Every time I run the "optimize withdrawal priority" it tends to put the Inherited IRA last in the order. That seems at odds with what I have learned elsewhere, which is to reduce your "forced income" early on so you have better control of your income in the future. Anyone have any ideas why Pralana is suggesting withdrawal from the Inherited IRA last? Is there a rationale for that that I may be missing?
The manual explains that it only examines the order for taxable, Roth, your tax deferred and spousal tax deferred (plus withdrawing proportionally from all), so it doesn't examine spending from the inherited IRA first in the Optimizer. Just looking at the 4 accounts it does consider over two time periods is 24 x 24 =576 + 24 (if the first period is proportional, it still has to check the other time period + 24 (if the second time period is proportional it still has to check the first + 1 (if both are proportional) = 625 combinations.
The number of combinations to check explodes factorially. If they were to consider just one more account, the number of combinations would be (5! x 5! + 5! + 5! + 1 = 14641) if I did the math right. You can run the withdrawal optimizer and then manually move the inherited IRA to the top and see how you compare with the optimizer results.
Ah, got it. Thank you kindly Richard. I have always planned to spend down the Inherited IRA early just to gain a tad more control over income. I am relieved that there isn't (necessarily) something I was missing. I took a look at the section you are referencing. They did include it in the ? contextual help for this section.
I think the help could be that it could be a tad clearer on this. For example, help states:
"OWP assumes inherited accounts, the HSA and College Savings Plan will fall after these accounts in the priority order." might be rewritten for clarity because when they say it "assumes they will fall after", it makes me think there is a rationale for it, rather than they are tacked on at the end because they are not part of the OWP algorithm. Your explanation is much clearer.
In any case, thank you!
I have a small inherited IRA (lifetime withdrawal) and have experimented with using it at various times of life using the Scheduled Withdrawal Table. In general, I have a hard time finding a reason to accelerate withdrawal since it does have tax protection, just not for as long as other IRAs.
The best chance of early withdrawals being useful was in a year when we were done with Roth Conversions, but were short of cash. Using the i-IRA avoided some capital gains taxes that were otherwise new lifetime taxes, whereas the I-IRA money was going to be taxed at some point anyway. Having high unrealized gains made it more likely to be attractive. Another driving force seemed to be if your income was otherwise low enough that reducing i-IRA RMDs could reduce taxation of SS benefits.
For us, Roth Conversions were the priority, so there was no case where using the i-IRA early was meaningfully better.
What's amazing is that for the paltry price of Pralana, you can investigate these kinds of things specific to your situation.
@ricke Thank you for that insight. I am also focusing on Roth Conversions over drawing down the Inherited IRA for now. My Inherited IRA is small relative to my portfolio, so, at the moment, I use it to pay my taxes and avoid estimated tax penalties. It's really interesting to play with these scenarios. I can't imagine how complex it is to build a tool like this, so hats off to the team that does it!
@ricke has explained this well, but the question often comes up in regard to other accounts. I.e. "Why is Pralana never using our Roth accounts for withdrawals?" Well, if your other accounts are at a 50/50 asset allocation and your Roth is 100/0, it's going to see the benefit of letting that grow as the RoR is going to be higher. Keep in mind that Roth at 100/0 is common during asset accumulation phase, but if you're going to be needing/using that money within five years, it probably shouldn't be all in equities.
Thanks Bill. I 100% agree. I was riding the "all stocks in Roth" train for a while, but it is apparent that no longer works for me for a few reasons:
1. Seeing my Roth get trounced in a downturn near retirement is stressful. I can weather the storm if I know I have some bonds/fixed as well in the Roth.
2. I may very well use the Roth to supplement my income for ACA purposes. The alternative is to hold Tbills in taxable, but then I have more ordinary income..
3. If I do tax loss harvesting with my the stocks in my taxable account during a downturn, I might want to pull some cash. To keep my AA the same, swapping bonds for stocks in the Roth during a downturn is more favorable to me than doing the same thing in tax deferred. I'd rather not grow my tax deferred obligations any further. I don't know how well this will work, but it's something I have pondered.
So, yes, I am keeping the Roth a mix. Helps me sleep at night.
Enjoyed your book by the way!
The issue you describe is the biggest problem with Pralana's competitors, where the user tries to optimize asset location for tax efficiency and the program incorrectly drives Roth Conversion answers as program tries to get rid of the slower growing tax deferred in order to get money into the faster growing Roth and taxable. What's really happening of course is it is dumping bonds to buy stocks. Pralana's Mode 1 suffers from the problem as well and that's the reason for Stuart developing Mode 2.
