:22 am

Re: [**Update] - Metrics to Compare Roth Conversion Scenarios ??

Post by marcopolo »

Lee_WSP wrote: Sat Jul 31, 2021 1:58 pm
WoodSpinner wrote: Sat Jul 24, 2021 8:42 pm Metrics Considered:
  • Marginal Tax Rates vs. Future Marginal Rates (Fed + CA State)
  • Total Portfolio Value
  • Total Tax Adjusted Portfolio Value
  • Total Taxes (Fed +IRMAA + State) paid
  • Return on Investment (Tax Adjusted Portfolio Delta/ Total Taxes Paid Delta (Scenario compared to No Conversions)
  • Internal Rate of Return ( Tax Adjusted Portfolio Delta/ Total Taxes Paid Delta (Scenario compared to No Conversions)

I'm not seeing the update, but answering the question again, the two options I haven't crossed off are the most relevant, but neither are perfect for the reason I stated in my post above. You could die before it breaks even.

I think of the two remaining, marginal rates is still the best, but let's set that aside. Let's look at the other one.

Prima facie, it makes a lot of sense. Except that the calculation itself is flawed. Unless you run it through an actual lifetime simulator, you're not going to get an accurate result. But if you do run it through an actual simulator and calculate the total amount of after tax withdrawals for a given lifetime, that is the correct metric and should get you similar results to the marginal rate analysis, but with shorter lifespans preferring no or less conversions.
What is this break even period you speak of?
When considering equal tax rates, and tax adjusted portfolio value (TAPV), as FiveK is discussing, the conversions are break even immediately.
Consider a simple example where there is $10,000 in a TDA, and the combined tax rate is 30% (due to other income sources), and is expected to stay the same for the remainder of the analysis period (equal tax case).

The TAPV is $7000 with no conversions.
Converting the entire amount and paying the 30% tax leaves $7000 in a Roth with no tax liability, so a TAPV of $7000.
The break even is immediate.

It makes no difference when you convert or withdraw, thanks to the commutative property of multiplication
If you wait until there is 10x gain, then the TDA account is worth $100k, with TAPV of $70k, after paying $30k in taxes
If you had converted at the $10k value, paid the $3k in taxes, the Roth account would also grow 10x, leaving you with $70k in the Roth, so the same $70k TAPV.

This also illustrates the absurdity of using taxes paid as a metric. The TAPV is the same even though in one case you have paid 10x the cumulative taxes.

One can do a little better by paying the tax out of a taxable account, But, then the break even is also immediate, but then the conversion pulls ahead due to effectively having moved some money from a taxable account to a tax free account.


Can you provide a simple numerical example where there is >0 break even period under the equal tax rate scenario?
Once in a while you get shown the light, in the strangest of places if you look at it right.
User avatar
Lee_WSP
Posts: 10403
Joined: Fri Apr 19, 2019 5:15 pm
Location: Arizona

Re: [**Update] - Metrics to Compare Roth Conversion Scenarios ??

Post by Lee_WSP »

marcopolo wrote: Sat Jul 31, 2021 3:32 pm
Lee_WSP wrote: Sat Jul 31, 2021 1:58 pm
WoodSpinner wrote: Sat Jul 24, 2021 8:42 pm Metrics Considered:
  • Marginal Tax Rates vs. Future Marginal Rates (Fed + CA State)
  • Total Portfolio Value
  • Total Tax Adjusted Portfolio Value
  • Total Taxes (Fed +IRMAA + State) paid
  • Return on Investment (Tax Adjusted Portfolio Delta/ Total Taxes Paid Delta (Scenario compared to No Conversions)
  • Internal Rate of Return ( Tax Adjusted Portfolio Delta/ Total Taxes Paid Delta (Scenario compared to No Conversions)

I'm not seeing the update, but answering the question again, the two options I haven't crossed off are the most relevant, but neither are perfect for the reason I stated in my post above. You could die before it breaks even.

I think of the two remaining, marginal rates is still the best, but let's set that aside. Let's look at the other one.

Prima facie, it makes a lot of sense. Except that the calculation itself is flawed. Unless you run it through an actual lifetime simulator, you're not going to get an accurate result. But if you do run it through an actual simulator and calculate the total amount of after tax withdrawals for a given lifetime, that is the correct metric and should get you similar results to the marginal rate analysis, but with shorter lifespans preferring no or less conversions.
What is this break even period you speak of?
When considering equal tax rates, and tax adjusted portfolio value (TAPV), as FiveK is discussing, the conversions are break even immediately.
Consider a simple example where there is $10,000 in a TDA, and the combined tax rate is 30% (due to other income sources), and is expected to stay the same for the remainder of the analysis period (equal tax case).

The TAPV is $7000 with no conversions.
Converting the entire amount and paying the 30% tax leaves $7000 in a Roth with no tax liability, so a TAPV of $7000.
The break even is immediate.

It makes no difference when you convert or withdraw, thanks to the commutative property of multiplication
If you wait until there is 10x gain, then the TDA account is worth $100k, with TAPV of $70k, after paying $30k in taxes
If you had converted at the $10k value, paid the $3k in taxes, the Roth account would also grow 10x, leaving you with $70k in the Roth, so the same $70k TAPV.

This also illustrates the absurdity of using taxes paid as a metric. The TAPV is the same even though in one case you have paid 10x the cumulative taxes.

One can do a little better by paying the tax out of a taxable account, But, then the break even is also immediate, but then the conversion pulls ahead due to effectively having moved some money from a taxable account to a tax free account.


Can you provide a simple numerical example where there is >0 break even period under the equal tax rate scenario?
Well, under the tax adjusted portfolio metric, it's an immediate breakeven. But under the actual dollars spent or withdrawn metric, which most people actually care about, if you die before you withdraw, you don't break even.
JoinToday
Posts: 1110
Joined: Sat Mar 10, 2007 8:59 pm

Re: Metrics to Compare Roth Conversion Scenarios ??

Post by JoinToday »

WoodSpinner wrote: Sat Jul 31, 2021 9:36 am
marcopolo wrote: Sat Jul 31, 2021 3:03 am
At the risk of repeating myself and others.

If you are going to do this kind of detailed analysis, it seems really odd to do two things:

1) You are using cumulative taxes paid as a metric. As shown in a previous post cumulative taxes paid is a terrible metric to use to evaluate Roth Conversions.
Upthread there was a link to the Kitces article on Tax Equilibrium throughout your Retirement. See https://www.kitces.com/blog/tax-rate-eq ... nversions/

A graph of cumulative taxes paid seems to be a pretty useful way to visually see this, see the No Conversion and IRMAA Tier3 lines and notice the slope and bend points that occur when Conversions end and RMDs begin.

Secondly the amount of taxes paid between scenarios is important, it affects Cashflow during different periods.
2) You are correctly using tax adjusted portfolio value. So you clearly understand that portions of your portfolio should be derated by the taxes due. But, for some reason you are still ignoring tax adjusted asset allocation. You seem to be ignoring that most of the perceived gains you are getting is from taking on more risk. To do a consistent comparison (why go through all this detailed analysis If not to compare Apple to Apples scenarios), you would need to use equivalent tax adjusted asset allocation. If you are OK with higher equity allocation, you should have that in the less conversions cases as well.
Perhaps this is a difference between an academic exercise and a real world one (well at least a model of my real world). In my world, I am not going to Tax Adjust my asset allocation but I do acknowledge that there is a risk difference. I am comfortable with keeping my overall Asset Allocation and Rebalancing strategies consistent across all scenarios.

WoodSpinner
In my modelling, the differences between different roth conversion plans is surprisingly small, after considering the reduced future IRMAA surcharge, eliminated the future 3.8% NIIT, and 0% future LTCG tax rates. If you can do this easily, change your rate of return for both equity and bonds to the same rate of return, or change the rate of return for your Traditional IRA = Roth IRA. Does this change the results for the benefits of Roth conversions?

In my earlier modelling, I let excel optimize the roth conversions. Excel front loaded the Roth conversions, which was surprising to me. Turns out it was due entirely to the rate of return assumed for the two IRAs (trad., Roth). I have 100% bonds in my traditional IRA, and equity in my Roth (for now) with a higher rate of return. If you are not holding your AA constant, you are comparing apples to oranges. Whether the difference is significant or not should be considered. It may not be significant. But you should at least quantify the difference.
I wish I had learned about index funds 25 years ago
JoinToday
Posts: 1110
Joined: Sat Mar 10, 2007 8:59 pm

Re: [**Update] - Metrics to Compare Roth Conversion Scenarios ??

Post by JoinToday »

Lee_WSP wrote: Sat Jul 31, 2021 3:41 pm ....

Well, under the tax adjusted portfolio metric, it's an immediate breakeven. But under the actual dollars spent or withdrawn metric, which most people actually care about, if you die before you withdraw, you don't break even.
I don't think most people on this board care about actual dollars spent or withdrawn. I certainly don't.

I get the feeling most people here have more money than they need, and are concerned about their heirs also. Total spendable money by me, + wife, + heirs is what I want to maximize. Or maximize after tax portfolio value.

Breakeven time is not relevant in my modelling, I never consider it.
I wish I had learned about index funds 25 years ago
marcopolo
Posts: 8446
Joined: Sat Dec 03, 2016 9:22 am

Re: [**Update] - Metrics to Compare Roth Conversion Scenarios ??

Post by marcopolo »

Lee_WSP wrote: Sat Jul 31, 2021 3:41 pm
marcopolo wrote: Sat Jul 31, 2021 3:32 pm
Lee_WSP wrote: Sat Jul 31, 2021 1:58 pm
WoodSpinner wrote: Sat Jul 24, 2021 8:42 pm Metrics Considered:
  • Marginal Tax Rates vs. Future Marginal Rates (Fed + CA State)
  • Total Portfolio Value
  • Total Tax Adjusted Portfolio Value
  • Total Taxes (Fed +IRMAA + State) paid
  • Return on Investment (Tax Adjusted Portfolio Delta/ Total Taxes Paid Delta (Scenario compared to No Conversions)
  • Internal Rate of Return ( Tax Adjusted Portfolio Delta/ Total Taxes Paid Delta (Scenario compared to No Conversions)

I'm not seeing the update, but answering the question again, the two options I haven't crossed off are the most relevant, but neither are perfect for the reason I stated in my post above. You could die before it breaks even.

