Looking for the magic formula for Roth conversions vs RMDs

[ILikeStarTrek]

I thought, at first, that that model looked a bit odd, so I went back and looked at it some more, and I think it makes sense now. It is telling you to do about 5 years of IRA2ROTH conversions, going into the 22% bracket to do so. After that, your IRA balance becomes low enough that your RMDs will never cause you to exceed the 12%/15% bracket, and so no more conversion is desirable.

I emailed iORP, and within an hour or so they got back to me and fixed the bug, telling me to re-run. Indeed, it works now for the ones that were failing before.

One quick experiment I did now that it's working is to run with no Roth conversion vs 22% and 32% Roth conversion. My income with Roth conversions goes up by $1-2k compared to no conversions, so doing the conversions helps. I think the assets are low enough that as you said, it's not really taking the conversions up to the 32% bracket even if I tell it to go that high. But it's doing some conversions and even though they don't make a dramatic difference, they help some.

This was just an initial experiment and now that I see how useful this tool is, I want to play around for it. For instance:

- It doesn't seem to allow one to specify the max income that one wants, but rather generates the max income where one runs out of money at the end of the planning horizon. I can kind of get around that by giving a planning horizon of 120. I don't expect to live this long of course, but by doing this, when I am in my 90s and even early 100s, there is still plenty of money left so it makes the tool essentially generate its own max income for me where my assets won't run out in my lifetime.

- The default assumption of stock return is 6%. If I change that to 8 or 9%, it changes the results a lot! I also need to play with a lower than 100% stock allocation, which is the tool default. I am pretty aggressive in my stock AA but it's not quite 100%.

- I also have some real estate rental income that I haven't told the tool about. I thought of entering it as a pension but that wouldn't be quite right either since, for instance, $30k of real estate cash flow might correspond to $10k of taxable income. In that respect, telling the tool there is a 10k/yr pension would get the tax numbers right; but that doesn't quite capture it either, because I view the real estate cash flow as indexed to inflation so $20k extra wouldn't reflect the cash flow 20 yrs from now. I don't think there is a way to tell the tool I have some non-taxable income indexed to inflation.

So I have a good bit more to experiment with. But this tool is a very good start and I'm glad to know about it!

 

[Huston55]

I emailed iORP, and within an hour or so they got back to me and fixed the bug, telling me to re-run. Indeed, it works now for the ones that were failing before.

One quick experiment I did now that it's working is to run with no Roth conversion vs 22% and 32% Roth conversion. My income with Roth conversions goes up by $1-2k compared to no conversions, so doing the conversions helps. I think the assets are low enough that as you said, it's not really taking the conversions up to the 32% bracket even if I tell it to go that high. But it's doing some conversions and even though they don't make a dramatic difference, they help some.

This was just an initial experiment and now that I see how useful this tool is, I want to play around for it. For instance:

- It doesn't seem to allow one to specify the max income that one wants, but rather generates the max income where one runs out of money at the end of the planning horizon. I can kind of get around that by giving a planning horizon of 120. I don't expect to live this long of course, but by doing this, when I am in my 90s and even early 100s, there is still plenty of money left so it makes the tool essentially generate its own max income for me where my assets won't run out in my lifetime.

- The default assumption of stock return is 6%. If I change that to 8 or 9%, it changes the results a lot! I also need to play with a lower than 100% stock allocation, which is the tool default. I am pretty aggressive in my stock AA but it's not quite 100%.

- I also have some real estate rental income that I haven't told the tool about. I thought of entering it as a pension but that wouldn't be quite right either since, for instance, $30k of real estate cash flow might correspond to $10k of taxable income. In that respect, telling the tool there is a 10k/yr pension would get the tax numbers right; but that doesn't quite capture it either, because I view the real estate cash flow as indexed to inflation so $20k extra wouldn't reflect the cash flow 20 yrs from now. I don't think there is a way to tell the tool I have some non-taxable income indexed to inflation.

So I have a good bit more to experiment with. But this tool is a very good start and I'm glad to know about it!

Happy it’s now working for you.

Regarding your RE income, isn’t it sheltered by expenses & depreciation?

 

[ILikeStarTrek]

Regarding your RE income, isn’t it sheltered by expenses & depreciation?

