Trad to Roth conversion - more or less

[boarder42]

As we near the withdrawal stages I am digging deeper into how much we should convert.  I was planning to go all the way to the top of the 12% bracket which may make sense in a down market but I don't think it makes sense long term. 

Top of 12% ~106k currently

ACA Cost - 5607
Tax Cost - 6500
Total Cost to covert - 12,107

vs

90k

ACA Cost - 2979
Tax Cost - 3712
Total cost to covert - 6691

Savings converting 16k less 5416

so the out of pocket cost to convert 16k more is 5416 or 33%

if we go down further to 81.5k conversion we get

ACA Cost - 1707
Tax Cost - 2824
Total cost to convert - 4531

so the out of pocket cost to convert 8500 more is 2160 or 25%

i really wouldn't feel comfortable going lower than the 81500 conversion plus we have to keep that level or our healthcare costs go up due to kids going on CHIP. 

starting to think we should only convert enough to keep the kids off CHIP long term i can always pay 31% to convert money including fed and state taxes.  it also lowers our expenditures in the first 5-10 years which will pay off more long term in theory. 

we will have exhausted current roth contributions by age 40 or 5 years in, as well as our HSA money we can freely tap due to existing receipts.  So future monies will likely come from this conversion and either sale of taxable assets or margin borrowed against taxable assets. 

Need to find someone who will lend me money against my IRA assets

[seattlecyclone]

Need to find someone who will lend me money against my IRA assets

Yeah...that's a prohibited transaction. Don't do that.

[seattlecyclone]

But back to your main question, I think you've definitely hit on the fact that ACA tax credit phase-outs act as a pretty high add-on to your regular tax bracket. There's something to be said for realizing as much income as possible in years when you're not on ACA marketplace insurance, doing more Roth contributions/conversions before FIRE and/or waiting until you're Medicare age. Your current marginal rates of 25-33%, you can realize a ton of income in a lower bracket than that when you don't have ACA phase-outs to worry about.

[boarder42]

But back to your main question, I think you've definitely hit on the fact that ACA tax credit phase-outs act as a pretty high add-on to your regular tax bracket. There's something to be said for realizing as much income as possible in years when you're not on ACA marketplace insurance, doing more Roth contributions/conversions before FIRE and/or waiting until you're Medicare age. Your current marginal rates of 25-33%, you can realize a ton of income in a lower bracket than that when you don't have ACA phase-outs to worry about.

My marginal rate is 28% today. Realizing today also takes money out of compounding for you as well. I also expect healthcare not insurance but care to get cheaper over 25 years thru tech and someone finally doing something to fix why we overpay for bad services so much. And a public option to become available. It will likely be AGI based but the trend is more credits. And possibly hsa being allowed to purchase it

[seattlecyclone]

Realizing today also takes money out of compounding for you as well.

Not really. Paying 25% now and then letting the rest compound is no different from compounding the whole amount and then paying 25% later. Either way you end up keeping 75% of what you would if taxes didn't exist.

I also expect healthcare not insurance but care to get cheaper over 25 years thru tech and someone finally doing something to fix why we overpay for bad services so much. And a public option to become available. It will likely be AGI based but the trend is more credits. And possibly hsa being allowed to purchase it

These might all be good reasons to hold off on recognizing unnecessary income before those reforms come to pass.

[boarder42]

Yes but if I don't need 20k in 5 years why roll it today as it stops some of that from compounding due to going for the govt. Keeping as much money in my account as possible today helps whether downturns longer right?   Bc that tax is flowing out today and not compounding for me anymore.

[seattlecyclone]

Yes but if I don't need 20k in 5 years why roll it today as it stops some of that from compounding due to going for the govt. Keeping as much money in my account as possible today helps whether downturns longer right?   Bc that tax is flowing out today and not compounding for me anymore.

Part of your traditional IRA is already earmarked for the government. They give you the choice of giving them their share now or later. Your share compounds the same either way. Who cares whether their share compounds or not?

[boarder42]

Yes but if I don't need 20k in 5 years why roll it today as it stops some of that from compounding due to going for the govt. Keeping as much money in my account as possible today helps whether downturns longer right?   Bc that tax is flowing out today and not compounding for me anymore.

Part of your traditional IRA is already earmarked for the government. They give you the choice of giving them their share now or later. Your share compounds the same either way. Who cares whether their share compounds or not?

In withdrawal it matters to some level right. Money is cut from 1 million to 500k. That 80k I move out at 500k is more valuable to me than them. So if you move out more earlier it hurts sorr. You're assuming it's all equal when money comes out which it never is. Long term and withdrawals aside I agree with the simplistic equation of whichever is cheaper.

[seattlecyclone]

I don't really follow what you're saying. Are you saying that if the market dropped you'd tighten your belt and your tax rate would be lower than you originally planned, which would make traditional look like the better choice in hindsight?

[Mr. Green]

This is a complicated problem. I have spent lots of energy considering it and what could be the best path forward, as we are now in the beginning stages of this process.

If you are simply looking to pay as little tax as possible now, a blend of income from taxable accounts (long-term capital gains) and Roth IRA conversions is best. In 2021, married folks can convert $25,100 from tIRA to Roth tax-free. If you have children, that really juices the amount you can convert by allowing you to offset your federal taxes with the child tax credit. If planned for correctly, in 2021 a married couple could have an income of $106,150 consisting solely of Roth conversions and long-term capital gains and pay zero federal tax. If you have enough kids I suppose it could be even more, especially with the newly expanded child tax credit.

However, there's a long-term angle to consider as well. If your tIRA balances are high enough (ours are), solely focusing on minimizing tax burden now may lead to tIRA balances that are so large by the time you hit Required Minimum Distributions that you're paying significantly higher taxes on an income in old age that is so large you likely can't spend it all because your mobility and health are declining. This aspect is especially worth considering if you knew balancing current and future taxes would result in a larger income now that you could enjoy while younger and healthier. The obvious caveat here is that we don't know exactly what future taxes will be so this optimization won't be perfect.

The ACA "tax" will happen no matter what. The only way you can limit that is by either choosing a cheaper bronze plan so that the loss of premium tax credits is less or relocating to another place where health insurance premiums are lower overall. There are potential drawbacks to both of these though. You may not want to move, or someone in your family may have health problems that make a bronze plan more expensive than a silver plan after considering the cost of care in addition to insurance premiums.

As to @boarder42's comment about leaving money untaxed to lessen the impact of SORR, this really depends on whether you need all those dollars, which is what @seattlecyclone was getting at. If you need to spend 80k per year it doesn't matter whether your portfolio is 1 million or 500k. However, if you can spend less than 80k (belt-tightening), or if you were simply creating an 80k income in the name of tax efficiency and aren't actually spending all that money each year, then having a larger portfolio balance is going to be better for SORR.

However, if your portfolio balance is high enough that you're not concerned about running out of money, and your long-term plan is to create a certain level of income yearly no matter what because that brings down your tIRA balance over time to make RMDs more equitable to current taxation, then performing a Roth conversion in the midst of a crash is a nice little bonus since you get to convert more shares for the same dollar amount.

Even after FIRE, I propose that someone with a huge tIRA balance should consider being more aggressive with Roth conversions, rather than conservative. Over time your portfolio will grow faster than the income tax bracket thresholds are increased. That means the current tax year allows you to convert the largest percentage of your portfolio for the same tax cost. There are caveats with this as well since tax rates can change and markets can crash, but considering the overall long-term trend of markets going up, delaying will only lead to converting smaller and smaller percentages of your portfolio if your intention is to convert up to a certain tax bracket threshold for tax efficiency.

I have gone so far as to create an Excel spreadsheet that calculates our overall tax rate (federal + state + ACA premiums) each year so that I can model tax efficiency over our lifetimes. At a 40-45k annual spend, being the most tax-efficient now (paying $0 federal tax) leads to ridiculous RMDs later in life, assuming average historical returns continue into the future. Like 200k in RMDs ridiculous (in today's dollars). So it would actually behoove us to perform larger Roth conversions now to make that tax rate more even across our lifetimes.

[monarda]

I'm liking this discussion.

I'm thinking backwards, targeting the trad IRA so the RMD would be (maximally) close to your annual expected spend (at 70+), minus other sources of income, like pensions?

I'm less than 9 years away from 70, so I need to think about my total trad/Roth balance ratio, which is currently 1.27
and I'm contributing significantly more to trad (mostly in 403b) while I still work full time. Counting IRA alone, the ratio is 0.43
 

[boarder42]

I have gone so far as to create an Excel spreadsheet that calculates our overall tax rate (federal + state + ACA premiums) each year so that I can model tax efficiency over our lifetimes. At a 40-45k annual spend, being the most tax-efficient now (paying $0 federal tax) leads to ridiculous RMDs later in life, assuming average historical returns continue into the future. Like 200k in RMDs ridiculous (in today's dollars). So it would actually behoove us to perform larger Roth conversions now to make that tax rate more even across our lifetimes.

