I'm seeking feedback form the Pralana community as to whether the way I'm looking at Roth conversions seems reasonable - you may find this a useful way to think about the economic return on Roth conversions. I'm using an internal rate of return (IRR) calculation to see if Roth conversions are justifiable or not for just calendar year 2025 - I'm not going to consider future Roth conversions at this time for various reasons. Here's a description of the process. I run two scenarios using Pralana online. Scenario 1 is no Roth conversion. Scenario 2 is a Roth conversion for just 2025 using whatever constraints I prefer (could be no constraints). All other scenario inputs are identical, so the only factor is the amount of Roth conversion. I then observe the balance sheet at the end of the projection and subtract the terminal Scenario 1 effective balance from the Scenario 2 effective balance and call that my "additional return". I then observe the difference in total taxes paid between Scenario 1 and Scenario 2 in calendar year 2025. I call that my "initial investment". I then calculate the internal rate of return of investing the "initial investment" and realizing a return at the end of the projection period of an amount equal to the "initial investment" plus the "additional return" over a period of years equal to the number of years in the projection. The concept is simply this: (1) you invest an amount of cash flow equal to the difference in total taxes paid in the year of the Roth conversion; (2) at the end of the projection period, you realize an amount equal to your "initial investment" plus any excess effective balance of Scenario 2 over Scenario 1. If the two terminal scenario effective balances were exactly equal, you were just paid back your "initial investment" and had a 0% return. For my particular situation, I found that after 27 years, I received an internal rate of return of only about 2% per year for the outlay of extra taxes paid for a Roth conversion in 2025. I concluded that it wouldn't make sense for me to risk that initial investment for such a small expected return given the uncertainty of future tax rates and other important variables.
Does this IRR approach seem reasonable, or do you have an improvement to suggest?
Thanks, Joel
Since the money to pay the taxes came out of the portfolio, this 2% return per year on the taxes paid is over and above the rate of return of the portfolio. Is there some alternative you can think of where you can make 2%/year more than your portfolio return without taking excessive risk? If so, sign me up too! Over 27 years, that 2%/year means the taxes paid recouped their value plus the portfolio growth rate plus 71%.
That said, I do not actually see IRR as the approach to understanding Roth Conversion math. The tax obligation in your IRA grows along with the investments you hold in that account, so just because your plan is to not access the money for a long time, does not make the obligation go away, the obligation keeps growing. Instead, I would start the evaluation by understanding your goals for the money.
If you have heirs you care about, they will end up paying the tax, so the best way to evaluate Roth Conversions in that case is to set an Effective Tax Rate as an estimate of the rate your heirs will pay upon your death. When you do the full circle accounting like that, you will see that you are never behind with a Roth Conversion that is done at the same tax rate as you/your heirs will pay later, so the concept of an investment doesn't make sense, the commutative property of multiplication means that if you know your tax rate, it doesn't matter when you pay it, you will end at the same spot.
If you are giving your IRA away via QCDs and charitable bequests, then using zero for the Effective Tax Rate is fine, but Roth Conversions inherently wouldn't make sense as the best money to give to charity is money that's never been taxed.
If you do not have heirs you care about and are not giving it away to charity, then you can set the Effective Tax Rate to zero, but you should be enjoying your money while you are alive instead of letting people you don't care about blow your dough after you're gone. In that case, set a very great age, like 100, for your planning horizon and then use a variable withdrawal method designed to exhaust the portfolio at that age (maybe leaving an modest allowance for long term care). The typically used methods in Pralana for that are Actuarial or Consumption Smoothing. If you die before then, nothing lost, you were spending the maximum safe amount, if you surprise yourself and live to a great age, you've planned for that.
A simple thing you can do to minimize the need for Roth Conversions in the first place is to put your bonds preferentially in your tax deferred account. That will slow the growth of that account, reducing RMDs. This is tax efficient as it matches the taxation of bond dividends (ordinary income) to that of the account, leaving your taxable space for your stocks, where the primary growth mechanism is taxed as long term capital gains. This is way pros will tell you to hold your money and the good news is that you picked the right calculator tool. Other tools can't do the math right, but you can with Pralana using Advanced Portfolio Modeling - Mode 2.
