Roth conversion confusion

[Magus]

Not sure I totally understand your comment.

In essence, when you put away income in a traditional pre-tax account and take the deduction now, you are saving at the marginal tax rate. If you are in the 35% tax bracket, you save paying 35 cents for every dollar you defer.

Later, when withdrawing funds, you pay taxes at your effective tax rate which in my case is half of my marginal rate.

It’s really marginal both ways but when you are working the money you are deducting is at the top if the brackets while making money in retirement it’s generally starting from the bottom of the brackets (0% rate for $26k for married for example). If you have no pension or no significant non retirement assets your approach works but if you have other income you can’t just use effective rate in retirement.

In general, over 95% of folks have a lower effective rate in retirement than while working so Roths are way overrated but there are times they work (inherit a ton, investments did phenomenally well beyond market, pension that’s over half your base income). Most people also forget to adjust tax brackets and std deduction in retirement for inflation and massively overestimate what they will owe.

I should note you’d need about $300k in retirement income (married couple) right now to have a 16% effective fed tax rate in retirement - and thats assuming its not LT Capital gains income. Very few people have that income period, much less retirement. Even if you are modeling $300k in say 20 years, keep in mind with inflation you’ll be about the equiv of $180k income today and closer to 10-11% effective rate.

 

[RunningBum]

Figure out your tax, in dollars. Now add $100 of income, from a conversion or tIRA withdrawal. Figure out your tax with that extra $100. The difference between the two is your marginal rate. It is not only the tax bracket you're in, but it may include pushing another $100 of cap gains or $85 of SS to be taxed. It may also include the loss of 8.5% of ACA subsidy, or a new IRMAA tier.

This is what matters, the impact of that extra $100 or however much you want to convert now or withdraw later, not the overall rate you had.

A simple example. Suppose you have low enough income or high enough deductions to owe $0 in taxes. That doesn't mean that no matter how much more income you add you will still owe $0 in taxes. As you incrementally add income to the equation, you will see where you start incurring taxes.

 

[SecondCor521]

question for you Roth Conversion experts...when converting to the top of the tax bracket do you use effective tax rate % or tax bracket % to decide whether or not to do?

For instance, I am in 35% tax bracket but after deductions, donations, etc my effective tax rate is closer to 16%. So, would I convert up to the top of the 22% bracket or up to top of 35% bracket?

Not an expert but I've read a fair amount and thought a fair amount.

I look at marginal rate now vs. marginal rate at age 75 to decide how much to convert.

Basically I am considering two scenarios:

1. Income of $X + $1 this year and income of $Y at age 75.
2. Income of $X this year and income of $Y + $1 at age 75.

For me, it looks like that additional dollar at age 75 will cost me about 33 cents in taxes. So if I can convert it today at less than that, I do; if not, I don't. And for that analysis, it's the marginal tax rate on that dollar both now and later that matters.

But once I make the decision to convert an additional dollar, I can repeat the analysis with another dollar. And another. I keep converting dollars until I hit a dollar that would cost me 34 cents or more to convert, and that's where I stop.

There can be so-called "tax humps" where some dollars might need to be converted at, say, 50 cents of tax cost, but past those dollars there can be more dollars converted at, say, 22 cents of tax cost. In this case, one can make an aggregate decision based on the height and width of the tax hump as well as the width of the "tax valley" beyond to see if it makes sense to convert a collection of dollars. In my tax situation I don't think I have any or will have any, but they definitely exist for some people and it's in my tax notes to consider the possibility.

(There are other factors other than the tax arbitrage of course.)

 

[Gumby]

I have never found effective tax rate to be a particularly useful number. Like RunningBum, I am interested in the marginal effect that a particular action has on my total tax burden. I am beyond 0% capital gains territory, already have 85% of my social security taxed, don't get an ACA subsidy and will be reimbursed for any IRMAA surcharges charges if I ever incur them, so for me, it's pretty much only my marginal rate that matters. If I convert now, my marginal taxes are at 22% of the conversion. I never contributed any of the money in my tIRA at a marginal rate less than 24%, so I'm money ahead no matter what.

But the real question is whether I convert now and pay the tax at 22% or don't convert and potentially pay 24% or higher later when I hit RMDs. My own spreadsheet shows that if all the tax brackets stay as current and inflation adjust together, then the RMDs will never drive us into the 24% bracket (although we will hit NIIT eventually because it has no inflation provision) even if I convert nothing between now and then. It also tells me that in the likely event the young wife survives me, she will be in the 24% bracket even if I convert all the way up to the top of the 22% bracket every year between now and RMDs, but she will never get to the 32% bracket even if I do not convert a thing.