Your whole post sums up the conundrum nicely.  Any chance you can share this on google sheets or PM me this b/c I was about to build the exact same thing.  I feel like splitting the difference may make the most sense - I dunno either way because the 4% SWR is extremely conservative we're likely to end up with piles of RMDs like you say unless we maximize your conversion. 

Also I'm strongly considering margin instead of selling my taxable assets.  This would lead to an even larger reason to have more dry powder in roth conversions backlog in the event of a downturn so i could offset the losses without selling taxable shares.

[seattlecyclone]

The ACA "tax" will happen no matter what.

Yes and no. It will end when you (and your spouse, if applicable) are on Medicare. It may end sooner than that depending on legislative changes in the intervening time. It ends at high incomes, once the subsidy phases out completely. The phase-out causes marginal rates up to 30% for ACA insurance purchasers with rather low incomes. Then, unusually, you see your tax rate drop significantly once your income raises to the point where the subsidy is gone. If you have decided that you do want to convert a big chunk of your balance before Medicare age, you may find that it's optimal to alternate between years where your income is way above the phase-out range (so some of it is only taxed at the regular 12-22% brackets instead of 25-30%), and years where your income is very low so you get a sweet subsidy.

At a 40-45k annual spend, being the most tax-efficient now (paying $0 federal tax) leads to ridiculous RMDs later in life, assuming average historical returns continue into the future. Like 200k in RMDs ridiculous (in today's dollars). So it would actually behoove us to perform larger Roth conversions now to make that tax rate more even across our lifetimes.

Two things about $200k RMDs:
1) A married couple with $200k AGI, no ACA coverage, is on the cusp between the 22% and 24% bracket. A married couple with $50k AGI on ACA is paying nearly 30%. Even though the income amount is vastly more if you wait until your 70s, the percentage taxed away is less.
2) Having so much in your IRA that you have $200k RMDs means you likely ended up with a lot more money than you need. What's your plan for the excess cash? If you have charitable ambitions there's a provision of the tax code letting you give the first $100k of your RMD straight to a charity and it's completely excluded from income. That brings you down to a $100k AGI (or potentially zero if you and your spouse have the same amount in your IRAs and both take full advantage of this). That will bring you down a bracket or two, making the advantage of limiting your income during ACA years even greater.

[BicycleB]

Excellent discussion!!

[Mr. Green]

The ACA "tax" will happen no matter what.

Yes and no. It will end when you (and your spouse, if applicable) are on Medicare. It may end sooner than that depending on legislative changes in the intervening time. It ends at high incomes, once the subsidy phases out completely. The phase-out causes marginal rates up to 30% for ACA insurance purchasers with rather low incomes. Then, unusually, you see your tax rate drop significantly once your income raises to the point where the subsidy is gone. If you have decided that you do want to convert a big chunk of your balance before Medicare age, you may find that it's optimal to alternate between years where your income is way above the phase-out range (so some of it is only taxed at the regular 12-22% brackets instead of 25-30%), and years where your income is very low so you get a sweet subsidy.

True, if someone's income needs to be large enough that they get no ACA premium subsidies, it doesn't make sense to think of it as a tax. Most folks here probably won't be FIREing with those kinds of base expenses, though. It's valuable to understand how steep the marginal "tax" rate is on the loss of subsidies as someone's income moves from 150% of the federal poverty level to 400% and beyond.

A lot of the "how much do I need to convert" decision really just depends on each person's allocation between taxable, tIRA, and Roth IRA accounts. And of course how early they're retiring. This makes it basically impossible to offer universal advice, but threads like these can help people better understand their personal variables better.

We don't have a choice but to do sizeable Roth conversions because our taxable funds will run dry well before 59 1/2, even with good returns. It's possible that we could exhaust taxable funds in our 50s and then follow that up with a 72t SEPP on the tIRAs but that still feels like trying to land the plane as it's running out of gas. Too many variables for me to be comfortable with that scenario.

At a 40-45k annual spend, being the most tax-efficient now (paying $0 federal tax) leads to ridiculous RMDs later in life, assuming average historical returns continue into the future. Like 200k in RMDs ridiculous (in today's dollars). So it would actually behoove us to perform larger Roth conversions now to make that tax rate more even across our lifetimes.

Two things about $200k RMDs:
1) A married couple with $200k AGI, no ACA coverage, is on the cusp between the 22% and 24% bracket. A married couple with $50k AGI on ACA is paying nearly 30%. Even though the income amount is vastly more if you wait until your 70s, the percentage taxed away is less.
2) Having so much in your IRA that you have $200k RMDs means you likely ended up with a lot more money than you need. What's your plan for the excess cash? If you have charitable ambitions there's a provision of the tax code letting you give the first $100k of your RMD straight to a charity and it's completely excluded from income. That brings you down to a $100k AGI (or potentially zero if you and your spouse have the same amount in your IRAs and both take full advantage of this). That will bring you down a bracket or two, making the advantage of limiting your income during ACA years even greater.

If we ended up with excessive tIRA balances in our 70s, we would absolutely do something charitable with a bunch of that money. I'd rather see the money put to use while I'm alive than just willing it to some organizations after we're dead.

In yet another example of why this whole calculation is so personal based on individual circumstances, I'll use my own financial picture as an example. We could live on 40-45k per year comfortably, probably for the rest of our lives (while adjusting for inflation). However, if our income was 20-30k per year higher now, while we're young, we would probably spend some of that money. Maybe it would be a little more travel or more luxurious travel. Maybe we'd choose to support some charitable causes now, albeit in smaller amounts. We're also hopefully about to have a child or children. I don't know if that kid will end up with special needs or any other thing that might require more income than our 40-45k projection. So it would be smart to take some of those Roth conversions and allow them to marinate. For me, it's about having access to as much of my money while young as possible, while still being tax efficient. So it behooves me to look past the steep marginal "tax" rate of losing ACA premium subsidies to the long-term picture. I'm not going to try to be perfect, but the difference between all I have to do, and what I could do, is so big that the smart move is paying some extra tax.

Future changes (legislation, etc.) could absolutely render current moves obsolete. There's nothing I can do about those kinds of unknowns.  You can't overthink those things or you'll go crazy. The long-term modeling is already complicated enough as it is.

Ironically, here's an example from my model which is already out of date because it assumes a subsidy cliff at 400% of the federal poverty level and a $2,000 child tax credit. At age 42 (year 2025), we're using a bronze ACA plan. Our income bumps right up against 400% FPL for a family of 3, which is just over 86k in 2021. I don't guess at future inflation, just use today's dollars. Our Roth conversions are just over 46k. We have 5k in rental income but the remainder is dividends and long-term capital gain harvesting. The $2,000 child tax credit means federal income tax is less than $1,000 (the newly expanded credit is even more beneficial). ACA premiums for the year come in just over 3k, using 2021's premium amounts as a guide.

Considering ACA premiums as a tax, because we could get those down to $0 if we really wanted, our total tax rate (state + federal + ACA premiums) comes out to ~9% of our income, or about $7,500. We could make that tax rate less than 5% (no federal tax, no ACA premium, and 5.25% state tax on income over $21,500). But that leaves us with less access to money while young, and likely those ridiculous RMDs in old age.

I need to update my model for the latest ACA changes passed in April. Sigh. I'm satisfied with the broad strokes though. I will try to sanitize it a bit and toss it up here in case anyone wants to look it over. It ain't fancy, I'll say that.

Also, a lot of this may only matter so immensely for those FIREing fairly young. If I was 50 some of this wouldn't be very important because I have a 72t SEPP as an option, access to all my money in a decade, and the child(ren) ship has already sailed. There are significantly fewer unknown variables that would affect that person's financial future.

[secondcor521]

It's valuable to understand how steep the marginal "tax" rate is on the loss of subsidies as someone's income moves from 150% of the federal poverty level to 400% and beyond.

This might be helpful to you:

https://seattlecyclone.com/aca-premium-tax-credits-2021-edition/

particularly the second graph.

h/t of course to @seattlecyclone.

[Mr. Green]

It's valuable to understand how steep the marginal "tax" rate is on the loss of subsidies as someone's income moves from 150% of the federal poverty level to 400% and beyond.

This might be helpful to you:

https://seattlecyclone.com/aca-premium-tax-credits-2021-edition/

particularly the second graph.

h/t of course to @seattlecyclone.

I will dig into that for my model refresh. I still have one of @seattlecyclone's ACA posts bookmarked from 2015! This stuff is complicated enough that great explanations and visuals are valuable to hang on to. :)

[seattlecyclone]

The ACA "tax" will happen no matter what.

Yes and no. It will end when you (and your spouse, if applicable) are on Medicare. It may end sooner than that depending on legislative changes in the intervening time. It ends at high incomes, once the subsidy phases out completely. The phase-out causes marginal rates up to 30% for ACA insurance purchasers with rather low incomes. Then, unusually, you see your tax rate drop significantly once your income raises to the point where the subsidy is gone. If you have decided that you do want to convert a big chunk of your balance before Medicare age, you may find that it's optimal to alternate between years where your income is way above the phase-out range (so some of it is only taxed at the regular 12-22% brackets instead of 25-30%), and years where your income is very low so you get a sweet subsidy.