Most of it is sheltered, but it still comes out a net positive. That's why I said about $30k of cash flow can look like $10k of taxable income.

 

[Curmudgeon]

- The default assumption of stock return is 6%. If I change that to 8 or 9%, it changes the results a lot! I also need to play with a lower than 100% stock allocation, which is the tool default. I am pretty aggressive in my stock AA but it's not quite 100%.

If I'm not mistaken, the default analysis just assumes constant returns year after year. This is pretty optimistic - it exposes you to sequence of returns risk - and so you will get much higher yearly spending results than you would in, say, FireCalc, which looks at more realistic sequences of returns. I haven't played around yet with iORP's Monte Carlo analysis, but this might be a good next step.

So then, a major concern (given the above), is this: If realistic, or worst-case, sequnce of returns are considered, are the ira2roth conversions recommended by iORP still 'optimal'? Selling stocks during a bear market, in order to do Roth conversion, could in many cases be the wrong thing to do.

 

[Jackanory]

I'm only age 55 so I have quite a few years between now and RMDs, but it seems like one cannot start planning too soon since it could take a number of years to implement a gradual conversion strategy to avoid RMDs.

You should probably hold off until you're 59.5, lest you incur the 10% early withdrawal penalty -- should give you plenty of time to puzzle things out!

 

[ILikeStarTrek]

So then, a major concern (given the above), is this: If realistic, or worst-case, sequnce of returns are considered, are the ira2roth conversions recommended by iORP still 'optimal'? Selling stocks during a bear market, in order to do Roth conversion, could in many cases be the wrong thing to do.

On the other hand, selling stocks during a bull market to do Roth conversion could be the right thing to do. It is often assumed that when one runs out of money in a withdrawal-rate scenario, it happens during the 1st 10 years, but often that is not the case in the tool projections. If the bear market happens right after I turn 70.5 then I still have all the high assets previously accumulated, for taxes via RMDs. This gets back to where it gets pretty fuzzy for me. In any case, I agree that more research is needed and I too want to play with the Monte Carlo analysis in iORP.

E.g. another thing is, even with constant annual stock returns, the amount of Roth conversion will be different if one projects 6% return vs 10% return. I believe I read that the 20-yr S&P return is closer to the 6% while the 30 or 40 yr return is closer to the 10%. So when using a tool such as this, which figure does one use in deciding on Roth conversions? Since we're ultimately guessing at the future, maybe there comes a point where we look at multiple possibilities and pick either what we consider an average one or most likely one.

You should probably hold off until you're 59.5, lest you incur the 10% early withdrawal penalty -- should give you plenty of time to puzzle things out!

My understanding is that a Roth conversion can be done at any age without penalty. It's not the same as a withdrawal. Incidentally, a withdrawal from a 401k plan of a company that one worked for until age 55 is allowed without penalty. Here the rule is different from the age 59.5 rule for an IRA or a previous company's 401k.

 

[Curmudgeon]

On the other hand, selling stocks during a bull market to do Roth conversion could be the right thing to do.

Absolutely! I'm just pointing out the limitations of the model used by ORP, and situations where what it recommends would not actually be optimal.

My understanding is that a Roth conversion can be done at any age without penalty. It's not the same as a withdrawal.

That's correct.

 

[pb4uski]

......Selling stocks during a bear market, in order to do Roth conversion, could in many cases be the wrong thing to do.

Not sure if I agree or disagree, but in any event that "problem" is very easily avoided by just selling stocks in the tIRA and buying the same stocks in the Roth. That is what I have done for years.

 

[Curmudgeon]

Not sure if I agree or disagree, but in any event that "problem" is very easily avoided by just selling stocks in the tIRA and buying the same stocks in the Roth. That is what I have done for years.

No, I'm talking about the stock you would need to sell in order to pay the taxes on your Roth conversion. Regardless of how you get it, it needs to be subtracted from your net worth, is gone forever, and compounds the sequence of returns problem.

 

[pb4uski]

If I'm not mistaken, the default analysis just assumes constant returns year after year. This is pretty optimistic - it exposes you to sequence of returns risk - and so you will get much higher yearly spending results than you would in, say, FireCalc, which looks at more realistic sequences of returns. I haven't played around yet with iORP's Monte Carlo analysis, but this might be a good next step.