I think of the two remaining, marginal rates is still the best, but let's set that aside. Let's look at the other one.

Prima facie, it makes a lot of sense. Except that the calculation itself is flawed. Unless you run it through an actual lifetime simulator, you're not going to get an accurate result. But if you do run it through an actual simulator and calculate the total amount of after tax withdrawals for a given lifetime, that is the correct metric and should get you similar results to the marginal rate analysis, but with shorter lifespans preferring no or less conversions.
What is this break even period you speak of?
When considering equal tax rates, and tax adjusted portfolio value (TAPV), as FiveK is discussing, the conversions are break even immediately.
Consider a simple example where there is $10,000 in a TDA, and the combined tax rate is 30% (due to other income sources), and is expected to stay the same for the remainder of the analysis period (equal tax case).

The TAPV is $7000 with no conversions.
Converting the entire amount and paying the 30% tax leaves $7000 in a Roth with no tax liability, so a TAPV of $7000.
The break even is immediate.

It makes no difference when you convert or withdraw, thanks to the commutative property of multiplication
If you wait until there is 10x gain, then the TDA account is worth $100k, with TAPV of $70k, after paying $30k in taxes
If you had converted at the $10k value, paid the $3k in taxes, the Roth account would also grow 10x, leaving you with $70k in the Roth, so the same $70k TAPV.

This also illustrates the absurdity of using taxes paid as a metric. The TAPV is the same even though in one case you have paid 10x the cumulative taxes.

One can do a little better by paying the tax out of a taxable account, But, then the break even is also immediate, but then the conversion pulls ahead due to effectively having moved some money from a taxable account to a tax free account.


Can you provide a simple numerical example where there is >0 break even period under the equal tax rate scenario?
Well, under the tax adjusted portfolio metric, it's an immediate breakeven. But under the actual dollars spent or withdrawn metric, which most people actually care about, if you die before you withdraw, you don't break even.
I don't understand this at all. Are you spending less because you spent some money on Roth conversions?!?
I don't know anyone that does that.

If you spend the same amount on your self, and a little on Roth Conversions, how is that negatively impacting you before you die? The portfolio and the heirs still break even (under equal tax rate assumption).
Once in a while you get shown the light, in the strangest of places if you look at it right.
User avatar
Lee_WSP
Posts: 10403
Joined: Fri Apr 19, 2019 5:15 pm
Location: Arizona

Re: [**Update] - Metrics to Compare Roth Conversion Scenarios ??

Post by Lee_WSP »

marcopolo wrote: Sat Jul 31, 2021 3:59 pm
I don't understand this at all. Are you spending less because you spent some money on Roth conversions?!?
I don't know anyone that does that.

If you spend the same amount on your self, and a little on Roth Conversions, how is that negatively impacting you before you die? The portfolio and the heirs still break even (under equal tax rate assumption).
You spent less because you're dead. Although technically you neither spent nor saved any more since you're dead and that's the end of the scenario.

You are neither better nor worse off, you are dead. You paid more taxes before you died than you had to though.

Your heirs will probably pay less in taxes because of what you did though.

It all depends on the metric you are using which one is superior.

I don't think it's a great metric, but it does capture the finality of death.

Also, you can setup the model to be second to die or second to die and then ten years or death plus ten years.
User avatar
FiveK
Posts: 15742
Joined: Sun Mar 16, 2014 2:43 pm

Re: Metrics to Compare Roth Conversion Scenarios ??

Post by FiveK »

Lee_WSP wrote: Sat Jul 31, 2021 2:47 pm
FiveK wrote: Sat Jul 31, 2021 2:26 pm
Lee_WSP wrote: Sat Jul 31, 2021 1:51 pm
FiveK wrote: Sat Jul 31, 2021 1:24 pm Why do you think the Total After Tax Portfolio Value (TATPV) is not immediately breakeven? Given equal marginal tax rates for current and future conversions, I would expect "no change" in TATPV when the conversion occurs.
Because you could die before it actually breaks even.
Because it is immediately break even, how can one die before it breaks even? :confused
If you pay taxes today with the plan to pay less taxes tomorrow, but tomorrow never comes, you paid more taxes in life than necessary.

That's the theory, obviously one's heirs still benefit, which complicates the calculation, but not the underlying theory.
Thank you for explaining that perspective. Not sold on the theory, but vive la différence. Don't know if that is also woodspinner's perspective...?

Also a good example of why "rules of thumb" are problematic in this area: when one digs into the pertinent details behind such rules, one often finds the applicability much narrower than one would hope.
User avatar
Lee_WSP
Posts: 10403
Joined: Fri Apr 19, 2019 5:15 pm
Location: Arizona

Re: Metrics to Compare Roth Conversion Scenarios ??

Post by Lee_WSP »

FiveK wrote: Sat Jul 31, 2021 5:53 pm
Lee_WSP wrote: Sat Jul 31, 2021 2:47 pm
FiveK wrote: Sat Jul 31, 2021 2:26 pm
Lee_WSP wrote: Sat Jul 31, 2021 1:51 pm
FiveK wrote: Sat Jul 31, 2021 1:24 pm Why do you think the Total After Tax Portfolio Value (TATPV) is not immediately breakeven? Given equal marginal tax rates for current and future conversions, I would expect "no change" in TATPV when the conversion occurs.
Because you could die before it actually breaks even.
Because it is immediately break even, how can one die before it breaks even? :confused
If you pay taxes today with the plan to pay less taxes tomorrow, but tomorrow never comes, you paid more taxes in life than necessary.

That's the theory, obviously one's heirs still benefit, which complicates the calculation, but not the underlying theory.
Thank you for explaining that perspective. Not sold on the theory, but vive la différence. Don't know if that is also woodspinner's perspective...?

Also a good example of why "rules of thumb" are problematic in this area: when one digs into the pertinent details behind such rules, one often finds the applicability much narrower than one would hope.
I don't particularly like it either, but it's not completely defective so long as you model the plus ten years.

I've kind of given up trying to fully grasp wood spinners metric. To be honest.
User avatar
Topic Author
WoodSpinner
Posts: 3505
Joined: Mon Feb 27, 2017 12:15 pm

Re: Metrics to Compare Roth Conversion Scenarios ??

Post by WoodSpinner »

JoinToday wrote: Sat Jul 31, 2021 3:19 pm
1. I can't follow this thread very well, in part because I think your metrics are not correct, as others have noted. This has complicated the thread above, making it difficult for me to follow.
I hear you — part of the issue is I am learning, refining and remodeling to gain insights. My thinking and opinions are evolving.
2. In the plot above, why isn't the tax adjusted portfolio value the same for all scenarios at age 62? (I assume you are 62, this is t=0; I may be in error). Regardless of what I do in the future, my portfolio has a certain tax adjusted portfolio value today, determined primarily by the tax rate I assume for trad-IRA and LTCG).
Good question, the model is correct (took me a bit to figure it out) and here is what is going on.

With No Conversions, the Marginal Tax Rate is 18% (12% Federal, 6% State), after all the AGI doesn’t have any Roth Conversions.

The other Conversion scenarios have a higher Marginal Rate (31.3%, 33.3%, 33.33%) for IRMAA Tier1-3.

This changes the Tax Adjusted Portfolio Value since the Marginal Rate is used to Discount the Tax Adjusted Value.
3. I may have missed this, but are you considering your heir's tax brackets in your analysis?
I haven’t extended the model to deal with my heirs brackets at this point, it is part of my plans.
4. I think I am largely aligned with FiveK here. If I can do Roth conversions today that minimize taxes (in real dollars) or maximize spendable money (= after tax money) for me, my wife, and my heirs, that is what I want to do. I am not sure why you don't think the Total After Tax Portfolio Value (TATPV) changes when a Roth conversion is performed. Break even point is immediate, as FiveK states. If I die in a year of two, my wife & heirs benefit from reduced taxes. But the benefits mostly boil down to (a) tax rates now vs the future (tax arbitrage), and (b) reducing tax drag if taxes are paid from after tax account.
Will reply to FiveK directly, below ….

Appreciate your questions!

WoodSpinner
WoodSpinner
User avatar
Topic Author
WoodSpinner
Posts: 3505
Joined: Mon Feb 27, 2017 12:15 pm

Re: Metrics to Compare Roth Conversion Scenarios ??

Post by WoodSpinner »

FiveK wrote: Sat Jul 31, 2021 2:26 pm
Lee_WSP wrote: Sat Jul 31, 2021 1:51 pm
FiveK wrote: Sat Jul 31, 2021 1:24 pm Why do you think the Total After Tax Portfolio Value (TATPV) is not immediately breakeven? Given equal marginal tax rates for current and future conversions, I would expect "no change" in TATPV when the conversion occurs.
Because you could die before it actually breaks even.
Because it is immediately break even, how can one die before it breaks even? :confused
Within a Scenario I agree that the portfolio has an immediate break even since we are using the Tax Adjusted Portfolio Value and discounting by the Marginal Rate.

However, this is NOT true when you compare scenarios to each other. The main issue is:
- Marginal Tax Rates used to calculate the Tax Adjusted Portfolio Value are different for each scenario over time.
  • Marginal rate (at 62) for No Conversions is 18% (Fed 12%, CA 6%)
  • Marginal Rates (at 62) for IRMAA Tier1-3 are (31.3%, 33.3%, and 33.3%).
- Once RMDs begin at 72 the Marginal Rates are much closer:
  • Marginal rate (at 72) for No Conversions is 37.3% (Fed 28%, CA 9.3%)
  • Marginal Rates (at 72) for IRMAA Tier1-3 are (37.3%, 36%%, and 33%).
* AGI changes based on RMDs and SS COLA adjustments.
** Tax rates are assumed to revert in 2026 per TCJA and brackets increase with inflation by 2%

This was a really great question! It took a bit of digging to properly understand things. Thanks!

I will need to update the observations up-thread….

Thanks

WoodSpinner
Last edited by WoodSpinner on Sat Jul 31, 2021 7:29 pm, edited 1 time in total.
WoodSpinner
User avatar
Topic Author
WoodSpinner
Posts: 3505
Joined: Mon Feb 27, 2017 12:15 pm

Re: Metrics to Compare Roth Conversion Scenarios ??