True, if someone's income needs to be large enough that they get no ACA premium subsidies, it doesn't make sense to think of it as a tax. Most folks here probably won't be FIREing with those kinds of base expenses, though. It's valuable to understand how steep the marginal "tax" rate is on the loss of subsidies as someone's income moves from 150% of the federal poverty level to 400% and beyond.

Right. I think there's some "standard advice" that it's good to keep your income as similar as possible from year to year to minimize your tax burden in the long run. This advice depends on a progressive tax structure, where the tax rates at lower incomes are lower than the tax rates at higher incomes. Better to keep your income just below the top of a tax bracket every year than to venture above that some years and pay tax at a higher rate.

The ACA actually creates something of a regressive tax structure, where you have a pretty high overall marginal rate in the subsidy phase-out range and then once your subsidy is exhausted your rate goes down to whatever the base tax bracket is at that point. Income stability is best for a progressive tax because you avoid the higher rates at higher incomes, but income fluctuation is best for a regressive tax, because you have more of your long-term overall income in the lower-rate range that occurs at higher income levels.

That's why I'm saying that if your FIRE plans depend on a Roth conversion ladder, try modeling the effect of switching from no subsidies one year to huge subsidies the next, compared to having medium subsidies every year. You may find that the oscillating income produces a better result.

A lot of the "how much do I need to convert" decision really just depends on each person's allocation between taxable, tIRA, and Roth IRA accounts. And of course how early they're retiring. This makes it basically impossible to offer universal advice, but threads like these can help people better understand their personal variables better.

I completely agree.

In yet another example of why this whole calculation is so personal based on individual circumstances, I'll use my own financial picture as an example. We could live on 40-45k per year comfortably, probably for the rest of our lives (while adjusting for inflation). However, if our income was 20-30k per year higher now, while we're young, we would probably spend some of that money. Maybe it would be a little more travel or more luxurious travel. Maybe we'd choose to support some charitable causes now, albeit in smaller amounts. We're also hopefully about to have a child or children. I don't know if that kid will end up with special needs or any other thing that might require more income than our 40-45k projection. So it would be smart to take some of those Roth conversions and allow them to marinate. For me, it's about having access to as much of my money while young as possible, while still being tax efficient. So it behooves me to look past the steep marginal "tax" rate of losing ACA premium subsidies to the long-term picture. I'm not going to try to be perfect, but the difference between all I have to do, and what I could do, is so big that the smart move is paying some extra tax.

Yeah that's totally fair. My personal situation is that we have almost enough in taxable accounts to sustain a pretty lean FIRE, and a bunch of Roth basis (helped by mega backdoor contributions), so the access piece is much less of a concern. The pre-tax retirement balances we have are more in a "old man money" and/or "probably not going to need it" bucket, so the focus is more on figuring out when the absolute cheapest time to realize that income will be. If you know you need to realize that income relatively soon that's a different ballgame. Do try to model the income oscillation thing.

Future changes (legislation, etc.) could absolutely render current moves obsolete. There's nothing I can do about those kinds of unknowns.  You can't overthink those things or you'll go crazy. The long-term modeling is already complicated enough as it is.

For sure. I tend to plan as though the laws will remain as they are today, because the potential universe of changes is infinite. I know of course that the laws will change, but I don't claim to have any special insight into how those changes will look. The best I can hope for is that I have a diverse enough set of accounts that it's easy to adapt my plans to the changes as they happen, however they might look.

[Simpli-Fi]

Definitely lot to learn…ptf

[Caoineag]

Ah yes. We get to play this game now. Plan to do some Roth conversions, some trad Ira withdrawals and try to keep the tax rate reasonable but we may end up doing alternating years if it makes more sense. We are very young retirees so don't want to do the 72t withdrawals. Aca subsidies definitely complicated things. What a problem to have.

[boarder42]

Ah yes. We get to play this game now. Plan to do some Roth conversions, some trad Ira withdrawals and try to keep the tax rate reasonable but we may end up doing alternating years if it makes more sense. We are very young retirees so don't want to do the 72t withdrawals. Aca subsidies definitely complicated things. What a problem to have.

Why are you planning trad ira withdrawals at 39. Do you not have a 5 year bridge in other accounts.  We have 25 years of this so 72t doesn't sound fun to me.

[MustacheAndaHalf]

As we near the withdrawal stages I am digging deeper into how much we should convert.  I was planning to go all the way to the top of the 12% bracket which may make sense in a down market but I don't think it makes sense long term. 

Top of 12% ~106k currently
ACA Cost - 5607
Tax Cost - 6500
Total Cost to covert - 12,107

vs

90k
ACA Cost - 2979
Tax Cost - 3712
Total cost to covert - 6691

Setting aside ACA Cost for a moment, you are paying 4.1% tax on the first $90k, and then 17.4% on the next $16k.  That doesn't sound like the 12% tax bracket to me.  But with those numbers, avoiding the added 33.4% cost (ACA + Tax) makes sense, and stopping at $90k seems more efficient.

The larger question is predicting future tax rates.  Corporate tax rates seem to be a race to the bottom, so those seem more likely to fall than personal tax rates.  Personal tax rates seem to be at historic lows, so my guess would be they go higher over the upcoming decades.  I'd expect the 12% bracket is a good deal now.  In either event, a Roth Conversion brings certainty: you know you'll pay 12% tax, instead of the future tax rate.

[boarder42]

As we near the withdrawal stages I am digging deeper into how much we should convert.  I was planning to go all the way to the top of the 12% bracket which may make sense in a down market but I don't think it makes sense long term. 

Top of 12% ~106k currently
ACA Cost - 5607
Tax Cost - 6500
Total Cost to covert - 12,107

vs

90k
ACA Cost - 2979
Tax Cost - 3712
Total cost to covert - 6691

Setting aside ACA Cost for a moment, you are paying 4.1% tax on the first $90k, and then 17.4% on the next $16k.  That doesn't sound like the 12% tax bracket to me.  But with those numbers, avoiding the added 33.4% cost (ACA + Tax) makes sense, and stopping at $90k seems more efficient.

The larger question is predicting future tax rates.  Corporate tax rates seem to be a race to the bottom, so those seem more likely to fall than personal tax rates.  Personal tax rates seem to be at historic lows, so my guess would be they go higher over the upcoming decades.  I'd expect the 12% bracket is a good deal now.  In either event, a Roth Conversion brings certainty: you know you'll pay 12% tax, instead of the future tax rate.

this is a combination of state and federal taxes with the child tax deductions my state tax is 5.4% so 12+5.4 = 17.4%

[Caoineag]

Ah yes. We get to play this game now. Plan to do some Roth conversions, some trad Ira withdrawals and try to keep the tax rate reasonable but we may end up doing alternating years if it makes more sense. We are very young retirees so don't want to do the 72t withdrawals. Aca subsidies definitely complicated things. What a problem to have.

Why are you planning trad ira withdrawals at 39. Do you not have a 5 year bridge in other accounts.  We have 25 years of this so 72t doesn't sound fun to me.

We had more than 5 years but spent it on a house. I won't need to do more than a couple of years of trad Ira withdrawals which is why I  won't do a 72t. Just a couple of years of higher tax rate, not the end of the world. We have a lot of assets in trad Ira form right now. I have some in Roth iras but not enough to completely avoid the trad Ira withdrawals. We still are wealthier than when we retired so I can't feel too heart broken about it.

[Mr. Green]

I overhauled my long-term FIRE spending/tax model today and it revealed some things I was not prepared for. Even though some of the data is specific to my personal situation, I think it's a valuable process to share. Having a large enough W2 income to retire in your 30s probably means a small Roth balance, and large taxable and tIRA balances. So I'm guessing there'd be others out there in a similar situation as us. Our portfolio split is currently ~60% tIRA/35% taxable/5% Roth.

If you want to download and dig into the specific data of the model, feel free, but I'm going to talk about the broader trends here because that is what was so surprising.

I first made this model because I knew that continuing to create an income just big enough for us to live on would inevitably lead to obscene account balances in old age. The question I wanted to answer was, "How large could our income be during the younger years of our lives (when we'd be more likely to spend it) while being as tax efficient as possible over the long term?"

A couple of general points about the data before I dive into the trends.....

• I used a 5% growth rate, which is a bit conservative since the long-term figure in the US is closer to 7% before inflation
• State taxes are specific to North Carolina
• ACA premiums are specific to the bronze and silver plans I've chosen based on my County of residence
• I consider ACA premiums a "tax" because we could lower them to $0, but Medicare premiums are not since base premiums are a set amount. So ACA premiums are reflected in the overall tax percentage but Medicare premiums are not.
• The percentage of our taxable accounts that are long-term capital gains are specific to us
• Assumes one child and the associated child tax credit
• Income tax thresholds are Married Filing Jointly

At first, I thought making our tax rate as even as possible over our lifetimes would be best. However, it became apparent that doing this would leave a huge portfolio balance in old age. The model quickly struck down that idea.