The other thing you left out of your analysis is the reduction in tax drag that occurs when you do Roth Conversions and pay taxes out of taxable. That gives you an immediate reduction in tax drag, that keeps growing year after year. Then when RMDs start, it also reduces any accumulation of funds in taxable that would cause further tax drag. If you have heirs you care about, the Roth money doesn't need to come out until the 10th year, whereas the inherited IRA money had RMDs to come out in 10 years, so Roth money stays out of your heirs taxable accounts for an average of 5 years longer. The benefits of tax drag savings start very small but really add up over time. In our case, shrinking dividends means making it easier to avoid IRMAA surcharges, shrinking NIIT obligations and reducing the SS tax torpedo where the phase in of taxes on SS benefits turns the 22% bracket into 40.7%.
@ricke Thank you for your comments. In particular, you are correct in observing that the IRR I was calculating is actually an INCREMENTAL IRR, not an ABSOLUTE IRR.
On a technical point, it's not an incremental IRR above the portfolio return. It's an incremental return above a complex blend between (1) cash return (initially, added taxes come out of cash), and subsequently, (2) the portfolio investment return, depending upon annual cash flows and floor/ceiling cash constraints. One can actually measure the opportunity cost of paying extra Roth conversion taxes by running a third scenario in the Pralana calculator with no Roth conversions and adding the extra tax as a miscellaneous expense; the reduction in the terminal effective balance is the accumulated opportunity cost. This could be used to make the IRR calculation more precise.
In the real world, heirs will not always come out ahead right away, or possibly even at all, because your heirs' effective tax rate and your marginal tax rate for the Roth conversion will not always align - nor should they have to. My scenarios bear this fact out. For me at least, the "effective balance" break-even year takes a while to show up in the projections. For example, the Roth conversion optimizer may throw you into a higher marginal tax bracket than the effective tax rate that you choose for your heirs. And even when I add some constraints to tamp down Roth conversions, I find this break-even year delay to still be true. I suspect my situation is not unique.
You also questioned whether there was any risk at all to making a Roth conversion. Here too, I disagree. While this is not tested in the Pralana calculator, what if immediately after taking a large Roth conversion your portfolio declines by a significant amount, and doesn't recover for many years (this has happened in the past). There will be a lot less traditional IRA balance remaining from which to generate tax savings. This sequence of returns risk is very real and could definitely make a Roth conversion uneconomic. Perhaps more importantly, I think there may be a great deal of uncertainty around what future tax rates will be for you and your heirs. Not only may tax rates change in the future, but your heirs may inherit IRA's and other assets from other family members, which could throw them into higher tax brackets. Or maybe your heirs are forced to take a lower-paying job and find themselves in a lower tax bracket. I found that changing the effective tax rate makes a huge difference in whether a Roth conversion makes sense or not. Finally, none of us knows our "expiration date". And assuming that my assertion that the break-even year takes a while to show up in the real world, it may turn out that the Roth conversion doesn't improve the tax situation for the heirs if you expire before the break-even point.
All this is to say that I believe there's a great deal of risk/uncertainty around the benefits of a Roth conversion, and one should demand some type of "risk premium", however you want to measure that amount. I'm thinking about the IRR as one potential measure of the added return for taking that risk. If you believe that you can set up a Roth conversion that is bullet proof from an economic standpoint, I applaud your skill, and agree that you wouldn't need any risk premium as compensation.
Joel
Not sure I understand your test case, but to isolate the effects of Roth Conversions, if you keep your asset allocation and cash holding constant, you can isolate Roth Conversion effects. If you think I said there was no risk, re-read my words, I specifically said "at the same tax bracket" as I was trying to illustrate that the typical gap in the argument for people that start with an IRR approach is to ignore about the tax liability on the residual balance. If you account for that and want to use IRR as a way to judge what to do, that works, I approach it somewhat differently, but it's all about trying to judge the risk.
You can make some tests in Pralana about what would have happened in bad bear markets by using the Historical Sequence Analysis (starting in say 1929 or 1965) on the Historical Analysis menu and test with and without conversions. To mitigate bear market risks, I make what I hope are reasonably conservative assumptions, 5% real for stocks and 2% real for bonds. When using the Historical Analysis, it shows that I'm no higher than the 20%-30% percentile (so 70-80% of historical cases would have fared better).