So then, a major concern (given the above), is this: If realistic, or worst-case, sequnce of returns are considered, are the ira2roth conversions recommended by iORP still 'optimal'? Selling stocks during a bear market, in order to do Roth conversion, could in many cases be the wrong thing to do.

Actually no... if I run Firecalc using the exact same assumptions as I use for i-orp I get higher spending with Firecalc.

 

[pb4uski]

No, I'm talking about the stock you would need to sell in order to pay the taxes on your Roth conversion. Regardless of how you get it, it needs to be subtracted from your net worth, is gone forever, and compounds the sequence of returns problem.

No again... in my case it reduces cash, which is replenished when I rebalance.

IMO, your concern is minor in the scheme of things as long as one manages their AA to reasonable bounds

 

[Curmudgeon]

Actually no... if I run Firecalc using the exact same assumptions as I use for i-orp I get higher spending with Firecalc.

Hmm, interesting; I saw just the opposite and I assumed it was due to (1) more efficient taxation from iORP and (2) the assumed optimism I described. I'll have to take another look to see if I can figure out what's going on.

 

[pb4uski]

Actually... strike that... I just noticed that the spending in the ORP report is described as after-taxes. If I gross it up for taxes to be consistent with FIRECalc then my FIRECalc 95% success rate spending (which includes taxes) is ~85% of ORP's spending grossed up for taxes.

In any event, I would use FIRECalcs stochastic analysis for deciding withdrawals rather than ORP's or QLP's deterministic analysis.

I think ORP is interesting to see the tax impact of Roth conversions. It is suggesting that I do higher Roth conversions... to the top of the 22% bracket for 2018 to when we start SS, and then we'll be in the 15% tax bracket after that. I think it is reverting the old tax brackets after the one in the new law expire and figures that 22% now is better than 25% later. Something to think about for me.

 

[Curmudgeon]

Actually... strike that... I just noticed that the spending in the ORP report is described as after-taxes.

I think I was missing some revenue streams (SS/pension) when I ran FC - with those streams back in, the numbers are much closer.

In any event, I would use FIRECalcs stochastic analysis for deciding withdrawals rather than ORP's or QLP's deterministic analysis.

Yes, definitely. Always use the right tool for the right job.

 

[Brat]

My analysis is much more basic. DH is kicking 80 so hard his toes are bruised so his MRDs are fast escalating. We don't have a lot of taxable income right now so recharacterizing ~ $10T from his traditional IRA to his Roth at 12% is chump change, in 2 years we are $ ahead and our combined lifespan is much more than 2 years. The kids, or grandkids, will inherit both traditional and ROTH IRAs. Would they rather pay income taxes on $10T in the traditional IRA or inherit $8.8T money via a Roth?

If we need ROTH $ at the end of life we will be will be at max RMDs from our traditional IRAs, higher tax brackets, and having tax free accounts to tap would be optimal.

Also, the lower marginal tax rates will likely be 'readjusted' in the future.

 

[MrLoco]

It depends on what your other income is before the Roth conversions and what constraints exist. For example, if you were living off a huge slug of taxable savings in low-interest FDIC savings accounts then you might have some room for Roth conversions even if ACA subsidies is a constraint. There's probably no getting around the tax torpedo depending on your situation.

What I find best is start with constraint, then work backwards to the Roth conversion. The also test what happens if you do Roth conversions beyond the constraint and what the "net" economic cost of those conversions is.

I understand what you are saying. But for those who do not wish to convert Trad to Roth IRA for fear of losing an ACA subsidy, I think this potentially carries a huge opportunity cost. For example:
Person A has $1M in a bank savings account (money market, CD, whatever) earning 0.5% ($5,000/year). This allows him to keep $7,000 in an ACA subsidy.
Person B had $1M invested in an S&P 500 fund in 2017 for a return of $210,000....but gave up $7,000 in an ACA subsidy.
YEs....this is an extreme case but you see the point. Don't give up returns which could pay for an ACA subsidy many times over.

 

[MrLoco]

I think the best course of action may be to convert some of your Trad IRA to Roth's ....but not all. This is the course of action I am taking since in later years I want to have assets in 3 types of accounts: Taxable; tax deferred, and tax free. And then withdraw as tax efficiently as possible.