Post by WoodSpinner »

JoinToday wrote: Sat Jul 31, 2021 3:50 pm In my modelling, the differences between different roth conversion plans is surprisingly small, after considering the reduced future IRMAA surcharge, eliminated the future 3.8% NIIT, and 0% future LTCG tax rates. If you can do this easily, change your rate of return for both equity and bonds to the same rate of return, or change the rate of return for your Traditional IRA = Roth IRA. Does this change the results for the benefits of Roth conversions?

In my earlier modelling, I let excel optimize the roth conversions. Excel front loaded the Roth conversions, which was surprising to me. Turns out it was due entirely to the rate of return assumed for the two IRAs (trad., Roth). I have 100% bonds in my traditional IRA, and equity in my Roth (for now) with a higher rate of return. If you are not holding your AA constant, you are comparing apples to oranges. Whether the difference is significant or not should be considered. It may not be significant. But you should at least quantify the difference.
We have very different starting points for the IRA, mine 52/48. During Conversions equities are shifted from the IRA to Roth (taxes paid from the IRA Bonds). This will shift to 27/73 assuming a IRMAA Tier2 conversion scenario.

The overall Portfolio AA is 60/40, rebalanced yearly. I am not adopting a Tax Adjusted Asset Allocation (see discussion upthread).

I believe I am correctly modeling (at least for my situation) the shift in Asset Location of Equities from the IRA to the Roth.

WoodSpinner
WoodSpinner
JoinToday
Posts: 1110
Joined: Sat Mar 10, 2007 8:59 pm

Re: Metrics to Compare Roth Conversion Scenarios ??

Post by JoinToday »

WoodSpinner wrote: Sat Jul 31, 2021 6:34 pm .....
2. In the plot above, why isn't the tax adjusted portfolio value the same for all scenarios at age 62? (I assume you are 62, this is t=0; I may be in error). Regardless of what I do in the future, my portfolio has a certain tax adjusted portfolio value today, determined primarily by the tax rate I assume for trad-IRA and LTCG).
.....

Good question, the model is correct (took me a bit to figure it out) and here is what is going on.

With No Conversions, the Marginal Tax Rate is 18% (12% Federal, 6% State), after all the AGI doesn’t have any Roth Conversions.

The other Conversion scenarios have a higher Marginal Rate (31.3%, 33.3%, 33.33%) for IRMAA Tier1-3.

This changes the Tax Adjusted Portfolio Value since the Marginal Rate is used to Discount the Tax Adjusted Value.


3. I may have missed this, but are you considering your heir's tax brackets in your analysis?

I haven’t extended the model to deal with my heirs brackets at this point, it is part of my plans.


WoodSpinner
In one of my spreadsheets, I model my tax adjusted portfolio value differently than what you do. For all scenarios, I discount the traditional IRA by the same tax rate. If I have no roth conversions and a low marginal tax rate for a few years, my portfolio is not worth more than if I were doing large roth conversions with a higher marginal rate. For what it is worth, I assume 25% or 30% of my traditional IRA will be taxes. I also assume either 0% tax rate on after tax account (stepped up basis upon our death) or 10% to accommodate Fed LTCG tax and Calif tax. I think I use the same numbers for my heir's tax rate (traditional IRA only).
I wish I had learned about index funds 25 years ago
User avatar
FiveK
Posts: 15742
Joined: Sun Mar 16, 2014 2:43 pm

Re: Metrics to Compare Roth Conversion Scenarios ??

Post by FiveK »

WoodSpinner wrote: Sat Jul 31, 2021 7:05 pm
  • Marginal rate (at 62) for No Conversions is 18% (Fed 12%, CA 6%)
  • Marginal Rates (at 62) for IRMAA Tier1-3 are (31.3%, 33.3%, and 33.3%).
You get a year free of IRMAA concerns at age 62 - might be a good year to make a large Roth conversion.
- Once RMDs begin at 72 the Marginal Rates are much closer:
  • Marginal rate (at 72) for No Conversions is 37.3% (Fed 28%, CA 9.3%)
  • Marginal Rates (at 72) for IRMAA Tier1-3 are (37.3%, 36%%, and 33%).
That seems to indicate that pre-RMD (not sure how SS benefits enter here) conversions up through the top of the 24% bracket could be beneficial - especially if taxes are paid with cash on hand. Although, it's probably a relatively "flat optimum," meaning a wide range of conversion amounts provide about the same result. Is that what you see?
ChrisC
Posts: 1475
Joined: Tue Jun 19, 2012 9:10 am
Location: North Carolina

Re: Metrics to Compare Roth Conversion Scenarios ??

Post by ChrisC »

RetiredCSProf wrote: Fri Jul 30, 2021 11:28 am
celia wrote: Thu Jul 29, 2021 5:44 pm
In regards to the primary question, the "metric" or goal is to maximize your spendable money. ... The easiest guess is to figure out what will happen if one does not convert, then make the best-estimate of a Roth conversion plan you can. If you over-convert or under-convert, oh well, at least you tried, and you can always make adjustments each year as you go along.

I want to say that even when we were converting, our portfolio still kept on growing, but I can't really prove that since the the inheritance was partly taxable, tax-deferred, and Roth. ... But it would be interesting to hear of anyone else whose portfolio kept on growing through the years they did Roth conversions.
My tax-deferred portfolio has continued to grow in retirement while taking partial Roth conversions, which I started in 2015, at age 67. I had nothing in Roth when I retired at age 64. My TDA at the end of 2012 (when I stopped contributing to TDA) had a balance of about $400K. I think my AA was about 70/30, but I would need to review my records to check. I have converted about $200K and plan to continue making partial conversions for a few more years.

My current TDA balance is about $600K; AA of 56/39/5. I withdrew $12K from TDA in 2013 to fill gap between pension and expenses, spread Roth conversions over several years, and withdrew RMDs/QCDs at age 70, 71, and 73 (this year). I could have slowed the growth in TDA more by starting sooner (in 2013), adjusting sooner to a more conservative AA in TDA, and by converting bigger chunks.

I am not sure how to measure the growth in TDA, given that not all withdrawals were for Roth conversions, my conversions were spread unevenly across multiple years, and my AA fluctuated.
Isn't growth in TDAs and Roth accounts solely dependent on asset make-up in the accounts? Our conversions haven't kept up with the growth in my TDAs, and our Roth accounts have grown significantly as a result of moving several highly appreciated individual stocks from TDA brokerage accounts to the Roth brokerage accounts (and the index funds also moved have done well too). We wish we had done more sizeable conversions during the first few years after our retirement in 2013, instead of the major conversions we've done the last 4 years where we've averaged $150K in conversions. We maintain an 80-20 allocation across TDA and Roth accounts. In May of this year, I did a conversion of $150K; as of today, growth in my TDA accounts is at $270K.

If we were to convert an extra $10k this year, with spouse scheduled to receive SS retirement benefits in September, we'll end up in the 32% bracket and trip into tier 4 of IRMAA -- I can't see myself going there. The major metric for us is leaving as much as we feel comfortable to leave to our heirs in Roth accounts. We don't contemplate spending any of our funds in our TDA or Roth accounts for our consumption purposes. Indirectly, we're maximizing my heir's spendable money (and they will likely be in very high tax brackets) or backstopping them for potential bad events.
User avatar
Topic Author
WoodSpinner
Posts: 3505
Joined: Mon Feb 27, 2017 12:15 pm

Re: Metrics to Compare Roth Conversion Scenarios ??

Post by WoodSpinner »

FiveK wrote: Sat Jul 31, 2021 8:35 pm
WoodSpinner wrote: Sat Jul 31, 2021 7:05 pm
  • Marginal rate (at 62) for No Conversions is 18% (Fed 12%, CA 6%)
  • Marginal Rates (at 62) for IRMAA Tier1-3 are (31.3%, 33.3%, and 33.3%).
You get a year free of IRMAA concerns at age 62 - might be a good year to make a large Roth conversion.
Agreed, good idea!

- Once RMDs begin at 72 the Marginal Rates are much closer:
  • Marginal rate (at 72) for No Conversions is 37.3% (Fed 28%, CA 9.3%)
  • Marginal Rates (at 72) for IRMAA Tier1-3 are (37.3%, 36%%, and 33%).
That seems to indicate that pre-RMD (not sure how SS benefits enter here) conversions up through the top of the 24% bracket could be beneficial - especially if taxes are paid with cash on hand. Although, it's probably a relatively "flat optimum," meaning a wide range of conversion amounts provide about the same result. Is that what you see?
[/quote]

I am using the OpenSocialSecurity recommendation (Spouse at 65 and myself at 70).

The 24% bracket is really similar to IRMAA Tier3 — I can include those results as well. My thought is that this is too aggressive after 62.

WoodSpinner
WoodSpinner
User avatar
Topic Author
WoodSpinner
Posts: 3505
Joined: Mon Feb 27, 2017 12:15 pm

Re: Metrics to Compare Roth Conversion Scenarios ??

Post by WoodSpinner »

ChrisC wrote: Sat Jul 31, 2021 8:40 pm
RetiredCSProf wrote: Fri Jul 30, 2021 11:28 am
celia wrote: Thu Jul 29, 2021 5:44 pm
In regards to the primary question, the "metric" or goal is to maximize your spendable money. ... The easiest guess is to figure out what will happen if one does not convert, then make the best-estimate of a Roth conversion plan you can. If you over-convert or under-convert, oh well, at least you tried, and you can always make adjustments each year as you go along.

I want to say that even when we were converting, our portfolio still kept on growing, but I can't really prove that since the the inheritance was partly taxable, tax-deferred, and Roth. ... But it would be interesting to hear of anyone else whose portfolio kept on growing through the years they did Roth conversions.
My tax-deferred portfolio has continued to grow in retirement while taking partial Roth conversions, which I started in 2015, at age 67. I had nothing in Roth when I retired at age 64. My TDA at the end of 2012 (when I stopped contributing to TDA) had a balance of about $400K. I think my AA was about 70/30, but I would need to review my records to check. I have converted about $200K and plan to continue making partial conversions for a few more years.