I never act on the automated Roth Optimizer alone, I always make manual tests, especially of various IRMAA tiers. If Pralana recommends a large conversion above your eventual tax bracket, I would want to play with alternative strategies to understand why the oddball recommendation - maybe it's avoiding things like SS tax phase-in, LTCG taxes, IRMAA, NIIT, etc. in the near future that might be reasonably predictable, but maybe it's avoiding an IRMAA tier in a decade and that would not very predictable.
Roth Conversions are not all-or-nothing. I find that the first ones are very attractive, but once you reach a certain point, there is a broad, relatively flat, optimum, where you can make lots more conversions that gain only a little. I make a judgment call, trying to weigh the chance that returns won't work out like my crystal ball vs. the chance that things might go even better than that, plus the unpleasant possibility that either I or my wife might pass early, leaving the other in a single bracket. My personal accept/reject is if the conversion doesn't increase terminal wealth by 10% of the converted value, I don't do the conversion. Since that is dependent on the years to go, maybe your IRR idea, if you properly account for the taxes on the residual IRA balance, is more fundamental.
For me the biggest unknow for whether or not to do Roth conversions is future tax rates (brackets). The rule of thumb is if you expect to be in a lower tax bracket in retirement, don't do Roth conversions. If you expect to be in the same or higher tax bracket, do conversions. Well, do you know what the tax brackets will be 10, 20, 30 years from now ????? Nobody does of course ????. My personal take is they HAVE to be higher! We can talk about that if you all would like ????. What you could try in PRC is increase future tax rates (apparently increasing by 25% is the max you can do, in PRC Gold anyway) and run your simulations and see what happens. Just a thought.
You can do a pretty good ROI on doing Roth conversions within a single scenario by first disabling Roth conversions, run a MC analysis, then do an optimize Roth conversions. Now look at the results, with the blue baseline line being no conversions and the red line being the 'optimized' in the results graph. It shows your crossover ROI point (perhaps too late in life to care?) and some savings numbers and percentages for both overall and on your taxes specifically.
Keep in mind there are other non-math reasons to do them, i.e as a form of insurance against:
1) One spouse passing away early and unexpectedly, leaving the other in the single taxpayer widow/widower penalty for the rest of their lives
2) As insurance against future egregious tax hikes (we have a lot of debt, so this is likely)
3) As a way to prepay the taxes to non-charity heirs as a gift.
4) Getting better rates on ACA premiums by lowering your MAGI using ROTH monies
5) Getting income based energy efficiency rebates by lowering your AGI using ROTH monies (if not outright canceled by present administration)
I know that Roth conversions have been discussed in several other threads, but, at the risk of redundancy 😏, I thought I would provide some Pizzaman rules of thumb for Roth conversions without needing higher math 🤗.
First of all, you should have enough money (assets) in your regular IRAs to equal your standard deduction. Standard deduction for a couple filing jointly is now $30,000/yr. Assuming a 30 year retirement (not including inflation) that would equal $900,000 in todays money. You didn't pay taxes on that money going in, and you won't pay taxes on it coming out 😘. The only time you should have more then this amount in your regular IRA is if your RMDs are less than your income needs. Run your PRC simulations 😶. If you RMDs are greater then your living expenses, you need to do more Roth conversions before your RMDs start.
Next, tax brackets. As I stated earlier, this is the most important metrics 😎. Take two people, person one puts all her money into 401k/regular IRA for 30 years. Person two puts all her money into Roth 401k/Roth IRA. If tax rates stay the same over those 30 years, who saved the most? Answer, they will have saved the exact same amount. So, if you think tax rates will go down over the next 30 years, no more Roth conversions needed.
Right now we are in near historic low tax rates. If the present administration has it's way, those rates will continue at the present great rates. How long will those great rates last?? You don't need a crystal ball to know they can't stay at those rates forever. From a political stand point the rates may only last two more years assuming the Democrats retake congress, or 4 years if a Democrat retakes the White house. But stay the same over the next 30 years????? So from a big picture stand point, Roth conversations in the next 2-4 years is a no brainer to me🤩.