In light of that, I will Roth convert now only if and when I need funds from my tIRA (unlikely) and, later, just take QCDs and RMDs and pay the taxes then. Of course, our investment returns could be much greater than I assumed in my spreadsheet (4% real) and that could drive us into higher brackets later, but that would be a good problem to have and I won't mind sharing a little more of the wealth with Uncle Sam if it occurs.

 

[pb4uski]

Roth gives your heirs another decade of tax-free growth...

[-]And then a big-a$$ tax bill in that 10th year all else being equal.[/-] See post#38

 

[RockyMtn]

Thanks for the insight folks. Very helpful! While I'm still collecting consulting income and have significant dividends, pension, etc I guess I will forego the Roth conversions. When consulting retainer ends maybe I will reconsider.

 

[RunningBum]

Gumby, any thoughts about the sunset provision on the lower tax brackets? 22%/24% will revert back to 25%/28% in 2026 unless congress takes action. That would be a tiebreaker for me to convert if it looks like a wash keeping the current tax rates. Not a significant effect, but it may be something.

 

[CRLLS]

And then a big-a$$ tax bill in that 10th year all else being equal.

Please explain. I was under the impression that all (non-spousal) inherited Roth withdrawals are tax free. Is that not true? I also think they must be withdrawn totally within that 10-year period. If not, there are penalized at a very high rate, right?

Any interest earned on monies invested from a withdrawal are taxed but not the principal. I'd like to know if I am off base.

 

[SecondCor521]

Please explain. I was under the impression that all (non-spousal) inherited Roth withdrawals are tax free. Is that not true? I also think they must be withdrawn totally within that 10-year period. If not, there are penalized at a very high rate, right?

Any interest earned on monies invested from a withdrawal are taxed but not the principal. I'd like to know if I am off base.

I think @pb4uski missed the fact that it's a Roth.

Yes, in general Roth IRAs also follow the 10 year SECURE distribution period rules.

In general, all the distributions (both principal and earnings) from the Roth are going to be tax free. There may be exceptions if the original owner established the Roth within the five years prior to death.

 

[CRLLS]

I think @pb4uski missed the fact that it's a Roth.

Yes, in general Roth IRAs also follow the 10 year SECURE distribution period rules.

In general, all the distributions (both principal and earnings) from the Roth are going to be tax free. There may be exceptions if the original owner established the Roth within the five years prior to death.

That was my thought. I wanted to make sure that either I wasn't missing something or 2) others didn't take his statement to heart. Let's see what he says.

 

[Gumby]

Gumby, any thoughts about the sunset provision on the lower tax brackets? 22%/24% will revert back to 25%/28% in 2026 unless congress takes action. That would be a tiebreaker for me to convert if it looks like a wash keeping the current tax rates. Not a significant effect, but it may be something.

If the brackets stay the same and continue to inflation adjust the same, but the rate just increases, we would stay in the 25% bracket though RMDs and the young wife would be in the 28% bracket after I go. The only benefit would be the ability to take money at 22% for next 4 years (tax years 2022-25) instead of 25% or 28%. Currently, the 22% bracket runs from $83,550 to $178,150 (MFJ). Assume I can use the entire $94,600 for Roth conversion, then I save $2838 in taxes by taking that money this year instead of later. Or about $12k over those 4 tax years until the rate goes up in 2026. In the grand scheme of things, that's nothing to write home about. And it will eat up about $84,000 in my after tax cash to pay the taxes on those conversions (Roth conversions are substantially less beneficial if you have to use part of the tIRA withdrawal to pay the taxes.)

 

[pb4uski]

What your marginal tax rate/tax savings were when you deferred that income is interesting but not relevant because it is a sunk decision... nothing that you can do about it now.

So now, the more relevant things are the cost of doing a Roth conversion now vs the cost of an RMD later. When RMD time comes along and you have income from taxable savings, pensions, SS, etc, what do you expect your marginal tax rate to be (before considering the RMD)? Add in your RMD and compare the increase in tax to the RMD and that is your tax rate on the RMD. In our case, without the RMD we'll be near the top of the 12% bracket so some of the RMD will be at 12% but most of it will be at 22% and it will average out to about 17%.

The cost of a Roth conversion now is the incremental tax paid and any IRMAA costs, etc.
In our case because I haven't yet begun SS our taxes absent a Roth conversion would be nil. The cost of Roth conversions to the top of the 12% tax bracket is about 11.5%... a blend of the 10% and 12% tax brackets.