My current TDA balance is about $600K; AA of 56/39/5. I withdrew $12K from TDA in 2013 to fill gap between pension and expenses, spread Roth conversions over several years, and withdrew RMDs/QCDs at age 70, 71, and 73 (this year). I could have slowed the growth in TDA more by starting sooner (in 2013), adjusting sooner to a more conservative AA in TDA, and by converting bigger chunks.

I am not sure how to measure the growth in TDA, given that not all withdrawals were for Roth conversions, my conversions were spread unevenly across multiple years, and my AA fluctuated.
Isn't growth in TDAs and Roth accounts solely dependent on asset make-up in the accounts? Our conversions haven't kept up with the growth in my TDAs, and our Roth accounts have grown significantly as a result of moving several highly appreciated individual stocks from TDA brokerage accounts to the Roth brokerage accounts (and the index funds also moved have done well too). We wish we had done more sizeable conversions during the first few years after our retirement in 2013, instead of the major conversions we've done the last 4 years where we've averaged $150K in conversions. We maintain an 80-20 allocation across TDA and Roth accounts. In May of this year, I did a conversion of $150K; as of today, growth in my TDA accounts is at $270K.

If we were to convert an extra $10k this year, with spouse scheduled to receive SS retirement benefits in September, we'll end up in the 32% bracket and trip into tier 4 of IRMAA -- I can't see myself going there. The major metric for us is leaving as much as we feel comfortable to leave to our heirs in Roth accounts. We don't contemplate spending any of our funds in our TDA or Roth accounts for our consumption purposes. Indirectly, we're maximizing my heir's spendable money (and they will likely be in very high tax brackets) or backstopping them for potential bad events.
ChrisC,

It sounds like you are doing lots of conversions and keeping an eye out on the marginal brackets.

How are you deciding what is the right level to convert? How did you pick IRMAA Tier 3 vs another choice?

We have been fortunate with great market returns. My only thought is to revisit the Conversion strategy yearly and adapt. Are you doing something similar?

WoodSpinner
WoodSpinner
ChrisC
Posts: 1475
Joined: Tue Jun 19, 2012 9:10 am
Location: North Carolina

Re: [**Update] - Metrics to Compare Roth Conversion Scenarios ??

Post by ChrisC »

Woodspinner,

I'm not a spreadsheet person, though I did fumble my way through RPM at one time and gave up because of the myriad of variables that impact conversion decisions. We stumbled into IRMAA and stayed within tier 2 for the first few years of Medicare Part B enrollment-- and we were actually thinking of avoiding Medicare Part B altogether as I have very good retiree health insurance coverage. But only one of us is enrolled in Medicare Part B and in the future we might drop this Part B coverage for my wife. I've always thought the IRMAA surcharges were minor issues in the grander scheme of things, BTW.

As said before, our goal in conversions is driven by legacy concerns, and we have 3 children, one in the 24% and the other 2 in the 35% and 37%, tax brackets, So, with the elimination of stretch we made a greater effort to convert, and it's likely that all three of my children with great earnings potential will be at the 37% personal or trust income tax bracket.

We've decided to convert up to the 24% bracket the last 4 years (and we were already at the higher end of the 22% bracket without conversions based solely on pensions during those years). And we wanted to make sure at least one of us fully converted his or her TDA; strategically, it made sense for us to convert my wife's TDA as she had less to convert and was 2 years older than me, and we were then facing an RMD age of 70.5. We've given up our fully converting my TDA before I reach the new RMD age of 72, which we be in 4 years for me. We decided we'll probably continue to convert even after we get rolled by RMDs.

Marginal tax brackets for us are different depending on which spouse is converting the TDA. My wife's conversions came with a 4.5 state income tax hit in NC, whereas my conversions from my Federal TSP account are not subject to that 4.5 state income tax hit.

I'm evaluating conversions annually on the basis of what my gut tells me, though my accountant looks at our returns and thinks we should stop when I reach 72.
User avatar
Johnsson
Posts: 517
Joined: Mon Jul 17, 2017 2:28 pm

Re: Metrics to Compare Roth Conversion Scenarios ??

Post by Johnsson »

FiveK wrote: Sat Jul 31, 2021 8:35 pm
WoodSpinner wrote: Sat Jul 31, 2021 7:05 pm
  • Marginal rate (at 62) for No Conversions is 18% (Fed 12%, CA 6%)
  • Marginal Rates (at 62) for IRMAA Tier1-3 are (31.3%, 33.3%, and 33.3%).
You get a year free of IRMAA concerns at age 62 - might be a good year to make a large Roth conversion.
- Once RMDs begin at 72 the Marginal Rates are much closer:
  • Marginal rate (at 72) for No Conversions is 37.3% (Fed 28%, CA 9.3%)
  • Marginal Rates (at 72) for IRMAA Tier1-3 are (37.3%, 36%%, and 33%).
That seems to indicate that pre-RMD (not sure how SS benefits enter here) conversions up through the top of the 24% bracket could be beneficial - especially if taxes are paid with cash on hand. Although, it's probably a relatively "flat optimum," meaning a wide range of conversion amounts provide about the same result. Is that what you see?
I've run many scenarios. I agree there's a wide range of conversions that provide similar results. How much should be converted seems to mostly depend on your mix of pre-tax and after-tax money (more pre-tax $$ = more conversions needed). The main question and impact for me (and I think for most ready to convert now) should be... Pay taxes now or pay later (at some future rate)? I have cash in hand and want to get the largest conversions done before tax rates are currently scheduled to increase. Future rates will certainly impact when conversions should have been done. I don't want to discuss possible future tax rates... not allowed.

While I've enjoyed reading all the discussion, I think most of it is likely moot because of all the future unknowns for each of us. I believe the best answer is determined by modeling many different scenarios and convert to the level that 'feels' right for your situation (especially since there is the "flat optimum" you mention). Precise modeling will be made worthless as life throws changes at us.

DW and I are 60 this year. My current plan is SS at 67 and 70 for us (but may do 70 and 70 with more conversions).

We have about 60% of our funds (2MM) in Pre-tax, with enough in after-tax to pay taxes and get us to SS.

We will be converting to the top of 24% this year and next. Then into 12/15% thereafter, with a total of 1.3MM converted (with all conversions done by age 66, before SS)

We'll be at IRMAA-2 when 65 and 66, then -1 thereafter... the price of conversions.

As discussed previously in this thread, I may cut off the conversions early because the pre-tax will be all-bonds and it's likely to not grow as the Roth will. Still thinking about that one...
'In theory there is no difference between theory and practice. In practice there is.' Yogi Berra
RetiredCSProf
Posts: 1228
Joined: Tue Feb 28, 2017 3:59 pm

Re: Metrics to Compare Roth Conversion Scenarios ??

Post by RetiredCSProf »

Johnsson wrote: Sun Aug 01, 2021 5:34 am
As discussed previously in this thread, I may cut off the conversions early because the pre-tax will be all-bonds and it's likely to not grow as the Roth will. Still thinking about that one...
I posted a thread, a few month ago, asking whether it made sense to maintain my AA (70/30) across pre-tax and Roth, given that my pre-tax was becoming increasingly bond-heavy as I shifted equities in-kind from TDA to Roth. My question was prompted by my Fido advisor recommending that I keep my rollover IRA at 60/40 AA to promote continued growth, not just preservation of capital. His rationale was based on my intention to leave all Roth for inheritance and spend a portion of my TDA (deplete at 100-something).

I do not have a set stopping point for conversions. I am already collecting SS and taking RMDs. If all remaining provisions of the Tax Reform Act of 2017 sunset, then my taxes are more likely to decrease than increase. The biggest potential change in my effective tax rate will be if I get "promoted / demoted" from filing as HoH to single status. My dependent son recently graduated from college and is job-hunting.
Phil DeMuth
Posts: 236
Joined: Thu Mar 06, 2008 6:02 pm

Re: [**Update] - Metrics to Compare Roth Conversion Scenarios ??

Post by Phil DeMuth »

Why not just run the numbers through ETF Math Guy's free App and see if you can beat his recommendation, which is grounded in academic research?

Disclosure: ETF Math Guy is a friend but we have no financial relationship.

https://app.etfmathguy.com/
NancyABQ
Posts: 451
Joined: Thu Aug 18, 2016 3:37 pm

Re: [**Update] - Metrics to Compare Roth Conversion Scenarios ??

Post by NancyABQ »

Phil DeMuth wrote: Sun Aug 01, 2021 11:16 am Why not just run the numbers through ETF Math Guy's free App and see if you can beat his recommendation, which is grounded in academic research?

Disclosure: ETF Math Guy is a friend but we have no financial relationship.

https://app.etfmathguy.com/
Looks interesting, but:

1) Doesn't let me retire earlier than 60
2) Looks like it was written several years ago. Can't enter the year 2022 as start of retirement, for example
3) It got some internal error when I tried to run my numbers. Could be "server overloaded" or an error in the application, no idea what.

Maybe your friend could update it?
User avatar
cowdogman
Posts: 2072
Joined: Sat Dec 16, 2017 6:44 pm
Location: Washington State

Re: Metrics to Compare Roth Conversion Scenarios ??

Post by cowdogman »

Johnsson wrote: Sun Aug 01, 2021 5:34 am
I've run many scenarios. I agree there's a wide range of conversions that provide similar results. How much should be converted seems to mostly depend on your mix of pre-tax and after-tax money (more pre-tax $$ = more conversions needed). The main question and impact for me (and I think for most ready to convert now) should be... Pay taxes now or pay later (at some future rate)? I have cash in hand and want to get the largest conversions done before tax rates are currently scheduled to increase. Future rates will certainly impact when conversions should have been done. I don't want to discuss possible future tax rates... not allowed.

While I've enjoyed reading all the discussion, I think most of it is likely moot because of all the future unknowns for each of us. I believe the best answer is determined by modeling many different scenarios and convert to the level that 'feels' right for your situation (especially since there is the "flat optimum" you mention). Precise modeling will be made worthless as life throws changes at us.
Hear! Hear!

I've learned a lot from this thread, especially looking at tax-adjusted portfolio values and not trying to look at breakeven (even tho I can't get the amounts to balance as nicely as they do in theory).

I have been running Roth scenarios for months and the $ results are always the same. Both the year-to-year and the overall impact is minimal and the upfront dollar cost is high (acknowledging the after-tax portfolio argument). I think that is because of the fact that our after-tax account is larger than our tax-deferred account.