In Mode 2, the program keeps your asset allocation at your target over time while you hold different assets in different locations, so it is a more fair comparison for Roth Conversion calculations. The most fair comparison would be to keep the after-tax asset allocation equal over time, but that number is hard to come up with, so Mode 2 is the best available approach if you hold your bonds preferentially in tax deferred to make room for stocks in Roth and taxable
Note that Pralana does not currently track the tax costs for rebalancing in taxable when you use Mode 2, so the best way to model using Mode 2 is to keep bonds out of taxable and if they overflow from tax deferred, have them go to Roth.
You wrote: “In Mode 2, the program keeps your asset allocation at your target over time while you hold different assets in different locations, ...”
To keep Asset Allocation at your target over time you may need to manage and change your assets in different locations (Asset Location). The tool relies on the user to define the locations. But the optimum locations are not obvious.
Asset Location has up to 12 assets that need prioritization between them. I think the tool prioritize them according to the order the user insert them in the table (higher in the table – higher priority). This priority may not be the user’s intention and may not be the optimum.
I suggested, in another thread, adding an Asset Location optimizer and letting the tool have some freedom in determining Asset Location. The user should be able to configure certain Locations (such as bonds in Roth) and leave the rest to the tool to decide, usually to maximize EOL saving and minimize taxes. Rigid definition of full Asset Location may not be achievable in many cases.
I believe that for many people (including me) optimizing Asset Location will provide a better saving outcome than Roth Conversion. Also, optimizing Asset Location early in life is important since it is difficult to change Asset Location later if you have substantial unrealized capital gain in the taxable account without paying tax penalty.
I agree that optimizing asset location is a bite of the same apple of tax optimization as Roth Conversions and if started early, can significantly reduce the need for conversions. In our case, location optimization is less powerful than conversions, but helps without the pain of writing big tax checks to the government. As you say, it's important to set things up right early on to let the effects compound.
As for an optimizer for what goes where, that's really getting into fine details and runs up against even Pralana's ability to be precise in taxes. For instance, right now Pralana doesn't recognize that if you rebalance in taxable, you are likely going to realize some capital gains. That's normally a very minor thing, but if you are testing asset location and where to put overflow bonds once tax deferred is filled, it's missing a tax cost and can skew the result.
Also, it's rarely discussed, but the real parameter you are trying to hold constant to keep your risk constant when you optimize asset location is after-tax asset allocation. That's a bit of a slippery number, tax deferred has taxes due when withdrawn, but even there we only have a rough estimate for the lifetime average. So we typically ignore that and just do as Pralana does and treat all $ as equal when figuring asset allocation. But $1 of bonds in tax deferred at a 22% bracket is only $0.78 after tax, so clearly jamming all your bonds into tax deferred means foisting a larger share of your bonds onto the government, reserving stocks in other accounts for yourself.
Some of the benefit you calculate when you switch to a tax efficient allocation is due to increasing your after tax stock allocation - taking more risk. It's worth it to prove that to yourself. Test a Mode 1 portfolio with the same asset allocation all accounts in both a bull and a bear historical sequence and then switch to Mode 2 and put your bonds preferentially in tax deferred and repeat the bull and bear market tests. Mode 2 has higher highs and lower lows - more risk.
A few years ago, I tested and found that in our situation, an 80/20 Mode 2 portfolio behaved pretty close to an 84/16 equal allocation portfolio in all cases - average, bull and bear markets. I've been told by senior folks at bogleheads that the effective asset allocation shift is not the whole story, the really are benefits to optimizing asset location. I believe them and it certainly helps doing Roth Conversions, but because Pralana doesn't correct for after tax allocation, I can't really study it.
I would like Pralana at least have an option to use the "effective" tax rate in tax deferred in calculating asset allocation, so you could correct the biggest effect and certainly you would want that before trusting any results from an optimizer. But even taxable and Roth differ in risk (any losses in Roth are on you, but losses in taxable reduce dividends, possible future taxes on unrealized gains and you may even tax loss harvest). Roth and taxable are much hazier on how to equalize after-tax risk, so it's never going to be perfect.
Interesting points, thank you. I am generally very happy with the tool. Introducing Mode-2 was a big improvement for me. Until then I was considering a big Roth conversion – not anymore. I am still considering improving Asset Location in my taxable account once the market drops 40-50%.
