Understanding I-orp Logic

Non-investing personal finance issues including insurance, credit, real estate, taxes, employment and legal issues such as trusts and wills
Post Reply
User avatar
Peter Foley
Posts: 4100
Joined: Fri Nov 23, 2007 10:34 am
Location: Lake Wobegon

Understanding I-orp Logic

Post by Peter Foley » Mon Jul 18, 2016 3:10 pm

I ran the Optional Retirement Planner a few years ago and took another stab at it today with updated numbers. I found the result to be counter intuitive. I then read the FAQ's just to make sure I was understanding the reports.

The scenario proposed for my wife and me had us doing substantial IRA withdrawals and Roth conversions during the next 10 years. The result is a tax report as follows.
Year 1 -marginal tax bracket 35%
Year 2 - marginal tax bracket 33%
Years 3-6 - marginal tax bracket 28%
Years 7-26 - marginal tax bracket 25%
Years 27+ - marginal tax bracket 15%

Here are a couple statement from their FAQs

ORP reduces taxes on RMD withdrawals by lowering the IRA balance early in retirement, first by conversions, followed by withdrawals above the RMD level. After mid retirement IRA withdrawals are pegged to the RMD.


If IRA to Roth IRA conversions are allowed in these cases then conversions for a few years early in the plan that edge into a higher tax bracket are optimal. Only a relatively small portion of the withdrawal is taxed at the higher rate, not the whole withdrawal. . . . This leads to some counter intuitive withdrawal schedules but which make sense when you disassemble them.


My approach has been to do conversions just into the 25% bracket until age 70 or so, the let ourselves be pushed into a higher tax bracket when I am in my 80's (if, by chance I avoid the inevitable until that age). Personal longevity in my family makes it unlikely that RMDs on my benefits would ever push us into a bracket of 28% or higher.

Has anyone else questioned the results for the reason I present here?

User avatar
TimeRunner
Posts: 1194
Joined: Sat Dec 29, 2012 9:23 pm

Re: Understanding I-orp Logic

Post by TimeRunner » Mon Jul 18, 2016 3:19 pm

I questioned similar results a few months ago and wrote the author at the contact email address on the site. His reply:

"This is a topic that gets a lot of attention, to the point that I addressed it in a scholarly paper at http://www.iarfc.org/documents/issues/V ... Issue1.pdf, page 47.

Comfortable is the operative word here. Are you seeking comfort in protection from future tax increases or possible Social Security means testing? Or are you uncomfortable with coughing up a hunk of your retirement savings for a large tax bill in the early years of retirement? ORP is telling you that in terms of overall after-tax income it doesn't make much difference. Your comfort factor is a whole different ball game.

ORP's purpose is to provide a sandbox where prospective retirees can explore these issues."
One cannot enlighten the unconscious. | Endurance athletes are the Bogleheads of sports. | "I like people - I just don't want to be around 'em." - Russell Gordy

curmudgeon
Posts: 1326
Joined: Thu Jun 20, 2013 11:00 pm

Re: Understanding I-orp Logic

Post by curmudgeon » Mon Jul 18, 2016 4:36 pm

The first point, of course, is to make sure that you have gotten all the data in cleanly, and understand the meaning of each field correctly. If I play with it, I definitely have to sanity check the results because I occasionally misinterpret a definition (or enter monthly amounts for annual, or don't use 1000s, or...).

Once you have the baseline down correctly, then it is definitely worth running an number of variations to understand how they would impact the optimal path. In particular, you want to think about realistic life expectancies, not just "to 99 years to be safe". You want to think about scenarios such as one of you passing at age 75, while the other lives to 90. And also consider whether you hope to leave investments to family members or charity, and in what form those would be most useful.

In my mind what you want to do is to come up with a plan that is "good enough" and has reasonable outcomes across a wide range of scenarios. Trying for "perfectly optimal" may be self defeating.

User avatar
Peter Foley
Posts: 4100
Joined: Fri Nov 23, 2007 10:34 am
Location: Lake Wobegon

Re: Understanding I-orp Logic

Post by Peter Foley » Mon Jul 18, 2016 7:41 pm

The scholarly paper explaining the results starts with some assumptions (taxable, Roth, and tax deferred accounts are all available for retirement spending) and some motivations for IRA to Roth conversions. Those motivations are as follows, with only 1 through 3 considered:

Motivations for making IRA to Roth conversions include:
1. increasing retirement disposable income
2. prepositioning savings in the retiree’s estate to reduce the heirs’ personal income tax liability
3. preserving tax advantaged savings for heirs by circumventing the IRA’s RMD6
4. insuring against personal income tax rate increases during retirement


Note that 2 and 3 are specifically for the heirs benefit. Does this mean that the analysis favors an heirs tax liability more than the account owners?