One metric that I just started to notice this week is our after-tax/tax deferred allocation.

With no Roth conversions and no withdrawals from our TDA other than RMDs (which is our plan) our AT/TD balance at 71 is 69/31 and at 92 is 91/9 (primarily because a large chunk of our AT accounts are in (hopefully growing) stocks and the RMDs will be reducing the TDA accounts which are all bond funds).

With Roth conversions ($100K/year from 64 to 71--$800K) and no withdrawals from our TDA other than RMDs our AT/TD balance at 71 is 76/24 and at 92 is 93/7.

That made me feel better about having our heirs deal with TDAs.

As I said the effect of Roth conversions for us is minimal. Sometimes I wish I could make Roth conversions more attractive because it's more satisfying to do something than to do nothing. But based on current tax law (which we need to assume continues--otherwise the modeling would be all over the place) doing nothing seems to be the best bet--or as good a bet as any other (for us).
diy60
Posts: 972
Joined: Wed Sep 07, 2016 6:54 pm

Re: [**Update] - Metrics to Compare Roth Conversion Scenarios ??

Post by diy60 »

WoodSpinner wrote: Sat Jul 31, 2021 9:53 pm How are you deciding what is the right level to convert? How did you pick IRMAA Tier 3 vs another choice?
WoodSpinner, here is another method to add to the mix. Maybe you'll find some of it useful.

This is roughly how I made my determination on the size of my ongoing Roth conversions; 3 steps

1) PRESENT:
Step 1a (do nothing, status quo) : calculate base income and all related major costs (taxes, pension, IRMAA).
Step 1b (do something, Roth conversion) : prepare lookup table to determine costs over a broad range of conversion amounts.
Step 1c (calculate current marginal rate) : Using steps 1a as a base & 1b above, where current marginal rate = "current delta costs / associated change in current income". This produces a range of current marginal rates associated with changing delta costs and related changes in income.

2) FUTURE:
Step 2a (future do nothing result) : calculate base income as a result of status quo (taxes, SS, pension, RMD).
Step 2b (future do something result) : prepare lookup table to determine costs over broad range of converted amounts.
Step 2c (calculate future marginal rate) : where future marginal rate = "future delta costs / associated change future income". This produces a range of future marginal rates, calculations similar to Step 1c.

3) FINAL STEP :
set up solver routine (I use OpenSolver) to solve for convergence of current marginal rate = future marginal rate. My results indicate I should convert up to just below Tier 2 IRMAA. Since I believe Roth IRA has asymmetrical risks vs. Traditional IRA I'm not willing to gamble my assumptions are correct, so I stay just below Tier 1 IRMAA. I err on the side of too much traditional.

Notes :
Solver is not really necessary, you could do this with a few manual iterative Roth conversion amounts. Goal Seek may work as well, but I'm not familiar with that function.

My situation is simple, so I just capture the big cost hitters (taxes, IRMAA).

I currently use an average RMD withdrawal of 5% for the analysis period to keep the table simple. I have not expanded the analysis to include death of one spouse, nor the tax bracket of non-spousal heirs. Too many unknowns so I have not been motivated to expand my quick and dirty analysis.

I use taxable account dollars to fund conversion taxes.

I use current dollars since brackets and IRMAA are indexed to inflation. My tax deferred is nearly all bond-related, so not much growth in those accounts to worry about.

I don't understand how break even and ROI enter into the analysis, but my personality is to take a simple approach and move on.

I don't tax-adjust my portfolio in the analysis, though I probably should. (I do keep the total AA balance the same for current vs future to avoid taking on more risk).

That's it. I threw it together on one cold winter day. My write-up is way more complicated than the spreadsheet. My advice would be keep it simple and go for the low hanging fruit.

Best of luck.
McQ
Posts: 1425
Joined: Fri Jun 18, 2021 12:21 am
Location: California

Re: [**Update] - Metrics to Compare Roth Conversion Scenarios ??

Post by McQ »

In hopes of advancing the discussion of how to model Roth conversion outcomes, allow me to introduce McQuarrie’s rule:

Sooner or later, in any such discussion, a poster will introduce a simple 2-period game “for clarity,” in which the conversion occurs in period 1 and a total distribution occurs in period 2.

Outcomes will be summarized with reference to the commutative property of multiplication. The poster will cite this property as conclusive proof that if tax rates are the same in period 1 and period 2, then the Roth conversion must be a wash (or exactly breakeven); whereas if future tax rates are higher, the conversion was the right decision and must produce a better after-tax outcome. End of story.

The OP, who introduced the question of how to model outcomes over time, will protest that the example is too simple and does not clarify things at all. The 2-period poster will respond along the lines of “what part of simple arithmetic did I not understand?” The ensuing dialogue, if it occurs, will satisfy no one, until finally a third party will opine that there are too many future unknowns, just convert up until it feels right.

To advance the discussion, let me suggest that Woodspinner is attempting to model an n-period game, n > 2 (and mostly, > 20). Since he is attempting to model real world results under the US system of taxation, as it is today and as it will become, Woodspinner must model a very particular kind of n-period game. Properties:
1. n can never be known until the game is over
2. Fractional distributions from the TDA of 1/k occur in each of the n periods, and each such distribution is subject to the then prevailing rate of tax
3. The fraction 1/k is set by rule (aka, the RMD schedule), and initial fractions are very small, k about 27; however, fractions are progressive, growing each year up until a limit of k = 2

Once the problem is conceptualized in these terms, the commutative property of multiplication becomes irrelevant, because the no-conversion and the conversion scenarios are not exposed to tax at the same time or over the same n of periods.

A simple n=3 example may help. Appreciation is 10% in each period; all accounts are invested in the same asset; tax is always 24% ordinary, 15% capital gains and dividends.

n=1: Either a conversion of $100,000 is made, and tax at 24% is deducted, creating an initial Roth balance of $76,000; or no conversion occurs, leaving $100,000 in the TDA

n=2: if no conversion, there is now $110,000 in the TDA just before an RMD is taken. That (age 72) RMD will be $4029; tax of $967 is due. The remaining $3062 is invested in a taxable account. End year TDA balance after RMD is $105,971. End year taxable account balance is $3062. The Roth has appreciated to $83,600.

n=3: The TDA grows to $116 568 before the RMD, which at age 73 is $4415; tax of $1060 is due. The remaining $3356 is added to the taxable account. In the meantime, last year’s addition has appreciated by 8.5% after tax. The total and after-tax value of the taxable account is now $6678. The pretax end year TDA balance after RMD is $112,152. The Roth has appreciated to $91,960.

The game ends and after-tax values are computed. The TDA provides $85,236 after tax; with the taxable account, the no-conversion scenario has produced $91,914 of after-tax wealth.

Conclusion: Roth wins. Its $91,960 balance is $46 greater. Even though tax rates were constant.

In this three period game, the Roth “winnings” are paltry, giving a 2 year ROI on the $24,000 tax debit equal to an annualized 0.10%. But after 20 years, the advantage of the Roth will have compounded further, into something quite reasonable. In fact, after a dozen years the Roth advantage will be able to withstand a decline in future tax rates of several percentage points, as when a converter converts themselves out of the 24% bracket into the 22% bracket. After forty years, the Roth advantage may handle a decline of ten points in tax rate.

Finally, the greater the number of periods the game lasts, the greater the Roth advantage. It is compounding at a flat 10% after tax; everything else is compounding at a slower rate. Meantime, as k shrinks and shrinks, more and more of the TDA is lost to tax. Eventually, the Roth balance will exceed even the pretax values of the TDA plus the taxable account.
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.
JoinToday
Posts: 1110
Joined: Sat Mar 10, 2007 8:59 pm

Re: [**Update] - Metrics to Compare Roth Conversion Scenarios ??

Post by JoinToday »

McQ wrote: Sun Aug 01, 2021 5:27 pm

Conclusion: Roth wins. Its $91,960 balance is $46 greater. Even though tax rates were constant.

In this three period game, the Roth “winnings” are paltry, giving a 2 year ROI on the $24,000 tax debit equal to an annualized 0.10%. But after 20 years, the advantage of the Roth will have compounded further, into something quite reasonable. In fact, after a dozen years the Roth advantage will be able to withstand a decline in future tax rates of several percentage points, as when a converter converts themselves out of the 24% bracket into the 22% bracket. After forty years, the Roth advantage may handle a decline of ten points in tax rate.

Finally, the greater the number of periods the game lasts, the greater the Roth advantage. It is compounding at a flat 10% after tax; everything else is compounding at a slower rate. Meantime, as k shrinks and shrinks, more and more of the TDA is lost to tax. Eventually, the Roth balance will exceed even the pretax values of the TDA plus the taxable account.
The $46 is the tax drag on the RMD put in the after tax account (10% gain x 15% tax rate x $3062) [from what I can see]. This has led me to focus on reducing the amount of money in my after tax account. Pay tax from after tax account if at all possible, balancing the LTCG taxes with the yearly tax drag of money in my after tax account.

I don't think your example negates the commutative property that others have mentioned, and you mention in your post.

For me, the biggest drivers have been (and still are):
1. Tax rates, current vs future
2. reducing the number of years (= $$) I spend in the higher IRMAA brackets
3. With lower income in the future, my LTCG will drop to 0% for a portion of my dividends
4. Tax drag (as you demonstrated in your example above).
Last edited by JoinToday on Sun Aug 01, 2021 6:13 pm, edited 1 time in total.
I wish I had learned about index funds 25 years ago
marcopolo
Posts: 8446
Joined: Sat Dec 03, 2016 9:22 am

Re: [**Update] - Metrics to Compare Roth Conversion Scenarios ??

Post by marcopolo »

McQ wrote: Sun Aug 01, 2021 5:27 pm In hopes of advancing the discussion of how to model Roth conversion outcomes, allow me to introduce McQuarrie’s rule:

Sooner or later, in any such discussion, a poster will introduce a simple 2-period game “for clarity,” in which the conversion occurs in period 1 and a total distribution occurs in period 2.

Outcomes will be summarized with reference to the commutative property of multiplication. The poster will cite this property as conclusive proof that if tax rates are the same in period 1 and period 2, then the Roth conversion must be a wash (or exactly breakeven); whereas if future tax rates are higher, the conversion was the right decision and must produce a better after-tax outcome. End of story.