I also assume that we will spend less in old age (50k), and I'm probably being overly generous with that figure as it's more than we spend now. After watching my very healthy grandparents age, it's quite evident that spending really falls off in your 80s and beyond (maybe even 70s depending on health), unless it's increased healthcare spending. As we get into our 70s and 80s, Social Security and a bit of rental income will cover about half of our planned spending. The model indicates that 500k in our tIRAs will be enough to cover the remainder of our spending in old age (~25k), while growth keeps our tIRA balance healthy enough to cover long term care costs at the end of our lives.

So once we hit 72, Social Security, some rental income, and our RMDs are enough to cover all spending, essentially leaving our remaining Roth IRA funds to compound untouched until we die. So with that trend in mind, I started to examine how we could increase our Roth conversions and spending earlier in life to bring down the balance of our Roth IRA fairly low by the age of 72.

A side effect of having exhausted our taxable accounts by the age of 72 and having only small RMDs is that our taxable income in our 70s and beyond is basically nothing. As far as I can tell, this is simply the way it has to be if we want to maximize the access we have to our money when we're younger.

As I started increasing our Roth conversions over the next 20-30 years, it became evident that the real pinch point was going to be the age of 60 because we don't have access to Roth gains prior to that without penalty. Bearing that in mind, I started pulling larger Roth conversion amounts forward, even closer to now, than I had originally planned for. It got to a point where the model indicated that having maximum access to our money over the next 30 years meant immediately increasing Roth conversions to 70-80k per year. In doing so, our AGI rises to ~120k for almost a decade until our taxable accounts are exhausted and there are no more long-term capital gains to be had. After that, our AGI drops to ~85k per year for another decade until we finally reach the point that it doesn't make sense to do any more Roth conversions because we want to leave enough tIRA money for those RMDs to cover our spending in old age. Plus we can't convert everything to Roth IRAs before we get on Medicare or we'll have no way to create income since we exhausted taxable accounts already.

Immediately maximizing Roth conversions while still maintaining a consistent, low overall tax rate from year to year does eventually have the side effect of "running out" of convertible money. Once we hit that point, we only convert the minimum needed to qualify us for private ACA insurance. Once our income becomes minimal, estimated to be in our late 50s, we basically pay no taxes. So we actually end up with a model that shows us paying little to no tax from our late 50s through the end of our lives. Taxable accounts have been exhausted, most all tIRA money has been converted to Roth, and what tIRA money remains barely goes above income tax exemptions. Again, this seems to be something that has to happen if we're to maximize access to our money while young.

So....shortcomings of the model. Our portfolio is obviously not going to grow at 5% every year. Sequence of returns is going to affect this. But 5% is also a bit conservative, historically speaking, so maybe our spending could be even higher than this. Future legislation could require the model to change.

And of course, the obligatory note that this is just a model that I created to gain better insight into long-term trends for income and tax efficiency. This will not dictate our spending but it can be a very useful guide.

And if anyone should find an error, please let me know!

I suppose I shouldn't be surprised that the model shows large, immediate Roth conversions are needed to maximize access to our money when 60% of it currently sits in tIRAs. We could choose to not convert that much money, leaving higher account balances in old age but why? It seems the prudent approach, given that the future is unknown, would be to maximize access to our money at any given time, even if it means paying a little bit more tax than we'd have to only covering our base expenses. I do have to say I was pleasantly surprised to see that we could spend 80-100k over the next 30 years while having a marginal tax rate of only ~12%. And that includes ACA insurance premiums. That's pretty damn good.


Edit: Attachment of my model was removed due to a request for an updated/improved version, which can be found downthread here

[Mr. Green]

Oh, and I almost forgot. @seattlecyclone I was thinking about the income fluctuation you were talking about. I could definitely see how it might ultimately work out better for the ACA premiums alone to alternate high and low years. I had planned to try and work this into my model and see what it would look like but I was pleasantly surprised with how low the overall "tax" rate was, thanks to using a Bronze insurance plan that had pretty small premiums even at higher income levels. I'm still trying to wrap my head around how fluctuating our income every other year might improve this. It's a lot!

[seattlecyclone]

Oh, and I almost forgot. @seattlecyclone I was thinking about the income fluctuation you were talking about. I could definitely see how it might ultimately work out better for the ACA premiums alone to alternate high and low years. I had planned to try and work this into my model and see what it would look like but I was pleasantly surprised with how low the overall "tax" rate was, thanks to using a Bronze insurance plan that had pretty small premiums even at higher income levels. I'm still trying to wrap my head around how fluctuating our income every other year might improve this. It's a lot!

It is a lot! Here's a quick example. Let's suppose you're a married couple with no kids, 40 years old, looking to average $75k of Roth conversions every year. No other income.

Scenario 1: Smooth income ($75k/year)
AGI: $75k
Income tax: $5,593
Health insurance (second-cheapest silver plan): $6,388 (after $2,568 subsidy)
Annual tax plus insurance premiums: $11,981
Two-year total: $23,962

Scenario 2: Fluctuating income ($25k one year, $125k the next)
Year 1:
AGI: $25k
Income tax: $0
Health insurance (second-cheapest silver plan): $16 (after $8,940 subsidy) <-- Sweet cost-sharing subsidies lower the out-of-pocket costs too!
Annual tax plus insurance premiums: $16

Year 2:
AGI: $125k
Income tax: $13,481
Health insurance (second-cheapest silver plan): $8,956 (no subsidy)
Annual tax plus insurance premiums: $22,437

Two-year total: $22,453

You can see that in this scenario you save about $1,500 every two years in tax + insurance premiums by varying your income, plus every other year your insurance comes with super-low out-of-pocket costs.

Adding kids into the mix complicates things (not just tax-wise, of course!). In my state a couple with two kids and income below $84k will have the kids on Apple Health (Medicaid/CHIP). This threshold may vary by state, I'm not sure. If you want to have the whole family on the same insurance plan you either need to have your income low enough for the adults to be on Medicaid too (<$36k/year), or high enough to have the kids disqualify (>$84k). The alternating $25k/$125k income scenario would flip you between both extremes, rather than being on private coverage all the time if you didn't have kids. I think the child tax credit would be the same either way under this year's rules, but under last year's rules you'd lose some of it with such a low income, especially if it's all unearned income from Roth conversions. So many variables!

[ender]

If you are simply looking to pay as little tax as possible now, a blend of income from taxable accounts (long-term capital gains) and Roth IRA conversions is best. In 2021, married folks can convert $25,100 from tIRA to Roth tax-free. If you have children, that really juices the amount you can convert by allowing you to offset your federal taxes with the child tax credit. If planned for correctly, in 2021 a married couple could have an income of $106,150 consisting solely of Roth conversions and long-term capital gains and pay zero federal tax. If you have enough kids I suppose it could be even more, especially with the newly expanded child tax credit.

One consideration you aren't thinking here is once you start "losing" the refundable portions of credits, you are still paying a marginal tax rate even if your federal tax owed is still $0 or negative.

Going from $0 owed/$5k refund to $0/$0 still means you had a marginal tax rate.

That doesn't change the conclusion you came to necessarily but it does change how you are thinking about the "tax-free" aspect.

However, there's a long-term angle to consider as well. If your tIRA balances are high enough (ours are), solely focusing on minimizing tax burden now may lead to tIRA balances that are so large by the time you hit Required Minimum Distributions that you're paying significantly higher taxes on an income in old age that is so large you likely can't spend it all because your mobility and health are declining. This aspect is especially worth considering if you knew balancing current and future taxes would result in a larger income now that you could enjoy while younger and healthier. The obvious caveat here is that we don't know exactly what future taxes will be so this optimization won't be perfect.

Even after FIRE, I propose that someone with a huge tIRA balance should consider being more aggressive with Roth conversions, rather than conservative. Over time your portfolio will grow faster than the income tax bracket thresholds are increased. That means the current tax year allows you to convert the largest percentage of your portfolio for the same tax cost. There are caveats with this as well since tax rates can change and markets can crash, but considering the overall long-term trend of markets going up, delaying will only lead to converting smaller and smaller percentages of your portfolio if your intention is to convert up to a certain tax bracket threshold for tax efficiency.

One of the things I stressed when I gave a "To Roth or Not" presentation last year was looking at which failure mode is more palatable. Paying too much now? Or later?

Personally, for me the failure mode of paying too much in taxes because I took the guaranteed tax savings now is less of an emotional negative than paying too much in taxes to lock in Roth.

I also think a risk people do not consider is how Roth may be implicitly taxed in the future. I do not think it's likely Roth withdrawals will be directly taxed but I would not be surprised if they change to for example impact your ACA subsidies or SS taxation. There is an assumption in a lot of FIRE calculations that Roth growth never will negatively impact your financial situation or have any tax impact.