So I figure that I'm saving about 5.5% compared to what I will pay later and that is conservative, because if the RMDs happen after one of us dies or is inherited by my kids the tax rate on the RMDs is likely to be much higher than 17%. And then there is always the prospect of higher tax rates in the future.

 

[pb4uski]

I think @pb4uski missed the fact that it's a Roth.

Yes, in general Roth IRAs also follow the 10 year SECURE distribution period rules.

In general, all the distributions (both principal and earnings) from the Roth are going to be tax free. There may be exceptions if the original owner established the Roth within the five years prior to death.

That was my thought. I wanted to make sure that either I wasn't missing something or 2) others didn't take his statement to heart. Let's see what he says.

Yes, you are correct... I didn't realize that the IRA being referred to was a Roth and thought it was a traditional IRA with no tax basis.

 

[Gumby]

What your marginal tax rate/tax savings were when you deferred that income is interesting but not relevant because it is a sunk decision... nothing that you can do about it now.

So now, the more relevant things are the cost of doing a Roth conversion now vs the cost of an RMD later. When RMD time comes along and you have income from taxable savings, pensions, SS, etc, what do you expect your marginal tax rate to be (before considering the RMD)? Add in your RMD and compare the increase in tax to the RMD and that is your tax rate on the RMD. In our case, without the RMD we'll be near the top of the 12% bracket so some of the RMD will be at 12% but most of it will be at 22% and it will average out to about 17%.

The cost of a Roth conversion now is the incremental tax paid and any IRMAA costs, etc.
In our case because I haven't yet begun SS our taxes absent a Roth conversion would be nil. The cost of Roth conversions to the top of the 12% tax bracket is about 11.5%... a blend of the 10% and 12% tax brackets.

So I figure that I'm saving about 5.5% compared to what I will pay later and that is conservative, because if the RMDs happen after one of us dies or is inherited by my kids the tax rate on the RMDs is likely to be much higher than 17%. And then there is always the prospect of higher tax rates in the future.

Conversion to the top of the 12% bracket should be a no-brainer for everyone. Sadly, I'm already there without a conversion. (I guess it's not that sad, since it means we have decent income.)

 

[Exchme]

Wow. I got Pralana Gold and it is amazing. It basically confirmed my intuitive conclusion that the pay not or pay later washes out. If you live a few years the end results are almost identical including total estate and total taxes paid. The difference is the mix between tax differed, taxable, and Roths.

I plan to run a lot more simulations with this and then talk to both a CPA and a fee only FA to discuss whether there are gotchas I am missing. But big conversion(s) near-term are probably in our future.

Thanks for the recommendation.

You may be looking at "absolute" dollars, meaning treating a $ in tax deferred as equal to a $ in Roth or taxable. That's not usually a good way to value tax deferred. There is an option the Roth conversion page, under the graph to switch to "effective" $ where it applies a tax rate you input to tax deferred. You input that "effective" $ conversion on the Financial Assets - Management page and a typical value would be what you expect your heirs to pay on the conversion of the inherited IRA.

Once you switch to "effective" $ on the Roth page, all the outputs, including the tabular ones, switch to "effective" $.

 

[donheff]

Of course, our investment returns could be much greater than I assumed in my spreadsheet (4% real) and that could drive us into higher brackets later, but that would be a good problem to have and I won't mind sharing a little more of the wealth with Uncle Sam if it occurs.

Off topic a bit Gumby but I have been toggling back and forth between 4% and 6% to see the impact. What do you use for bond real returns?

 

[donheff]

You may be looking at "absolute" dollars, meaning treating a $ in tax deferred as equal to a $ in Roth or taxable. That's not usually a good way to value tax deferred. There is an option the Roth conversion page, under the graph to switch to "effective" $ where it applies a tax rate you input to tax deferred. You input that "effective" $ conversion on the Financial Assets - Management page and a typical value would be what you expect your heirs to pay on the conversion of the inherited IRA.

Once you switch to "effective" $ on the Roth page, all the outputs, including the tabular ones, switch to "effective" $.

Thanks. I have been using effective dollars to evaluate the value for heirs. So far it looks like some large conversions over the next few years pays off OK if I fund most the taxes for the conversions from taxable accounts.

If you use Pralana, maybe you can clarify my understanding of that setting for estimated tax rate for Roth conversions. As I understand the tabulation tables, that setting simply directly reduces the displayed value of the tax differed IRA accounts by the amount specified (24% was the default) to give a more realistic impression of what the inherited IRAs will be worth to heirs. That seems reasonable for most inheritors (they will be in lower marginal Fed brackets but the 10 year window to unload the inheritance could bump them right up if it is big). Also, if they are already in a high Fed bracket (e.g. my son) it may warrant cranking the default up to more realistically reflect the value of the tax differed portion. And in all cases, once you get the Fed rate right you need to add any state income tax that will be imposed on the IRA withdrawals. Does that sound about right?