The OP, who introduced the question of how to model outcomes over time, will protest that the example is too simple and does not clarify things at all. The 2-period poster will respond along the lines of “what part of simple arithmetic did I not understand?” The ensuing dialogue, if it occurs, will satisfy no one, until finally a third party will opine that there are too many future unknowns, just convert up until it feels right.

To advance the discussion, let me suggest that Woodspinner is attempting to model an n-period game, n > 2 (and mostly, > 20). Since he is attempting to model real world results under the US system of taxation, as it is today and as it will become, Woodspinner must model a very particular kind of n-period game. Properties:
1. n can never be known until the game is over
2. Fractional distributions from the TDA of 1/k occur in each of the n periods, and each such distribution is subject to the then prevailing rate of tax
3. The fraction 1/k is set by rule (aka, the RMD schedule), and initial fractions are very small, k about 27; however, fractions are progressive, growing each year up until a limit of k = 2

Once the problem is conceptualized in these terms, the commutative property of multiplication becomes irrelevant, because the no-conversion and the conversion scenarios are not exposed to tax at the same time or over the same n of periods.

A simple n=3 example may help. Appreciation is 10% in each period; all accounts are invested in the same asset; tax is always 24% ordinary, 15% capital gains and dividends.

n=1: Either a conversion of $100,000 is made, and tax at 24% is deducted, creating an initial Roth balance of $76,000; or no conversion occurs, leaving $100,000 in the TDA

n=2: if no conversion, there is now $110,000 in the TDA just before an RMD is taken. That (age 72) RMD will be $4029; tax of $967 is due. The remaining $3062 is invested in a taxable account. End year TDA balance after RMD is $105,971. End year taxable account balance is $3062. The Roth has appreciated to $83,600.

n=3: The TDA grows to $116 568 before the RMD, which at age 73 is $4415; tax of $1060 is due. The remaining $3356 is added to the taxable account. In the meantime, last year’s addition has appreciated by 8.5% after tax. The total and after-tax value of the taxable account is now $6678. The pretax end year TDA balance after RMD is $112,152. The Roth has appreciated to $91,960.

The game ends and after-tax values are computed. The TDA provides $85,236 after tax; with the taxable account, the no-conversion scenario has produced $91,914 of after-tax wealth.

Conclusion: Roth wins. Its $91,960 balance is $46 greater. Even though tax rates were constant.

In this three period game, the Roth “winnings” are paltry, giving a 2 year ROI on the $24,000 tax debit equal to an annualized 0.10%. But after 20 years, the advantage of the Roth will have compounded further, into something quite reasonable. In fact, after a dozen years the Roth advantage will be able to withstand a decline in future tax rates of several percentage points, as when a converter converts themselves out of the 24% bracket into the 22% bracket. After forty years, the Roth advantage may handle a decline of ten points in tax rate.

Finally, the greater the number of periods the game lasts, the greater the Roth advantage. It is compounding at a flat 10% after tax; everything else is compounding at a slower rate. Meantime, as k shrinks and shrinks, more and more of the TDA is lost to tax. Eventually, the Roth balance will exceed even the pretax values of the TDA plus the taxable account.

This is only true if you don't spend your RMD money, right?
The advantage only comes from the taxes due on the earnings of the saved RMDs, versus those dollars growing tax free in the Roth. If one actually spends their RMDs for living expenses, I don't think it matters how many years the conversions or withdrawals are spread over, as long as the rates are the same, the outcome remains the same spendable dollars.

If you agree that this scenario is applicable to the case where the money is not spent, then it is not clear to me why you would tax adjust the taxable account by the capital gains rate. It would pass to the heirs with step up basis.

The only advantage might be a miniscule tax drag from any taxes due on yearly dividends.
Once in a while you get shown the light, in the strangest of places if you look at it right.
User avatar
celia
Posts: 16774
Joined: Sun Mar 09, 2008 6:32 am
Location: SoCal

Re: [**Update] - Metrics to Compare Roth Conversion Scenarios ??

Post by celia »

I’ve noticed a couple of posters here saying that most of their tax-deferred is in bonds, so slower growth is expected there. That’s fine, but you need to remember that the bond rate historically is around 5 or 6%. That’s hard to admit when current rates are so low. So when bonds return to ‘normal’, the TDA may grow more than one is expecting.

I also notice a few people here who have been converting less than the annual TDA growth. The conversions helped slow down the TDA growth, but the accounts still GREW.
A dollar in Roth is worth more than a dollar in a taxable account. A dollar in taxable is worth more than a dollar in a tax-deferred account.
User avatar
FiveK
Posts: 15742
Joined: Sun Mar 16, 2014 2:43 pm

Re: [**Update] - Metrics to Compare Roth Conversion Scenarios ??

Post by FiveK »

McQ wrote: Sun Aug 01, 2021 5:27 pm To advance the discussion, let me suggest that Woodspinner is attempting to model an n-period game, n > 2 (and mostly, > 20). Since he is attempting to model real world results under the US system of taxation, as it is today and as it will become, Woodspinner must model a very particular kind of n-period game. Properties:
1. n can never be known until the game is over
2. Fractional distributions from the TDA of 1/k occur in each of the n periods, and each such distribution is subject to the then prevailing rate of tax
3. The fraction 1/k is set by rule (aka, the RMD schedule), and initial fractions are very small, k about 27; however, fractions are progressive, growing each year up until a limit of k = 2

Once the problem is conceptualized in these terms, the commutative property of multiplication becomes irrelevant, because the no-conversion and the conversion scenarios are not exposed to tax at the same time or over the same n of periods.
There are two additional assumptions that must be specified, lest people arrive at different conclusions because they brought different versions of those assumptions to the table:
4. Whether taxes are
a. paid from conversion funds, or
b. from "cash on hand".
5. Whether RMDs are
a. spent, or
b. invested.

Assuming 4a and 5a is the Simplest situation: the commutative property of multiplication continues to apply and thus remains relevant.

Assuming 4b and 5a is addressed by the first of the More complicated situations described in that wiki article section.

Assumption 5b is indeed a wrinkle not yet covered in the BH Traditional versus Roth article. It would be nice if a simple spreadsheet, such as the two mentioned in the section above that cover moving funds from taxable to Roth, were available to handle the forced movement of funds from traditional to taxable. Don't know if there is a closed form function for the product of [(investment return) - (RMD fraction)] over multiple years, or whether a user-defined function would be needed. That is, to avoid the need for a multi-row spreadsheet that would not be as user-friendly as the ones mentioned above.
User avatar
Topic Author
WoodSpinner
Posts: 3505
Joined: Mon Feb 27, 2017 12:15 pm

Re: [**Update] - Metrics to Compare Roth Conversion Scenarios ??

Post by WoodSpinner »

All,

To see a demonstration of the Tax Drag in action, you can inspect this simple Excel model. The size and impact of the drag will depend on a variety of the variables (in addition to time).

https://drive.google.com/file/d/1RRNghh ... sp=sharing

WoodSpinner
WoodSpinner
diy60
Posts: 972
Joined: Wed Sep 07, 2016 6:54 pm

Re: [**Update] - Metrics to Compare Roth Conversion Scenarios ??

Post by diy60 »

celia wrote: Sun Aug 01, 2021 6:24 pm I also notice a few people here who have been converting less than the annual TDA growth. The conversions helped slow down the TDA growth, but the accounts still GREW.
Keep in mind tax brackets and IRMAA brackets grow as well lessoning the impact of account growth. I often see folks commenting and modeling TDA at 10% without any consideration to brackets ratcheting upward. That would artificially skew results to more Roth conversions. I think Roth accounts have asymmetrical risks, hence I err to too much TDA. Each to their own.
diy60
Posts: 972
Joined: Wed Sep 07, 2016 6:54 pm

Re: [**Update] - Metrics to Compare Roth Conversion Scenarios ??

Post by diy60 »

FiveK wrote: Sun Aug 01, 2021 6:28 pm Assumption 5b is indeed a wrinkle not yet covered in the BH
True enough, but I wonder what is the impact of tax drag due to invested RMDs in the grand scheme of Roth conversions. Is it the 20% that provides 80% of the benefit, or is it the 80% that provides 20% benefit.
User avatar
FiveK
Posts: 15742
Joined: Sun Mar 16, 2014 2:43 pm

Re: [**Update] - Metrics to Compare Roth Conversion Scenarios ??

Post by FiveK »

diy60 wrote: Sun Aug 01, 2021 7:23 pm
FiveK wrote: Sun Aug 01, 2021 6:28 pm Assumption 5b is indeed a wrinkle not yet covered in the BH
True enough, but I wonder what is the impact of tax drag due to invested RMDs in the grand scheme of Roth conversions. Is it the 20% that provides 80% of the benefit, or is it the 80% that provides 20% benefit.
Great question! Of course the answer will be "it depends..." on a bunch of things. For someone with a 0% tax rate on taxable gains, there is no impact at all. For someone paying ~50% on taxable gains (e.g., California resident in the highest bracket with all gains ordinary income), it might be significant. Thus the "It would be nice if a simple spreadsheet..." comment.
marcopolo
Posts: 8446
Joined: Sat Dec 03, 2016 9:22 am

Re: [**Update] - Metrics to Compare Roth Conversion Scenarios ??

Post by marcopolo »

FiveK wrote: Sun Aug 01, 2021 8:29 pm
diy60 wrote: Sun Aug 01, 2021 7:23 pm
FiveK wrote: Sun Aug 01, 2021 6:28 pm Assumption 5b is indeed a wrinkle not yet covered in the BH
True enough, but I wonder what is the impact of tax drag due to invested RMDs in the grand scheme of Roth conversions. Is it the 20% that provides 80% of the benefit, or is it the 80% that provides 20% benefit.
Great question! Of course the answer will be "it depends..." on a bunch of things. For someone with a 0% tax rate on taxable gains, there is no impact at all. For someone paying ~50% on taxable gains (e.g., California resident in the highest bracket with all gains ordinary income), it might be significant. Thus the "It would be nice if a simple spreadsheet..." comment.
Or, if you follow the premise that the RMDs are never spent (thus accruing the compounded tax advantage), then there is little impact as heirs get stepped up basis. The advantage only shows up if you assume you don't spend it, but also assume that you do. :happy
Once in a while you get shown the light, in the strangest of places if you look at it right.
User avatar
FiveK
Posts: 15742
Joined: Sun Mar 16, 2014 2:43 pm

Re: [**Update] - Metrics to Compare Roth Conversion Scenarios ??