But the flip side is risk such as the 5 year conversion rule or Roth conversion process changes.

I have gone so far as to create an Excel spreadsheet that calculates our overall tax rate (federal + state + ACA premiums) each year so that I can model tax efficiency over our lifetimes. At a 40-45k annual spend, being the most tax-efficient now (paying $0 federal tax) leads to ridiculous RMDs later in life, assuming average historical returns continue into the future. Like 200k in RMDs ridiculous (in today's dollars). So it would actually behoove us to perform larger Roth conversions now to make that tax rate more even across our lifetimes.

This gets to the failure mode analysis I talked about before.

If I have a larger than anticipated tax burden in retirement due to RMDs I... have "won."

[secondcor521]

And if anyone should find an error, please let me know!

Two errors, and a couple of things you could consider changing.

1.  The RMD divisors you are using are the old ones.  New ones were published by the IRS in the last few months that will apply to you by the time you're 72.  The new ones are higher, meaning that the percentage withdrawals are lower.  Probably won't affect your model and solution too much if at all.

2.  The standard deduction increases when you hit age 65, and also when your spouse hits age 65.

Other considerations:

You're ignoring IRMAA, although with your goals and solution it probably won't affect you, so probably not a real issue.

In general, I prefer to work in nominal dollars, so if I have a number for my SS in 2041 or whatever, I keep the number in 2041 dollars.  Doing so reasonably accurately requires me to inflate everything according to the various inflation rates.

You've increased your portfolios by 5%, but it looks like you haven't adjusted anything else in your spreadsheet.  The things that I adjust in my model are my SS amounts, the tax bracket amounts, the standard deduction amounts, the IRMAA brackets and amounts (again, possibly N/A for you), and the FPL amounts.

Some people argue that all of this stuff cancels out.  Maybe so, but it makes me feel better to model it all out in case it doesn't.  It can add a lot to complexity though, and more complexity can increase the chances of an error in the model.

I didn't dig into it for this, but in general since it sounds like you have a child, you might want to model the tax changes as that child gets older - your ACA family size will drop by one at some point, and your child tax credit changes (at age 6 and age 17 or so, IIRC).  Things of that nature.

Overall it looks quite nice though.

Another thing you might take a look at is i-orp (www.i-orp.com).  You can't probably bend it to model what you are trying to accomplish (more spendable when young) as I think the tool tries to maximize lifetime spending instead.  But it may be informative to play with.  Use the advanced version - there should be a checkbox at the top somewhere.

[Mr. Green]

It is a lot! Here's a quick example. Let's suppose you're a married couple with no kids, 40 years old, looking to average $75k of Roth conversions every year. No other income.

Scenario 1: Smooth income ($75k/year)
AGI: $75k
Income tax: $5,593
Health insurance (second-cheapest silver plan): $6,388 (after $2,568 subsidy)
Annual tax plus insurance premiums: $11,981
Two-year total: $23,962

Scenario 2: Fluctuating income ($25k one year, $125k the next)
Year 1:
AGI: $25k
Income tax: $0
Health insurance (second-cheapest silver plan): $16 (after $8,940 subsidy) <-- Sweet cost-sharing subsidies lower the out-of-pocket costs too!
Annual tax plus insurance premiums: $16

Year 2:
AGI: $125k
Income tax: $13,481
Health insurance (second-cheapest silver plan): $8,956 (no subsidy)
Annual tax plus insurance premiums: $22,437

Two-year total: $22,453

You can see that in this scenario you save about $1,500 every two years in tax + insurance premiums by varying your income, plus every other year your insurance comes with super-low out-of-pocket costs.

I think this might fall apart depending on the income level one chooses, the specific insurance plan costs and premium tax credit one gets, and other income variables.

I first tried doing this in my model for the first two years we're using a bronze plan. The smooth taxation method has our AGI at 130k. This happens while we're still depleting our taxable accounts so there are long-term capital gains involved. I should also highlight the fact that the model does two things semi-independently. I input how much I want our spending to be each year, before taxes and insurance premiums, and this adjusts our investment account balances accordingly (after adding the requisite taxes). I also adjust our AGI, mostly through Roth conversions, to determine what amount will result in similar taxation year to year. For the two years I've chosen, our spending is at 80k. We have LTCG in these early years and at that spending level, I can't get our income under 150% of the federal poverty level for a family of 3 ($32,580). The best I can do is about 44k. This is early enough that I can't really replace those funds from our taxable accounts with Roth principal, as there's only 30k there (we're still in the 5-year wait period for the start of these big conversions).

So I used 45k and 215k as my fluctuation years. All income tax in the examples is federal only.

Scenario 1: Smooth income ($130k/year)
AGI: $130k
Income tax: $4,840
Health insurance (cheapest bronze plan): $4,901
Annual tax plus insurance premiums: $9,741
Two-year total: $19,482

Scenario 2: Fluctuating income ($45k one year, $215k the next)
Year 1:
AGI: $45k
Income tax: $0
Health insurance (cheapest bronze plan): $0
Annual tax plus insurance premiums: $0

Year 2:
AGI: $215k
Income tax: $19,140
Health insurance (cheapest bronze plan): $11,870 (no subsidy)
Annual tax plus insurance premiums: $31,010

Two-year total: $31,010

Worth noting is the use of a Bronze plan in Year One of the fluctuating version, which is not efficient since it gives up CSRs. The lowest cost Silver plan for us in that instance would cost about $1,000 a year. I caught this after I reverted the model to look at another scenario. I guess one would have to consider their family health circumstances to determine if the extra $1,000 in costs upfront was worth using the more expensive Silver plan.

One shortcoming of the model this highlights is that it does not begin to spend Roth principal while there are still taxable funds left. When I created it, I did not envision that it would suggest an income so high, sustained over a long period. Using some Roth principal each year would reduce LTCG needed to fund our annual spending. While those gains are taxed at 0%, they do push our overall AGI up which still costs us additional taxes and increased health insurance premiums. I will look into adapting this concept to see how it would improve the projection.

If I run this same comparison in later years when there are no more LTCG impacting our AGI, here's what I get.

Scenario 1: Smooth income ($86k/year)
AGI: $86k
Income tax: $4,905
Health insurance (cheapest bronze plan): $1,125
Annual tax plus insurance premiums: $6,030
Two-year total: $12,060

Scenario 2: Fluctuating income ($32k one year, $140k the next)
Year 1:
AGI: $32k
Income tax: $0
Health insurance (cheapest silver plan): $0
Annual tax plus insurance premiums: $0

Year 2:
AGI: $140k
Income tax: $14,775
Health insurance (cheapest bronze plan): $5,812 (still some subsidy)
Annual tax plus insurance premiums: $20,587

Two-year total: $20,587

I think the variable that is making my examples look so much worse than yours is the use of a bronze plan. The reduced cost allows one's AGI to be higher before the cost of insurance starts to rise dramatically. How much really just depends on how expensive individual plans are in one's location and how big the premium subsidy is. Where we are, the value of the premium subsidy and the cost difference between a bronze and silver plan means we save 5k in premiums at the same AGI. At 79.5k AGI the bronze plan costs $0.88 a month and the silver plan we would use is $422 per month.

Should we find ourselves in a situation where we need enough medical care long-term that we're always choosing a silver plan I would definitely re-examine fluctuating AGI because I think it will be more comparable to your example.

[Mr. Green]

One consideration you aren't thinking here is once you start "losing" the refundable portions of credits, you are still paying a marginal tax rate even if your federal tax owed is still $0 or negative.

Going from $0 owed/$5k refund to $0/$0 still means you had a marginal tax rate.

That doesn't change the conclusion you came to necessarily but it does change how you are thinking about the "tax-free" aspect.

While true, the refundable portion of the child tax credit requires earned income, doesn't it? We have none, so we would forfeit that portion of the CTC regardless.

One of the things I stressed when I gave a "To Roth or Not" presentation last year was looking at which failure mode is more palatable. Paying too much now? Or later?

Personally, for me the failure mode of paying too much in taxes because I took the guaranteed tax savings now is less of an emotional negative than paying too much in taxes to lock in Roth.

I also think a risk people do not consider is how Roth may be implicitly taxed in the future. I do not think it's likely Roth withdrawals will be directly taxed but I would not be surprised if they change to for example impact your ACA subsidies or SS taxation. There is an assumption in a lot of FIRE calculations that Roth growth never will negatively impact your financial situation or have any tax impact.

But the flip side is risk such as the 5 year conversion rule or Roth conversion process changes.

Roth changes may be a risk but in our particular case, we are stuck working with what we can given where our money is located. If we waited until age 60 to access our tIRA money to avoid the 10% penalty, we leave 60% of our money to grow until a point where arguably most of our high spending years are over. In the face of my model telling me we could dramatically increase our spending, perhaps I need to re-examine a 72t SEPP for some portion of that increased spending. That could mitigate future tax treatment of Roth funds.