I cannot imagine that it is to the account owner's benefit to have conversions in the first year push them from the 15% tax bracket to the 35% tax bracket, and into the 33% tax bracket the following year. Three may, however, be some benefit to the heirs.

Note: Federal estate taxes are not an issue in the scenarios I ran for my wife and me.

User avatar
TimeRunner
Posts: 1194
Joined: Sat Dec 29, 2012 9:23 pm

Re: Understanding I-orp Logic

Post by TimeRunner » Fri Aug 12, 2016 5:08 pm

Update: I-ORP's incorporated a new feature that permits you to select capping Roth IRA conversions at various federal tax brackets, eg 15,25,28, etc, as well as none and unlimited. See: http://www.i-orp.com . Nice improvement.
One cannot enlighten the unconscious. | Endurance athletes are the Bogleheads of sports. | "I like people - I just don't want to be around 'em." - Russell Gordy

chuckb84
Posts: 199
Joined: Wed Oct 21, 2015 10:41 am
Location: New Mexico

Re: Understanding I-orp Logic

Post by chuckb84 » Fri Aug 12, 2016 6:31 pm

i-orp is broken right now?

Scenarios I was running a month ago now all return an error

"Model M168120xPnDJ6W14DM
will not solve, possibly because your tax bracket
limit may be over constricting your taxable income."

has something changed that I don't understand?

User avatar
TimeRunner
Posts: 1194
Joined: Sat Dec 29, 2012 9:23 pm

Re: Understanding I-orp Logic

Post by TimeRunner » Fri Aug 12, 2016 6:43 pm

chuckb84 wrote:has something changed that I don't understand?
Just a guess, but check that new setting for Tax Bracket in Roth Conversions - it may be defaulting to something too low. The help file says: "ORP assists you with this issue with an IRA to Roth IRA conversion drop down list gives you three options:
* No partial conversions,
* Allow conversions to the top of the selected Federal income tax bracket.
* Allow ORP to compute the true optimal conversion level.
The drop down list includes seven Federal income tax brackets. Select the bracket that fits your situation and ORP will make partial conversions with income taxes no higher than the top of the selected tax bracket. If you select too large of a tax bracket then ORP is running in the unconstrained mode. Pick a ceiling that is too low for your income situation and ORP will cough back an infeasible model; that is the model has no solution for the specified parameters."
One cannot enlighten the unconscious. | Endurance athletes are the Bogleheads of sports. | "I like people - I just don't want to be around 'em." - Russell Gordy

afan
Posts: 3005
Joined: Sun Jul 25, 2010 4:01 pm

Re: Understanding I-orp Logic

Post by afan » Fri Aug 12, 2016 6:46 pm

Same result. You can select a tax bracket cap, but if you do, the program will not run.

The results without a cap are crazy. It has me doing massive Roth conversions, while still working, at 43% combined state and federal tax bracket, staring at age 70. It seems to completely ignore the income earned past 70. It treats 70 as the age at which one stops working, no matter what you input as earned income after "retirement" or age earned income ends. It equates "starting SS" with "retirement".

The whole plan, as the creator indicates, is to maximize spending. It functions poorly, if at all to specify a spending level, simply indexed for inflation, and save the rest. It claims to do this, but throws away a fortune in taxes on Roth conversions. I have given up on it.
We don't know how to beat the market on a risk-adjusted basis, and we don't know anyone that does know either | --Swedroe | We assume that markets are efficient, that prices are right | --Fama

User avatar
TimeRunner
Posts: 1194
Joined: Sat Dec 29, 2012 9:23 pm

Re: Understanding I-orp Logic

Post by TimeRunner » Fri Aug 12, 2016 7:26 pm

afan wrote:Same result. You can select a tax bracket cap, but if you do, the program will not run.
IDK. I suggest contacting the programmer, James Welch, at orplanner@gmail.com . As it says on the report, include your "Model ID". I guess he logs each run and can look up your run by that report number. Let us know how it goes, as I'm trying to use this too. (Haven't hit your issue.)
One cannot enlighten the unconscious. | Endurance athletes are the Bogleheads of sports. | "I like people - I just don't want to be around 'em." - Russell Gordy

User avatar
Peter Foley
Posts: 4100
Joined: Fri Nov 23, 2007 10:34 am
Location: Lake Wobegon

Re: Understanding I-orp Logic

Post by Peter Foley » Fri Aug 12, 2016 10:17 pm

Thanks to later contributors for jumping in. The results without the cap were what I was seeing.