Here are some links on what may influence where taxes may be going in the future:
https://usafacts.org/answers/how-much-debt-does-the-us-have/country/united-states/
https://www.investopedia.com/terms/d/debtgdpratio.asp
4) Getting better rates on ACA premiums by lowering your MAGI using ROTH monies
5) Getting income based energy efficiency rebates by lowering your AGI using ROTH monies (if not outright canceled by present administration)
How do you model this is Pralana so you ensure that it uses Roth money during your ACA years, but not necessarily after?
@jkandell I still use Gold 2025 so my explanation is based on that. During the years you want to use ACA subsidies set your Withdraw Priority to Roth. Then under Scheduled Withdraw Table for those years input the amount of money to be withdrawn from your regular IRA to get whatever amount of tax break you are looking for. For enhanced subsidies this year you need to be below 150% of the federal poverty level. It is very unlikely the enhanced subsidies will be renewed for next year.
@pizzaman I'm in a different position, where I'm getting ACA subsidies (so want a low MAGI) but don't have enough Roth at this point to fill in. So my choice each year is between 401k contributions to lower MAGI (and do Roth conversions later in life), or ignoring ACA subsidy and instead doing Roth conversions now or do tax gain harvesting. No idea how to compare those three choices in Pralana, though my hunch is that taking the ACA now to my best ability is the best option.
Pralana won't do a tax gain harvesting calculation, but you can do it manually if it's in the first year or two. If it looks like you are $X below the LTCG taxation threshold, you could be pretty close by simply reducing your starting unrealized gains by $X. That's probably a small benefit compared to ACA credits or Roth Conversions. Pralana's Roth Optimizer will examine the impact of Roth Conversions on your ACA premium credits, though you would have to manually enter values for the 401K contribution and the Optimizer will only check ordinary tax brackets. After running the optimizer, you may want to check manually by looking at different ACA FPL multiples. Note that if your income already is higher than the ACA FPL multiple you select, the program just ignores your selection, so check on the Roth Conversion Results tab to see.
I believe the Roth conversion optimizer is accounting for circumstances like ACA subsidies (premium tax credits), IRMAA Medicare premium hikes, etc when it derives its optimal set of conversions. At least implicitly, as it's looking for the best overall result and now in Online (much enhanced algos) year by year. So, in my mind (please correct if I'm wrong), overriding the result by then fiddling with the FPL, IRMAA tuning levers on the Roth optimizer page is essentially saying, "I know this will cost me some money, but those things are important to me for other reasons..."
As well, the withdrawal optimizer is factoring all things in already, such as use of Roth monies to pay for expenses and thus keep tax brackets low in order to qualify for ACA subsidies, so why try to override that math when the algo is likely far more deep and sophisticated than the lower level calcs we do on our own (i.e. not being able to suss out all the down path side effects in so many areas?
Amirite? I'm just asking because I've seen countless examples of clients who love this stuff, retired and lots of time on their hands to tinker, overriding these and thinking it's a win because they see the premium tax credits "Look, free health care!" and then we do an A/B scenario comparison and see it's not the win they thougt it was. Future them is mad at current them.
@jkandell Yes, my example assumed that the person(s) is already in retirement, not yet getting Medicare, and has a good amount in Roth's. I think it does show the advantages of having a good amount in Roth's, and to encourage people not yet in retirement or early into it, to do what they can to increase their Roth levels ????.
@pizzaman This is a very interesting discussion. I am retired, 67, receiving a survivors SS benefit, and will start my own higher benefit at age 70. Although I care about my 2 sons and their tax situations as heirs, my primary concern is to maximize my use of money. I thought I had a sweet spot for conversions paying up to the first IRMAA tier until age 70, but just recently remembered that many people say to keep a certain amount in tax-deferred in case of having tax-deductible long-term care expenses in my old age. With such a complex tool as Pralana, I'm not sure that everything is set right to evaluate for that, but it looks like even if I don't convert any more after this year, that ship has sailed. So I don't know if I have "over-converted" or should keep converting. For what its worth, I could likely live to 100 based on family, and I definitely believe tax rates will go up.