While I understand that the government can do anything they damn well please, I have a hard time believing they would tax straight Roth withdrawals. That would be double taxation since we've already paid tax once on conversions. Also, if they decided to start taxing Roth gains, it's basically the same thing as a tIRA at that point so why not just eliminate the Roth account entirely. Are there other factors here I'm not thinking of?

This gets to the failure mode analysis I talked about before.

If I have a larger than anticipated tax burden in retirement due to RMDs I... have "won."

I'm definitely a conservative person. I want a larger than anticipated tax burden in retirement too. And for those that are comfortable spending what they already are for the rest of their lives, this is a useless exercise. One could always leave their money where it is and if something happens that does require more income they can change their process then, or if it's a one-time deal they can just take the 10% penalty of an early tIRA withdrawal. But we probably would increase our annual spending a bit. I don't know that I'd kick it up to the 80-100k per year that the model is indicating; it's hard for me to fathom spending that kind of money after years of 40-45k in spending. But I could see an increase to 60-70k, whether it's choosing to fund charitable efforts now, spending a little more on our kid, having more kids than we originally thought we wanted, taking care of a parent, etc.

For me, what the modeling can do is eliminate obscene possibilities. For instance, if I adjust our spending down to 50k for life (plus taxes on top) our RMDs start at 150k at age 72 and work up to 300k (in today's dollars) by age 90 or so. On a certain level, I feel like that is simply poor planning. At 80 our accounts are cumulatively worth $7.5 million and at 100 it's over $19 million. That's obscene. And this is using a growth rate less than the historical average. Though the growth rate is smooth in the model and we know it will not actually be that way.

We would choose to increase spending over our lifetimes, even if we planned to donate all the extra over our original spending amount, rather than allow the money to pile up like that. I think being able to use that money in various ways would be more rewarding, and we can still be conservative re: sequence of returns risk to ensure that we didn't overspend. Arguably, the model isn't telling us much different than what the 4% rule does. At $1.75 million, 4% is an annual spend of 70k per year.

Two errors, and a couple of things you could consider changing.

1.  The RMD divisors you are using are the old ones.  New ones were published by the IRS in the last few months that will apply to you by the time you're 72.  The new ones are higher, meaning that the percentage withdrawals are lower.  Probably won't affect your model and solution too much if at all.

I've found articles presenting proposed new divisors but the info on the IRS' website still shows the old ones. I probably just haven't found the right thing yet. I'll keep digging.

2.  The standard deduction increases when you hit age 65, and also when your spouse hits age 65.

I believe I took that into consideration, a $1,300 increase per person. I'm guessing you noticed the state standard deduction, which doesn't change in North Carolina.

You're ignoring IRMAA, although with your goals and solution it probably won't affect you, so probably not a real issue.

The model doesn't track increased Medicare premiums based on income since the goal was to not have an income over 85k by that point. And I don't have a good way of illustrating how the cost of medicare supplemental plans increase in cost with age (the model includes thosse), though I suppose one has little control over that so maybe it's not worth thinking about.

In general, I prefer to work in nominal dollars, so if I have a number for my SS in 2041 or whatever, I keep the number in 2041 dollars.  Doing so reasonably accurately requires me to inflate everything according to the various inflation rates.

You've increased your portfolios by 5%, but it looks like you haven't adjusted anything else in your spreadsheet.  The things that I adjust in my model are my SS amounts, the tax bracket amounts, the standard deduction amounts, the IRMAA brackets and amounts (again, possibly N/A for you), and the FPL amounts.

Some people argue that all of this stuff cancels out.  Maybe so, but it makes me feel better to model it all out in case it doesn't.  It can add a lot to complexity though, and more complexity can increase the chances of an error in the model.

I always work in nominal dollars (today's dollars) because I hate having to think about inflation across all the number values. The 5% growth rate I assume is based on the historical growth rate of 6.7% before inflation. If we include inflation that rate is more like 10% I believe.

I think the SS values should be somewhat self-adjusting via the increase in average wages until we're 60. So the 27k SS projects we'll receive at 67 will no doubt increase as wages do, though if they change things to address the coming trust exhaustion that will need to be accounted for.

I didn't dig into it for this, but in general since it sounds like you have a child, you might want to model the tax changes as that child gets older - your ACA family size will drop by one at some point, and your child tax credit changes (at age 6 and age 17 or so, IIRC).  Things of that nature.

I did account for $2,000 in CTC until the child is 17, then dropping to $500 at 18 and 19, then $0 after. Not assuming college in the name of being conservative with any benefits. I didn't bother with the increased CTC this year, as I believe it's only this year so we'll miss that.

Another thing you might take a look at is i-orp (www.i-orp.com).  You can't probably bend it to model what you are trying to accomplish (more spendable when young) as I think the tool tries to maximize lifetime spending instead.  But it may be informative to play with.  Use the advanced version - there should be a checkbox at the top somewhere.

I will take a look at that!

[2sk22]

I read through the strategies outlined here with considerable interest as I have also been thinking about Roth conversions. Let me start by admitting that my situation is very much in the nature of an extreme first world problem.

My wife and I have over $3M in pre-tax accounts ($1.7M in my tIRA and $1.5M in my wife’s 401k). I am in my late 50s and retired last year but my wife is still working and earning a fabulous salary. Since we get our health insurance through her company, we don’t have any ACA issues to deal with as of now.

We are already in a high tax bracket solely based on my wife's income. Consequently, it appears that I can’t even start my Roth conversions until she decides to retire - which is a few years away.  It appears that we have a window of opportunity for Roth conversions between when she decides to retire and RMDs kick in. Looking at the rules, one way or another, we are probably going to be paying hefty income taxes well into old age

[secondcor521]

New RMD tables.  Age 72 factor should be 27.3 if my spreadsheet is right:

https://www.taxwarriors.com/blog/updated-irs-life-expectancy-tables-mean-smaller-rmds-in-2022

As far as the standard deduction goes, I was looking at column AG on the first worksheet, but I just didn't look down far enough.

You didn't specifically reply about adjusting the standard deduction and tax brackets - they are increased by inflation every year, so you not adjusting them results in more aggressive taxation than will actually be the case, and apparently by quite a bit.  I'm 14 years older than you, so I have 14 years less of inflation adjustments before RMD age, and although the standard deduction for a single over 65 is $14,250 (I think that's the right 2021 number), by the time I start RMDs at age 72 it'll be $21,175.  (I assume 2% inflation.)

As mentioned before, to adjust your spreadsheet for that would add complexity though.  My spreadsheet just has the tax brackets in rows by year and does the basic tax calculations for each year in each row.  You could accomplish something similar by adding something like "...*1.02^(row()-3)" in cells like first worksheet cells AM4:AM66 whenever referring to thresholds from the 2021 taxes sheet.

(Of course, with the way things go these days, trying to predict taxes beyond the current year might be viewed as a fool's errand, so you could wave it away with that.  You might also just prefer to have the bias in there.  It also might not be that bad if you're going to update the worksheet yearly with the new constants on the "2021 taxes" sheet, as I think your situation will tend towards a reasonable optimum if you iterate each year.)

[secondcor521]

I read through the strategies outlined here with considerable interest as I have also been thinking about Roth conversions. Let me start by admitting that my situation is very much in the nature of an extreme first world problem.

My wife and I have over $3M in pre-tax accounts ($1.7M in my tIRA and $1.5M in my wife’s 401k). I am in my late 50s and retired last year but my wife is still working and earning a fabulous salary. Since we get our health insurance through her company, we don’t have any ACA issues to deal with as of now.

We are already in a high tax bracket solely based on my wife's income. Consequently, it appears that I can’t even start my Roth conversions until she decides to retire - which is a few years away.  It appears that we have a window of opportunity for Roth conversions between when she decides to retire and RMDs kick in. Looking at the rules, one way or another, we are probably going to be paying hefty income taxes well into old age

Probably so.  One minor optimization is if your wife retires early in a given calendar year, a Roth conversion that year (and every year between then and RMD age) might make sense.

[seattlecyclone]

We have LTCG in these early years and at that spending level, I can't get our income under 150% of the federal poverty level for a family of 3 ($32,580).

Why not? Just make the higher-income year be first instead of second. At the end of December in Year 1, realize a bunch of extra capital gains to fund your spending in Year 2. The only income that's really unavoidable is the dividends from your taxable account. VTSAX tends to pay about 2% (less this year though as dividends haven't kept pace with share prices), meaning you need to have in the ballpark of $1.5 million in taxable VTSAX for the dividends to approach your $32k number....which if that's the case, and taxable is only 35% of your portfolio, you can afford to spend a lot more than $85k annually.

I think the variable that is making my examples look so much worse than yours is the use of a bronze plan.

I don't think your choice of bronze vs. silver changes the math here very much. The subsidy is always calculated based on the price of the second-cheapest silver plan. The rate it phases out is the same regardless of what plan you pick, except at the lower end of the income range where your income is so low that the bronze plan is free, which means you're wasting some of the subsidy you could have had if you bought a better health plan. Even with some wasted subsidy in your low-income year it just impacts the bottom line up to the amount of wasted subsidy, which likely doesn't explain much of the discrepancy between your smooth vs. fluctuating income scenarios.