It seems key to get that "effective rate" discount right because it affects how the algorithms value your conversion savings/losses. In other words, the wash I am projecting only develops to the extend I leave IRA to the kids (as opposed to charities) and to their marginal tax rates at that time. Charitable giving is a whole other factor to keep in mind.

Does it sound like I am ready this right?

 

[RunningBum]

Assume I can use the entire $94,600 for Roth conversion, then I save $2838 in taxes by taking that money this year instead of later. Or about $12k over those 4 tax years until the rate goes up in 2026. In the grand scheme of things, that's nothing to write home about. And it will eat up about $84,000 in my after tax cash to pay the taxes on those conversions (Roth conversions are substantially less beneficial if you have to use part of the tIRA withdrawal to pay the taxes.)

That reminds me of another point in favor of conversions. You used up $84K in after tax cash, but you made over $375K available to you with no tax consequence. If you ever have a need for a large sum for some reason, it's better to have that money available in a Roth that you built up with planned tax payments rather than having to dip heavily into your tIRA with a large tax consequence.

Another tie breaker to convert, in my opinion.

Edit: That's just meant for general consideration, nothing specific to Gumby's, other than that's what prompted my though.

 

[Exchme]

Thanks. I have been using effective dollars to evaluate the value for heirs. So far it looks like some large conversions over the next few years pays off OK if I fund most the taxes for the conversions from taxable accounts.

If you use Pralana, maybe you can clarify my understanding of that setting for estimated tax rate for Roth conversions. As I understand the tabulation tables, that setting simply directly reduces the displayed value of the tax differed IRA accounts by the amount specified (24% was the default) to give a more realistic impression of what the inherited IRAs will be worth to heirs. That seems reasonable for most inheritors (they will be in lower marginal Fed brackets but the 10 year window to unload the inheritance could bump them right up if it is big). Also, if they are already in a high Fed bracket (e.g. my son) it may warrant cranking the default up to more realistically reflect the value of the tax differed portion. And in all cases, once you get the Fed rate right you need to add any state income tax that will be imposed on the IRA withdrawals. Does that sound about right?

It seems key to get that "effective rate" discount right because it affects how the algorithms value your conversion savings/losses. In other words, the wash I am projecting only develops to the extend I leave IRA to the kids (as opposed to charities) and to their marginal tax rates at that time. Charitable giving is a whole other factor to keep in mind.

Does it sound like I am ready this right?

Yes, that's a good summary, I would crank up the effective rate to match the expected marginal rates your heirs will see on the IRA they inherit from you.

When I make a change like that, I like to go to the Roth Conversion input page, click Disable Roth Conversions, then go to the Run Analysis page and click Update Active Analysis. Then return to the Roth page and click Enable Roth Conversions again. Then the graph and all the changes you are studying are up to date.

 

[Gumby]

Off topic a bit Gumby but I have been toggling back and forth between 4% and 6% to see the impact. What do you use for bond real returns?

By design, I currently have almost the exact same 65/35 asset allocation in each type of account (tax deferred and Roth). So I use the same 4% real return on a blended portfolio for each. I also assumed than RMD's will be put into the same funds with the same allocation in my taxable account. That is not 100% tax optimal (which would be to keep all fixed income out of taxable accounts), but it makes the analysis much simpler. I also assume that all subsequent gains in my taxable account after RMDs start will be ordinary income versus some capital gains. So it overstates the tax burden in the later years of RMDs and makes the best possible case for conversions now.

That reminds me of another point in favor of conversions. You used up $84K in after tax cash, but you made over $375K available to you with no tax consequence. If you ever have a need for a large sum for some reason, it's better to have that money available in a Roth that you built up with planned tax payments rather than having to dip heavily into your tIRA with a large tax consequence.

Another tie breaker to convert, in my opinion.

Edit: That's just meant for general consideration, nothing specific to Gumby's, other than that's what prompted my though.

That is a good point and I did consider it. Although it depends on how low you want to run your after tax account before RMD's kick in and start to refill it. My after tax money is currently either I-Bonds or cash. If I fully converted for the next eight years, it would run my after tax cash position very low, and I would have to stop buying I-Bonds. Our Roth accounts are already $300k+, so we have a fairly large slug of tax free money if we really need it.