I also feel that Uncle Sam will (hopefully) allow medical/dental deductions in future years.
( as they kept this provision in the new tax bill).
Hence, if I need Long term care or have high out of pocket healthcare expenses later in life; I should be able to offset taxes due from a Trad. IRA withdrawal to help pay for my LTC needs.
We shall see.

 

[bobbee25]

In my case I started converting IRAs to Roths after age 60. I determined how much I could convert without jumping into a high tax bracket.

It has worked great as now my RMD is low enough that I don't have to withdraw more than I actually require, I could withdraw more if needed.
If I hadn't converted I would be withdrawing more than needed and putting it in a taxable bank account. The rmd would be added to my pension and social security resulting in higher taxes.

Now all my monies grow tax free.

 

[pb4uski]

I understand what you are saying. But for those who do not wish to convert Trad to Roth IRA for fear of losing an ACA subsidy, I think this potentially carries a huge opportunity cost. For example:
Person A has $1M in a bank savings account (money market, CD, whatever) earning 0.5% ($5,000/year). This allows him to keep $7,000 in an ACA subsidy.
Person B had $1M invested in an S&P 500 fund in 2017 for a return of $210,000....but gave up $7,000 in an ACA subsidy.
YEs....this is an extreme case but you see the point. Don't give up returns which could pay for an ACA subsidy many times over.

You misconstrued what I wrote... I wasn't suggesting that someone go with a really conservative AA to lower their income for ACA subsidies.... I was just trying to design a scenario where their income before any Roth conversions was low so there was room to do Roth conversions and still qualify for ACA subsidies.

A better example might be where they have some taxable funds but have substantial tax-deferred funds and enough in Roths to live off... they would have no income before any Roth conversions so they would start the the ACA constraint, reduce it by the income on their taxable accounts and the Roth convert the difference.

 

[Scuba]

I'll explain what I have been doing and why.

The why first. I retired at 56 and have been living principally from taxable account savings. After our SS starts I expect we will be in the 25% tax bracket (22% under the new law) or perhaps even higher. Before our SS starts, we would not pay taxes as deductions/exemptions exceed our interest and my small pension and our qualified dividends and long-term capital gains are not subject to tax. So, I have a window of time from when I ERd until our SS starts that I can do Roth conversions and pay little in tax.

With that background, you'll read that many of us are doing Roth conversions to the top of the 15% tax bracket (12% under the new tax law). Since your OP said "I" rather than "we", I'll go through the calculations for a single for 2018 under the new tax law.

The top of the 12% tax bracket for a single person in 2018 is $38,700. The total income to get 0% capital gains is $38,600. It is stoopid that these numbers are different, but it is what it is.... I'll chose the lower of the two... $38,600. The standard deduction is $12,000. The sum of these two is $50,600. So if you had no other income, you could have $50,600 of Roth conversions and would pay $4,441.50 in tax... 8.8% of the Roth conversion... most likely a lot lower than the tax you avoided by deferring that income while you were working and also lower than what you would later pay if you took that money later once SS starts as a withdrawal, RMD or Roth conversion.

So in my case where I expect to pay 22% later, I'm saving 13.2%.. pretty substantial.

Now in reality, interest income, dividends, pension income etc effectively reduce the amount that you can convert and remain in the 12% tax bracket. So if you had $10,000 of income from other sources, you could only convert $40,600.

Over the last 5 years, I have converted almost 25% of my tax-deferred balances on the day that I retired and paid about 7.4% of the amount converted in federal tax. that's the good news.... the bad news is that because of growth my tIRA balance today is still a lot higher than when I retired despite converting that 25%... sort of a nice problem to have. Be aware, if you live in a state that has a state income tax then there will likely be state income tax on conversions too.

P.S. If your income includes qualified dividends or long-term capital gains then you will not want to exceed $50,600 because for each $100 the increase in tax is $27... a 27% incremental tax rate... this is because $100 of ordinary income is taxed at 12% but it pushes $100 of qualified income above the limit and that $100 of qualified income is taxed at 15% (for a total of 27%).
__________________

Pb4uski,
I found the above post very helpful. Would you please post a similar example but using numbers for married filing jointly status? Thank you!