A bigger factor is just the way the marginal rates work out given the second-cheapest silver plan that results from your combination of age/location/family size. The fluctuating income scenario requires your high-income year to have enough income taxed at the lower post-ACA rates to make up for all of the income taxed at the higher ACA phase-out rates. That's not happening for you with the numbers you were using. Of note is that you're still getting some subsidy (and therefore paying a marginal rate of 22% + 8.5% = 30.5%) at $140k AGI, while the childless couple in Seattle phases out of subsidies at around $105k and pays just 22% from then on. Based on the unsubsidized insurance cost you cite in your $215k income scenario, I calculate that your subsidies won't phase out completely until about $210k. That means even if you did a three-year cycle with two low income years and one high-income year you would still probably pay more than if you kept your income even.

I'm tempted to make a blog post with some different graphs to illustrate this concept.

[seattlecyclone]

One consideration you aren't thinking here is once you start "losing" the refundable portions of credits, you are still paying a marginal tax rate even if your federal tax owed is still $0 or negative.

Going from $0 owed/$5k refund to $0/$0 still means you had a marginal tax rate.

That doesn't change the conclusion you came to necessarily but it does change how you are thinking about the "tax-free" aspect.

While true, the refundable portion of the child tax credit requires earned income, doesn't it? We have none, so we would forfeit that portion of the CTC regardless.

Not this year! The child tax credit is fully refundable in 2021. Current law would have us revert back to the previous rules next year, though the Democrats in Congress have stated a desire to extend this year's rules into the future. I wouldn't be at all surprised to see them stick that provision into a future reconciliation bill, especially if the monthly checks prove popular.

[Mr. Green]

New RMD tables.  Age 72 factor should be 27.3 if my spreadsheet is right:

https://www.taxwarriors.com/blog/updated-irs-life-expectancy-tables-mean-smaller-rmds-in-2022

As far as the standard deduction goes, I was looking at column AG on the first worksheet, but I just didn't look down far enough.

You didn't specifically reply about adjusting the standard deduction and tax brackets - they are increased by inflation every year, so you not adjusting them results in more aggressive taxation than will actually be the case, and apparently by quite a bit.  I'm 14 years older than you, so I have 14 years less of inflation adjustments before RMD age, and although the standard deduction for a single over 65 is $14,250 (I think that's the right 2021 number), by the time I start RMDs at age 72 it'll be $21,175.  (I assume 2% inflation.)

As mentioned before, to adjust your spreadsheet for that would add complexity though.  My spreadsheet just has the tax brackets in rows by year and does the basic tax calculations for each year in each row.  You could accomplish something similar by adding something like "...*1.02^(row()-3)" in cells like first worksheet cells AM4:AM66 whenever referring to thresholds from the 2021 taxes sheet.

(Of course, with the way things go these days, trying to predict taxes beyond the current year might be viewed as a fool's errand, so you could wave it away with that.  You might also just prefer to have the bias in there.  It also might not be that bad if you're going to update the worksheet yearly with the new constants on the "2021 taxes" sheet, as I think your situation will tend towards a reasonable optimum if you iterate each year.)

I don't update any of the tax thresholds. I'm kinda operating under the assumption that the tax brackets will increase similarly to the standard deduction and that both of those will track general inflation. I know that could get out of whack a little, but I don't know how I could effectively add those subtle differences in.

I have the model set up in general for no inflation because I didn't want to be looking at values 50 years from now and have a hard time putting that in perspective because we think in today's dollars and relationship to material costs

[Mr. Green]

We have LTCG in these early years and at that spending level, I can't get our income under 150% of the federal poverty level for a family of 3 ($32,580).

Why not? Just make the higher-income year be first instead of second. At the end of December in Year 1, realize a bunch of extra capital gains to fund your spending in Year 2. The only income that's really unavoidable is the dividends from your taxable account. VTSAX tends to pay about 2% (less this year though as dividends haven't kept pace with share prices), meaning you need to have in the ballpark of $1.5 million in taxable VTSAX for the dividends to approach your $32k number....which if that's the case, and taxable is only 35% of your portfolio, you can afford to spend a lot more than $85k annually.

Fair point! Though our income in the second year is over $172,750 so our capital gains are no longer free. So moving those capital gains from the lower-income year would generate additional tax. I think this is only an issue with a substantially higher income. In the example you provided, shifting capital gains would likely still keep the big income year under $172,750 AGI so you could double up with no effect.

Though I could also see the possibility that you'd still be getting some ACA premium subsidies and those shifted LTCG would cost you that as well. Our SLCSP for three people is $1,496 per month, $17,952 per year, which at an upper limit of 8.5% of AGI, means premium subsidies don't fully phase out until 211,200.

I don't think your choice of bronze vs. silver changes the math here very much. The subsidy is always calculated based on the price of the second-cheapest silver plan. The rate it phases out is the same regardless of what plan you pick, except at the lower end of the income range where your income is so low that the bronze plan is free, which means you're wasting some of the subsidy you could have had if you bought a better health plan. Even with some wasted subsidy in your low-income year it just impacts the bottom line up to the amount of wasted subsidy, which likely doesn't explain much of the discrepancy between your smooth vs. fluctuating income scenarios.

Ah, you're right. I don't know what I was thinking there.

[boarder42]

One consideration you aren't thinking here is once you start "losing" the refundable portions of credits, you are still paying a marginal tax rate even if your federal tax owed is still $0 or negative.

Going from $0 owed/$5k refund to $0/$0 still means you had a marginal tax rate.

That doesn't change the conclusion you came to necessarily but it does change how you are thinking about the "tax-free" aspect.

While true, the refundable portion of the child tax credit requires earned income, doesn't it? We have none, so we would forfeit that portion of the CTC regardless.

Not this year! The child tax credit is fully refundable in 2021. Current law would have us revert back to the previous rules next year, though the Democrats in Congress have stated a desire to extend this year's rules into the future. I wouldn't be at all surprised to see them stick that provision into a future reconciliation bill, especially if the monthly checks prove popular.

I'm highly confident most of the last stimulus will be put into law for the future including child tax credit and dependent care credits. In addition to the ACA subsidy changes and elimination of the cliff.  If I'm wrong well I'll just go on Sedera

[Laura Ingalls]

This is the dance we have been dancing.  We have found alternating years helpful for reasons not mentioned on this thread.  We have a trickle of regular w-2 income.  We also have two offspring. One on high school and the other in a two year technical program.  Low years are good for FAFSA and EITC.   High years have involved selling appreciated stocks in taxable so that we have enough money to finance our lifestyle. 

Last year was not a FAFSA year as we will not have anyone in higher education during the 2022-2023 school year.  This year counts for younger offspring’s first year of college. 

Lots of moving parts and tax changes to digest too.  In the end there is probably no optimal just directional better or worse.

[ender]

While I understand that the government can do anything they damn well please, I have a hard time believing they would tax straight Roth withdrawals. That would be double taxation since we've already paid tax once on conversions. Also, if they decided to start taxing Roth gains, it's basically the same thing as a tIRA at that point so why not just eliminate the Roth account entirely. Are there other factors here I'm not thinking of?

A fairly reasonable situation I can think of is withdrawals of Roth gains being included in calculating things such as ACA subsidies, tax credits, and SS taxation.

There are of course more alarmist perspectives you could take such as wealth taxes factoring in your Roth accounts or straight up different taxation on Roth withdrawals (similar to how capital gains are taxed differently).

Personally, I think the likelihood of actual direct taxation is low. But I do think the chances Roth is indirectly taxed to be reasonably high in the timelines we talk about for FIRE.

[Mr. Green]

While I understand that the government can do anything they damn well please, I have a hard time believing they would tax straight Roth withdrawals. That would be double taxation since we've already paid tax once on conversions. Also, if they decided to start taxing Roth gains, it's basically the same thing as a tIRA at that point so why not just eliminate the Roth account entirely. Are there other factors here I'm not thinking of?

A fairly reasonable situation I can think of is withdrawals of Roth gains being included in calculating things such as ACA subsidies, tax credits, and SS taxation.

There are of course more alarmist perspectives you could take such as wealth taxes factoring in your Roth accounts or straight up different taxation on Roth withdrawals (similar to how capital gains are taxed differently).

Personally, I think the likelihood of actual direct taxation is low. But I do think the chances Roth is indirectly taxed to be reasonably high in the timelines we talk about for FIRE.

If Roth gains are the most likely point of attack then we'd get a reprieve until 60, which is most of our early retirement. Given how minimal the withdrawal of Roth gains are before 60, it seems like it wouldn't be very effective for influencing ACA subsidies or tax credits. Almost everyone is done with kids, careers are mostly over, and ACA participation is mostly over by that point. I mean, they could still do this if they want to target the likes of Peter Theil's $5 billion Roth IRA but that's not going to affect me one bit.