Post by FiveK »

marcopolo wrote: Sun Aug 01, 2021 8:32 pm
FiveK wrote: Sun Aug 01, 2021 8:29 pm
diy60 wrote: Sun Aug 01, 2021 7:23 pm
FiveK wrote: Sun Aug 01, 2021 6:28 pm Assumption 5b is indeed a wrinkle not yet covered in the BH
True enough, but I wonder what is the impact of tax drag due to invested RMDs in the grand scheme of Roth conversions. Is it the 20% that provides 80% of the benefit, or is it the 80% that provides 20% benefit.
Great question! Of course the answer will be "it depends..." on a bunch of things. For someone with a 0% tax rate on taxable gains, there is no impact at all. For someone paying ~50% on taxable gains (e.g., California resident in the highest bracket with all gains ordinary income), it might be significant. Thus the "It would be nice if a simple spreadsheet..." comment.
Or, if you follow the premise that the RMDs are never spent (thus accruing the compounded tax advantage), then there is little impact as heirs get stepped up basis. The advantage only shows up if you assume you don't spend it, but also assume that you do. :happy
Yes, that makes it even more complicated: tax drag while you are alive, then no LTCG due to a stepped up basis but a subsequent 10 years of tax drag vs. no tax drag "while you are alive plus 10 years" if there is a post-2019 inherited Roth IRA. And so on.... :)
User avatar
Topic Author
WoodSpinner
Posts: 3505
Joined: Mon Feb 27, 2017 12:15 pm

Re: Metrics to Compare Roth Conversion Scenarios ??

Post by WoodSpinner »

JoinToday wrote: Sat Jul 31, 2021 8:03 pm
WoodSpinner wrote: Sat Jul 31, 2021 6:34 pm .....
2. In the plot above, why isn't the tax adjusted portfolio value the same for all scenarios at age 62? (I assume you are 62, this is t=0; I may be in error). Regardless of what I do in the future, my portfolio has a certain tax adjusted portfolio value today, determined primarily by the tax rate I assume for trad-IRA and LTCG).
.....

Good question, the model is correct (took me a bit to figure it out) and here is what is going on.

With No Conversions, the Marginal Tax Rate is 18% (12% Federal, 6% State), after all the AGI doesn’t have any Roth Conversions.

The other Conversion scenarios have a higher Marginal Rate (31.3%, 33.3%, 33.33%) for IRMAA Tier1-3.

This changes the Tax Adjusted Portfolio Value since the Marginal Rate is used to Discount the Tax Adjusted Value.


3. I may have missed this, but are you considering your heir's tax brackets in your analysis?

I haven’t extended the model to deal with my heirs brackets at this point, it is part of my plans.


WoodSpinner
In one of my spreadsheets, I model my tax adjusted portfolio value differently than what you do. For all scenarios, I discount the traditional IRA by the same tax rate. If I have no roth conversions and a low marginal tax rate for a few years, my portfolio is not worth more than if I were doing large roth conversions with a higher marginal rate. For what it is worth, I assume 25% or 30% of my traditional IRA will be taxes. I also assume either 0% tax rate on after tax account (stepped up basis upon our death) or 10% to accommodate Fed LTCG tax and Calif tax. I think I use the same numbers for my heir's tax rate (traditional IRA only).
Our models do differ but I am not sure it would be worth your time to see if adjusting your approach will shift your strategy. There are a lot of moving parts.

WoodSpinner
WoodSpinner
User avatar
Topic Author
WoodSpinner
Posts: 3505
Joined: Mon Feb 27, 2017 12:15 pm

Re: Metrics to Compare Roth Conversion Scenarios ??

Post by WoodSpinner »

Johnsson wrote: Sun Aug 01, 2021 5:34 am
FiveK wrote: Sat Jul 31, 2021 8:35 pm
WoodSpinner wrote: Sat Jul 31, 2021 7:05 pm
  • Marginal rate (at 62) for No Conversions is 18% (Fed 12%, CA 6%)
  • Marginal Rates (at 62) for IRMAA Tier1-3 are (31.3%, 33.3%, and 33.3%).
You get a year free of IRMAA concerns at age 62 - might be a good year to make a large Roth conversion.
- Once RMDs begin at 72 the Marginal Rates are much closer:
  • Marginal rate (at 72) for No Conversions is 37.3% (Fed 28%, CA 9.3%)
  • Marginal Rates (at 72) for IRMAA Tier1-3 are (37.3%, 36%%, and 33%).
That seems to indicate that pre-RMD (not sure how SS benefits enter here) conversions up through the top of the 24% bracket could be beneficial - especially if taxes are paid with cash on hand. Although, it's probably a relatively "flat optimum," meaning a wide range of conversion amounts provide about the same result. Is that what you see?
I've run many scenarios. I agree there's a wide range of conversions that provide similar results. How much should be converted seems to mostly depend on your mix of pre-tax and after-tax money (more pre-tax $$ = more conversions needed). The main question and impact for me (and I think for most ready to convert now) should be... Pay taxes now or pay later (at some future rate)? I have cash in hand and want to get the largest conversions done before tax rates are currently scheduled to increase. Future rates will certainly impact when conversions should have been done. I don't want to discuss possible future tax rates... not allowed.

While I've enjoyed reading all the discussion, I think most of it is likely moot because of all the future unknowns for each of us. I believe the best answer is determined by modeling many different scenarios and convert to the level that 'feels' right for your situation (especially since there is the "flat optimum" you mention). Precise modeling will be made worthless as life throws changes at us.

DW and I are 60 this year. My current plan is SS at 67 and 70 for us (but may do 70 and 70 with more conversions).

We have about 60% of our funds (2MM) in Pre-tax, with enough in after-tax to pay taxes and get us to SS.

We will be converting to the top of 24% this year and next. Then into 12/15% thereafter, with a total of 1.3MM converted (with all conversions done by age 66, before SS)

We'll be at IRMAA-2 when 65 and 66, then -1 thereafter... the price of conversions.

As discussed previously in this thread, I may cut off the conversions early because the pre-tax will be all-bonds and it's likely to not grow as the Roth will. Still thinking about that one...
My best suggestion is to use your model and iterate (at least yearly) to see if your approach is still optimal. As you suggested, we are dealing with lots of unknowns and they will shift and reveal themselves over time.

Woodspinner
WoodSpinner
User avatar
Topic Author
WoodSpinner
Posts: 3505
Joined: Mon Feb 27, 2017 12:15 pm

Re: [**Update] - Metrics to Compare Roth Conversion Scenarios ??

Post by WoodSpinner »

All,

Just as an FYI I finally had a chance to update the Original Post in this thread to hopefully clarify my thinking and to help bypass some areas of discussion (e.g. RoI and IRR metrics) that weren't particularly insightful.

I do plan to convert to the top of the 24% bracket this year (my last before having to consider IRMAA).

Based on McQuarrie's Rule for Roth Conversions I decided to add another Scenario for Review, Top of IRMAA Tier-0. It turns out to be a really good alternative choice and I have decided to go with that as my approach.

I think this highlights the importance of making these decisions based on modeling and data analysis -- frankly it wasn't even on my radar as an alternative given my situation. Before starting this thread, my plan was to Convert to IRMAA-Tier-2 and this represents a significant improvement to the plan.
  • It will meet my Roth Conversion Goals
  • It will cost less in Taxes than the other alternatives under consideration
  • While it doesn't provide the Highest Tax Adjusted Portfolio, it isn't that far off.
My plan is to revisit this decision (per my Retirement Policy Statement) yearly so I can adjust the plan based on the latest information available.

Many thanks to everyone for the help and insights -- I am much more comfortable now with my decision going forward.

Image

Image

WoodSpinner
Last edited by WoodSpinner on Tue Aug 03, 2021 5:43 pm, edited 2 times in total.
WoodSpinner
DSBH
Posts: 740
Joined: Tue Jul 21, 2020 3:31 pm
Location: Texas

Re: [**Update] - Metrics to Compare Roth Conversion Scenarios ??

Post by DSBH »

WoodSpinner wrote: Sat Jul 24, 2021 8:42 pm Key Learnings:
  • Start with a well defined understanding of your Roth Conversion Goals
  • A robust tool or model is essential for understanding the nuances of your Income, Expenses, Cash-Flow, Taxes, Asset Growth, Asset Location etc. there are a lot of moving parts that need to be considered.
  • Conversion planning needs to be an iterative process. Taxes, Asset Growth, Unexpected Death, Inflation, Health issues etc. can all change your plan. Suggest reviewing the plan at least annually to fine tune and align.
  • The Widow(er) Tax Penalty is real and you need to understand what if any impact this will have on your Retirement Plans.
Excellent list, definitely applicable to me - bookmarked.

2 additional related learnings that I obtained:

1. Open a Roth IRA account even with 1 share of VTSAX (thanks BH David Jay :beer ) if you haven't done so, to start the 5-year clock,
2. If you're in the 12% bracket and even 22% and have a sizeable tax-deferred account, it may be good time spent looking at Roth IRA conversion options between now and 2025, before tax changes are scheduled.
John C. Bogle: "Never confuse genius with luck and a bull market".
User avatar
Topic Author
WoodSpinner
Posts: 3505
Joined: Mon Feb 27, 2017 12:15 pm

Re: [**Update] - Metrics to Compare Roth Conversion Scenarios ??

Post by WoodSpinner »

McQ wrote: Sun Aug 01, 2021 5:27 pm In hopes of advancing the discussion of how to model Roth conversion outcomes, allow me to introduce McQuarrie’s rule:
Professor,

I was actually thinking of another Rule you had postulated earlier (paraphrased)

“When in doubt choose the more conservative Roth Conversion Scenario — the future is uncertain “.

Thoughts?

WoodSpinner
WoodSpinner
User avatar
Lee_WSP
Posts: 10403
Joined: Fri Apr 19, 2019 5:15 pm
Location: Arizona

Re: [**Update] - Metrics to Compare Roth Conversion Scenarios ??