 

[Exchme]

By design, I currently have almost the exact same 65/35 asset allocation in each type of account (tax deferred and Roth). So I use the same 4% real return on a blended portfolio for each. I also assumed than RMD's will be put into the same funds with the same allocation in my taxable account. That is not 100% tax optimal (which would be to keep all fixed income out of taxable accounts), but it makes the analysis much simpler. I also assume that all subsequent gains in my taxable account after RMDs start will be ordinary income versus some capital gains. So it overstates the tax burden in the later years of RMDs and makes the best possible case for conversions now.

FWIW, Pralana Gold takes care of all the issues you listed. It lets you input the return of each asset class, allows you to do tax efficient asset placement (e.g. preferentially putting bonds in tax deferred) and allows you to input the % LTCG/Qualified Dividends and % Ordinary dividends/ST gains.

The main thing you have to be careful of is that capital gains from asset sales this year show up in your AGI and taxes for next year (otherwise the program would be unusably slow and possibly unstable). So if you are searching for an ACA subsidy, you have to examine the capital gains to make sure you aren't fooling yourself.

 

[Gumby]

FWIW, Pralana Gold takes care of all the issues you listed. It lets you input the return of each asset class, allows you to do tax efficient asset placement (e.g. preferentially putting bonds in tax deferred) and allows you to input the % LTCG/Qualified Dividends and % Ordinary dividends/ST gains.

The main thing you have to be careful of is that capital gains from asset sales this year show up in your AGI and taxes for next year (otherwise the program would be unusably slow and possibly unstable). So if you are searching for an ACA subsidy, you have to examine the capital gains to make sure you aren't fooling yourself.

Pralana sounds like it would be of great use to many people on this board, as the same questions arise regularly. I could have made my own spreadsheet more granular, but I started from scratch and got tired, so I made some simplifications. I am fairly certain I know which way those simplifications tilt the results. While certainly cruder than Pralana, I think it gave me sufficient information to draw a conclusion.

 

[target2019]

Not an expert but I've read a fair amount and thought a fair amount.

I look at marginal rate now vs. marginal rate at age 75 to decide how much to convert.

Basically I am considering two scenarios:

1. Income of $X + $1 this year and income of $Y at age 75.
2. Income of $X this year and income of $Y + $1 at age 75.

For me, it looks like that additional dollar at age 75 will cost me about 33 cents in taxes. So if I can convert it today at less than that, I do; if not, I don't. And for that analysis, it's the marginal tax rate on that dollar both now and later that matters.

But once I make the decision to convert an additional dollar, I can repeat the analysis with another dollar. And another. I keep converting dollars until I hit a dollar that would cost me 34 cents or more to convert, and that's where I stop.

There can be so-called "tax humps" where some dollars might need to be converted at, say, 50 cents of tax cost, but past those dollars there can be more dollars converted at, say, 22 cents of tax cost. In this case, one can make an aggregate decision based on the height and width of the tax hump as well as the width of the "tax valley" beyond to see if it makes sense to convert a collection of dollars. In my tax situation I don't think I have any or will have any, but they definitely exist for some people and it's in my tax notes to consider the possibility.

(There are other factors other than the tax arbitrage of course.)

I like this approach. Eventually you have to get down to the mechanics of tax filing, and can simply type in a conversion number to see how much additional tax you might pay for next year's conversion.

You do have to examine all your inputs, as next year you may have different 1099's and so on.

 

[SecondCor521]

I like this approach. Eventually you have to get down to the mechanics of tax filing, and can simply type in a conversion number to see how much additional tax you might pay for next year's conversion.

You do have to examine all your inputs, as next year you may have different 1099's and so on.

Yup. In fact that's what I do.

I'm a volunteer tax preparer, so I get access to next year's tax software for free in mid-December. I can then go in and do my tax return in the software and plug in different conversion amounts to get it exactly the way I want.

This way, the software automatically checks for phase-outs, credits, and other stuff which is in play in my tax life that year.

I then make the appropriate Roth conversions in late December.

Lather, rinse, repeat the next year.

 

[AvocadoRV]

Bogleheads Retirement Portfolio Calculator

Bogleheads has a very detailed retirement portfolio calculator that supports modeling for Roth Conversions with numerous assumptions, and provides details on tax savings to the individual and potential savings for heirs.



The Bogleheads page is here:
https://www.bogleheads.org/forum/viewtopic.php?t=97352


and the link to the spreadsheet is here:
https://www.dropbox.com/s/yf1aqu9eqxys1ft/Retiree Portfolio Model v22.3.xlsm?dl=0


Don't get frustrated by the amount of entries, just take your time, and continue to refine as your understanding of the operation of the spreadsheet increases.



I have found this spreadsheet to be extremely useful on an annual basis to estimate if and how much to convert.