Do you imagine this would happen as an attempt to end the Roth IRA without actually killing it? Taxing gains would make a Roth IRA no different than a taxable account. Why would people bother contributing to one at that point? Any taxation of withdrawals would make it the worst investment vehicle available because of the double taxation, also immediately eliminating contributions/conversions. I guess I'm struggling to understand the political motivation. Just because?

[ender]

Do you imagine this would happen as an attempt to end the Roth IRA without actually killing it? Taxing gains would make a Roth IRA no different than a taxable account. Why would people bother contributing to one at that point? Any taxation of withdrawals would make it the worst investment vehicle available because of the double taxation, also immediately eliminating contributions/conversions. I guess I'm struggling to understand the political motivation. Just because?

This is assuming that Roth IRA taxation would be exactly what capital gains taxation. Surely you can imagine ways it might be different (maybe it's a 10% tax  with graduation similar to capital gains, etc, lots of possibilities).

Anyways, my point here is assumptions go into many things in this equation. Your assumption here (which you obviously believe quite strongly) is that Roth gains will never have a negative tax impact and always remain 100% tax consequence free.

That's fine. Just make sure you recognize it as an assumption and analyze risk accordingly.

[Mr. Green]

Do you imagine this would happen as an attempt to end the Roth IRA without actually killing it? Taxing gains would make a Roth IRA no different than a taxable account. Why would people bother contributing to one at that point? Any taxation of withdrawals would make it the worst investment vehicle available because of the double taxation, also immediately eliminating contributions/conversions. I guess I'm struggling to understand the political motivation. Just because?

This is assuming that Roth IRA taxation would be exactly what capital gains taxation. Surely you can imagine ways it might be different (maybe it's a 10% tax  with graduation similar to capital gains, etc, lots of possibilities).

Anyways, my point here is assumptions go into many things in this equation. Your assumption here (which you obviously believe quite strongly) is that Roth gains will never have a negative tax impact and always remain 100% tax consequence free.

That's fine. Just make sure you recognize it as an assumption and analyze risk accordingly.

The thought of Roth funds becoming taxed in new ways has never crossed my mind before now, hence the questioning. Why is it that you feel so strongly that this will inevitably happen, because of the Peter Theil/Mitt Romney type examples?

I'm definitely going to take a fresh look at things with future taxation in mind but my options are somewhat limited since two thirds of our money is tied up in IRAs. It might end up being a case of "well that sucks that I didn't see that coming but since I've exhausted all our taxable funds I don't have much choice but to press forward."

[ender]

The thought of Roth funds becoming taxed in new ways has never crossed my mind before now, hence the questioning. Why is it that you feel so strongly that this will inevitably happen, because of the Peter Theil/Mitt Romney type examples?

Huh?

I've never been saying anything along these lines.

This is what I initially said, which kicked off this back/forth:

I do not think it's likely Roth withdrawals will be directly taxed but I would not be surprised if they change to for example impact your ACA subsidies or SS taxation. There is an assumption in a lot of FIRE calculations that Roth growth never will negatively impact your financial situation or have any tax impact.

[Mr. Green]

Personally, I think the likelihood of actual direct taxation is low. But I do think the chances Roth is indirectly taxed to be reasonably high in the timelines we talk about for FIRE.

I'm not trying to get into an argument about it. I was just wondering what info has you forming that opinion. I'm considering the idea of future Roth taxation for the first time here so any input is input worth looking at, as I currently have none.

[BicycleB]

A fairly reasonable situation I can think of is withdrawals of Roth gains being included in calculating things such as ACA subsidies, tax credits, and SS taxation.

...

Personally, I think the likelihood of actual direct taxation is low. But I do think the chances Roth is indirectly taxed to be reasonably high in the timelines we talk about for FIRE.

I hadn't thought of this, but it does seem plausible. It wouldn't invalidate the promise of tax free withdrawals or modify the Roth statute at all. It would just modify other benefits so that people with multiple resources (specifically Roth holders) don't get quite as much "dole" from ACA subsidies, SS tax breaks and so on as people lacking those same resources.

Re motivations, I can certainly imagine a continued trend toward viewing Roths as the province of fat cats like Thiel, rather than a tool of the everyday person. At that point, a populist party or a social justice party could each have reason to "stop fat cats from profiting at the trough of benefits for the needy" or some such.

Not saying it will happen, just agreeing that over 30-40 years, it easily could happen.

[ender]

^ is about what I'd say as well. We're talking timelines for most of us that are 30-50+ years if we FIRE.

That is a long time for taxes to materially stay the same.

For reference, Roth has only been around since 1997 (24 years).

[Mr. Green]

I spent yesterday thinking about @ender's comment about Roth taxation changing, pondering how I might modify my FIRE spending strategy to mitigate that risk, or perhaps even just diversify a bit. In my model we spend down taxable accounts until they're exhausted while doing Roth conversions. Once that bucket is empty we transition to spending Roth IRA principal until age 60. After 60 we spend Roth principal and gains, with RMDs from our tIRAs at age 72 and beyond.

Hypothetically speaking, accessing anything but Roth principal generates taxable income. Taking tIRA withdrawals early, Roth conversions, taxable withdrawals (assuming the account is equities w/ capital gains). Assuming one's taxable equities aren't approaching 100% capital gains, taxable funds represent the most access to your money for the same taxable income. tIRA withdrawals and Roth conversions are 1:1 with how much income they create. Roth principal is free but the total amount available in our case is low because we were high earners so our Roth account balances are small. The spread on our taxable dollars diminishes as capital gains go up but right now it's close to a 2:1 ratio of accessible money to income generated.

The value of that broader access was very apparent in the Spring of last year when we thought we'd have to buy a house with cash because a bank wouldn't lend to us (no "legit" incomes). We would not have been able to access a couple hundred thousand dollars without paying significantly higher taxes for the year if every dollar was taxable income.

So I thought perhaps I was approaching the model the wrong way. If I'm trying to give us access to as much of our money at any given time as possible without triggering penalties, etc. wouldn't it make more sense to leave as much of our taxable funds intact as possible? So I made a second model that uses taxable funds only until our Roth principal is large enough to cover spending and then we convert to Roth funds until age 60. However, we stop conversions at age 55 and the 5-year rule carries us those last 5 years. During those 5 years, we generate our income from long-term capital gains since our taxable accounts are highly appreciated now that we haven't used those funds in almost 20 years.

It shouldn't have come as a surprise but the newer model gave us much greater access to our money leading up to age 60. In the first model, since we had exhausted our taxable accounts, our only access was to any Roth principal we hadn't already spent. In the second model, we'd been spending Roth principal as fast as we were building it so that bucket was still very small but our taxable accounts had been left to grow. We have access to both principal and gains there so the difference was twice the available money, 800k vs. 400k.

Once we get past 60, tIRA funds and Roth IRA gains are available penalty-free so our whole portfolio opens up.

In the process of creating the second model, I made several improvements to the data set. Previously, I had capital gains as a static percentage of our taxable accounts. I knew this became inaccurate as years of gains accumulated so I addressed that by adding a couple columns to update the basis in our taxable accounts as money is spent (or not) each year. I also added a flag so that I could harvest capital gains (incurring the taxable event without actually spending down the taxable account). In the process of doing all this, I also had to change some formula inaccuracies for different calculations. I hadn't designed the model to withstand a higher income; I assumed our spending would stay low in the name of yearly tax efficiency. Now that I've changed what I'm asking the model to show me some faults were exposed, like not properly calculating long-term capital gains when total income after the standard deduction pushes into the third tax bracket.

Another interesting effect of the second model is that we end up with a much more even distribution of funds across taxable, tIRA, and Roth IRA buckets in the run-up to 60. I mentioned in an earlier post that right now our allocation is 60% tIRA/35% taxable/5% Roth. In the first model, at 60 our allocation is 20% tIRA/0% taxable/80% Roth. In the second model, our allocation at 60 is 22% tIRA/42% taxable/36% Roth. Both models basically have the same total portfolio balance at age 60. They're within 1% of each other.

I'm still digesting but I think the second model is the better path. The better distribution of funds across account types would provide more flexibility toward future legislative changes. Maybe Roth IRAs do get taxed in a way that affects us. Maybe long-term capital gains start getting taxed in the lowest two tax brackets. Perhaps the biggest concern with the first model is that as soon as we exhaust our taxable accounts, the only penalty-free way we have to generate income is Roth conversions. If future legislation eliminated the Roth conversion process we'd have no choice but to pay penalties on IRA withdrawals to support our spending until age 60. The model exhausts our taxable accounts at age 49. That's a long time until 60. Yikes!

Edit: typo

[BicycleB]

Another interesting effect of the second model is that we end up with a much more even distribution of funds across taxable, tIRA, and Roth IRA buckets in the run-up to 60.

That does seem more robust. You can vary details as conditions become apparent. Maybe the best thing now is the action path that preserves the biggest variety of options.