Post by Lee_WSP »

Metrics Considered:
Marginal Tax Rates vs. Future Marginal Rates (Fed + CA State)
This is a general useful metric -- it rarely makes sense to pay Higher Taxes now when you can pay Lower Taxes later. This is especially true if you will be leaving the IRA to beneficiaries in a lower Tax Bracket or to Charities.
Tax Equilibrium across Retirement (see Kitces Blog
This is a very useful metric to consider. My approach to using this metric is to plot the graphs of Total Cumulative Taxes paid over time (Fed, IRMAA, NIT, State etc.) and look at the slopes of the lines and any obvious bend points. This will highlight Conversion scenarios that are wasting lower marginal tax space or paying more in taxes than future income years.
Aren't these two more or less the same?
McQ
Posts: 1425
Joined: Fri Jun 18, 2021 12:21 am
Location: California

Re: [**Update] - Metrics to Compare Roth Conversion Scenarios ??

Post by McQ »

WoodSpinner wrote: Mon Aug 02, 2021 1:30 pm
McQ wrote: Sun Aug 01, 2021 5:27 pm In hopes of advancing the discussion of how to model Roth conversion outcomes, allow me to introduce McQuarrie’s rule:
Professor,

I was actually thinking of another Rule you had postulated earlier (paraphrased)

“When in doubt choose the more conservative Roth Conversion Scenario — the future is uncertain “.

Thoughts?

WoodSpinner
Hello Woodspinner: I stand by that rule, but it seemed too much common sense to think I could put my name on it. Whereas, the predilection toward 2-period simplification is widespread and widely accepted, and yet, for a modeler like you, surprisingly unhelpful (need to follow up later on JoinToday's insight; my 3 period example wasn't, quite).

Returning to the rule that did interest you: here is an elaboration. The future is uncertain, which makes a conservative approach wise, with "conservative" easily defined as the opposite of "all in." But how to operationalize "conservative?"

One simple approach: the higher the tax rate paid on the conversion, the less conservative it is. This rule is fairly widely held among BH. You can see it in people's comfort with converting through the 12% bracket. The same rule is visible when you see BH hesitate about converting at 22%. Now for you and me as professionals in California, 22% seems achingly, beautifully low. I'll never get there, Federal + State, absent some disaster. So if you could convert for Fed+State =22%, or even 25%, that would be low for you. But in Seattle or Austin or Orlando, not so much.

But you were flirting with busting through the 1st, 2nd, and 3rd IRMAA levels, for a combined tax rate of 24% +9.3% +4%, or 37.3% (more, post-TCJA). No matter what the conversion tax rate, you can make the spreadsheet work out, if your assumptions are hit, and especially in bequest mode (+ 10 years). But the higher the initial tax rate, the greater the odds that future uncertainty will cut against you, as when I had your daughter moving to Seattle before receiving her inheritance, or envisioned Peter Thiel inspired excise taxes against Roth balances.

So as an affluent Californian, the max conversion tax rate that qualifies as conservative in my book is 22% + 9.3%, i.e., stopping short of IRMAA #1. And even that only qualifies because you intend to remain in California all your life.
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.
RetiredCSProf
Posts: 1228
Joined: Tue Feb 28, 2017 3:59 pm

Re: [**Update] - Metrics to Compare Roth Conversion Scenarios ??

Post by RetiredCSProf »

McQ wrote: Mon Aug 02, 2021 3:35 pm
So as an affluent Californian, the max conversion tax rate that qualifies as conservative in my book is 22% + 9.3%, i.e., stopping short of IRMAA #1. And even that only qualifies because you intend to remain in California all your life.
Should I assume this characterization applies only to Californians filing as MFJ? Pension alone pushes me to the top of the federal 22% bracket for HoH or single filing.

It would be great to have models, measures, or rules that work universally, regardless of filing status. I think the "tax equilibrium" concept helps with that.

Also, how do you "stop short" of any IRMAA bracket, given the 2-year look-back and the annual CPI adjustment of IRMAA brackets? And then there was that year when one IRMAA bracket split itself in two. Can that happen again?
McQ
Posts: 1425
Joined: Fri Jun 18, 2021 12:21 am
Location: California

Re: [**Update] - Metrics to Compare Roth Conversion Scenarios ??

Post by McQ »

RetiredCSProf wrote: Mon Aug 02, 2021 4:43 pm
McQ wrote: Mon Aug 02, 2021 3:35 pm
So as an affluent Californian, the max conversion tax rate that qualifies as conservative in my book is 22% + 9.3%, i.e., stopping short of IRMAA #1. And even that only qualifies because you intend to remain in California all your life.
Should I assume this characterization applies only to Californians filing as MFJ? Pension alone pushes me to the top of the federal 22% bracket for HoH or single filing.

It would be great to have models, measures, or rules that work universally, regardless of filing status. I think the "tax equilibrium" concept helps with that.

Also, how do you "stop short" of any IRMAA bracket, given the 2-year look-back and the annual CPI adjustment of IRMAA brackets? And then there was that year when one IRMAA bracket split itself in two. Can that happen again?
Fair questions all, here is a stab at some answers:
1. Yes, I had Californians MFJ in mind, since at $175,999 AGI in 2021, just under IRMAA #1, they pay 31.3%, my ceiling for conservative-in-that-situation. Conversely, I would discourage a Texan couple from converting at 32% Federal-I don't believe that is low in that context. And if your personal projection is 24% once widowed, because pension (COLA'd??) puts you at the top of 22%, I don't think 22% is conservative for you today--unless you can project that you will fall into IRMAA as a widow/er. Then, 22% today isn't quite conservative, but it is not unconservative.

2. There's a Catch-22 on "universal." A better mathematician than me could set up a general model for conversion "odds;" but it would be a black box spitting out results impossible to vet. Any model that you could inspect, and confirm was relevant to your particular situation, would probably not be universal.

3. Re stopping short of IRMAA: today IRMAA #1 falls near the top of the AGI that keeps you in the 22% bracket. The income bracket adjusts for inflation each year and the IRMAA brackets also adjust. So IRMAA #1 is always going to be near the top of the 22% bracket--until Congress changes the rules, which of course, they have done repeatedly.

...which brings us back to the importance of being conservative about how much you convert today.
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.
McQ
Posts: 1425
Joined: Fri Jun 18, 2021 12:21 am
Location: California

Re: [**Update] - Metrics to Compare Roth Conversion Scenarios ??

Post by McQ »

JoinToday wrote: Sun Aug 01, 2021 6:11 pm
The $46 is the tax drag on the RMD put in the after tax account (10% gain x 15% tax rate x $3062) [from what I can see]. This has led me to focus on reducing the amount of money in my after tax account. Pay tax from after tax account if at all possible, balancing the LTCG taxes with the yearly tax drag of money in my after tax account.

I don't think your example negates the commutative property that others have mentioned, and you mention in your post.

For me, the biggest drivers have been (and still are):
1. Tax rates, current vs future
2. reducing the number of years (= $$) I spend in the higher IRMAA brackets
3. With lower income in the future, my LTCG will drop to 0% for a portion of my dividends
4. Tax drag (as you demonstrated in your example above).
Thank you to FiveK and others who responded so quickly. My learning continues: I see now from JoinToday’s response that I failed to construct a game that was multi-period in all aspects. Let me continue the game to n=4. Apologies for the slew of numbers—if only the BH implementation of BBC code permitted HTML tables!

N=4. The TDA grows to $123, 368 before the RMD, which at age 74 is $4838; tax of $1161 is due. The remaining $3677 is added to the taxable account. In the meantime the prior two additions have appreciated 8.5% after tax. The total and after-tax value of the taxable account is now $10,987. The end year TDA balance is $118,530. Applying tax of 24% to it, we have $90,082.50, or $100,980 in all, after-tax, no conversion.

But the Roth is worth $101,156, giving it an advantage now of $176.25. That’s more than the tax on capital gain & dividend this period ($95.67), plus the future value of the tax last period ($49.84); $30.74 excess for the Roth in all, relative to tax drag on the reinvested RMDs alone.

I haven’t succeeded yet in decomposing that $30.74 algebraically; but the logical place to look is the other source of tax drag in this multiperiod game: the $967 tax on the first RMD, lost to the government rather than reinvested.

In my first stab at n=3, the reinvested RMDs were a 2-period piece, hence the confusion. Apologies.

JoinToday, to your particular point: Tax rates now and future are certainly important if the gap is big, i.e., 12 to 22%, or 24% to 32%. But if the gap is small (22% to 24%, 24% to post-TCJA 28%), I find that tax drag, over longer periods, outweighs it.

And I am all for staying out of future IRMAA, generally the equivalent of +4.5% in marginal rate--if you take income to the exact top of that IRMAA bracket. Tens of thousands of percent on the first few dollars, if you don't.
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
Topic Author
WoodSpinner
Posts: 3505
Joined: Mon Feb 27, 2017 12:15 pm

Re: [**Update] - Metrics to Compare Roth Conversion Scenarios ??

Post by WoodSpinner »

Not sure if anyone took a look at the Tax Drag Example spreadsheet posted upthread. Reposting to provide support for McQ that the drag is real and will be more noticeable over time.
WoodSpinner wrote: Sun Aug 01, 2021 6:50 pm To see a demonstration of the Tax Drag in action, you can inspect this simple Excel model. The size and impact of the drag will depend on a variety of the variables (in addition to time).

https://drive.google.com/file/d/1RRNghh ... sp=sharing

WoodSpinner
WoodSpinner
User avatar
FiveK
Posts: 15742
Joined: Sun Mar 16, 2014 2:43 pm

Re: [**Update] - Metrics to Compare Roth Conversion Scenarios ??

Post by FiveK »

WoodSpinner wrote: Tue Aug 03, 2021 5:09 pm Not sure if anyone took a look at the Tax Drag Example spreadsheet posted upthread. Reposting to provide support for McQ that the drag is real and will be more noticeable over time.
WoodSpinner wrote: Sun Aug 01, 2021 6:50 pm To see a demonstration of the Tax Drag in action, you can inspect this simple Excel model. The size and impact of the drag will depend on a variety of the variables (in addition to time).

https://drive.google.com/file/d/1RRNghh ... sp=sharing

WoodSpinner
Looked at it but it wasn't "intuitively obvious" how to use it. Instructions?
Post Reply