I like the idea of being able to choose a cap although I'm still puzzled the logic that would generate massive Roth conversions and the taxes that would result.

randomguy
Posts: 5027
Joined: Wed Sep 17, 2014 9:00 am

Re: Understanding I-orp Logic

Post by randomguy » Fri Aug 12, 2016 10:32 pm

Peter Foley wrote:Thanks to later contributors for jumping in. The results without the cap were what I was seeing.

I like the idea of being able to choose a cap although I'm still puzzled the logic that would generate massive Roth conversions and the taxes that would result.



Do you have it set up so the ROTH earns a higher rate of return than the other accounts? Paying 40% taxes now is bad but if that money earns 8% instead of 4% for 40 years, that is still a good tradeoff. I did that (100% roth, 30% IRA) when I first tried it and it was amazing how aggressive it was at moving money into the roth.

2comma
Posts: 1059
Joined: Thu Jul 15, 2010 11:37 pm

Re: Understanding I-orp Logic

Post by 2comma » Sat Aug 13, 2016 3:06 am

I did the same thing as randomguy. Be very careful with your assumptions or your results will seem crazy. I'm not sure I'll ever feel comfortable with my understanding of i-orp to truly trust the results.
If I am stupid I will pay.

afan
Posts: 3005
Joined: Sun Jul 25, 2010 4:01 pm

Re: Understanding I-orp Logic

Post by afan » Sat Aug 13, 2016 8:09 am

Why would the Roth earn higher rates of return that an IRA or taxable account? You enter pre-tax returns and it applies the specified tax rates to the earning. At least, that is my understanding of what the instructions say.
We don't know how to beat the market on a risk-adjusted basis, and we don't know anyone that does know either | --Swedroe | We assume that markets are efficient, that prices are right | --Fama

curmudgeon
Posts: 1326
Joined: Thu Jun 20, 2013 11:00 pm

Re: Understanding I-orp Logic

Post by curmudgeon » Sat Aug 13, 2016 11:07 am

afan wrote:Why would the Roth earn higher rates of return that an IRA or taxable account? You enter pre-tax returns and it applies the specified tax rates to the earning. At least, that is my understanding of what the instructions say.

You might choose to hold more (or less) aggressive allocations in the Roth, perhaps because you have are planning to leave it to the next generation. This can have a major effect that is driven by the difference in expected return rather than tax efficiency. I would think that most of us should be using the same expected return for IRA and Roth to avoid this side effect.

randomguy
Posts: 5027
Joined: Wed Sep 17, 2014 9:00 am

Re: Understanding I-orp Logic

Post by randomguy » Sat Aug 13, 2016 5:10 pm

curmudgeon wrote:
afan wrote:Why would the Roth earn higher rates of return that an IRA or taxable account? You enter pre-tax returns and it applies the specified tax rates to the earning. At least, that is my understanding of what the instructions say.

You might choose to hold more (or less) aggressive allocations in the Roth, perhaps because you have are planning to leave it to the next generation. This can have a major effect that is driven by the difference in expected return rather than tax efficiency. I would think that most of us should be using the same expected return for IRA and Roth to avoid this side effect.


You can use the same allocation and returns for the ROTH, traditional and taxable to avoid the programming trying to push money into the highest return (basically asset allocation drift). But that doesn't allow you to model the common case of things like bonds in tax deferred and stocks in taxable and ROTH (with whatever adjustments you need to maintain your AA). Having higher than modeled growth in the the deferred account changes the optimal action slightly.

afan
Posts: 3005
Joined: Sun Jul 25, 2010 4:01 pm

Re: Understanding I-orp Logic

Post by afan » Fri Aug 19, 2016 11:50 am

So I emailed James Welch about what seemed to me to be anomalous results. For a guy doing this for free he is remarkably prompt and helpful.

I think his answer boils down to the program being for people with a certain set of goals, and it is not optimized for others.

In brief:
I plan to work well past 70 1/2. Thus, I will have a period when I have earned income, Social Security and RMDs. ORP has me doing my Roth conversions during this period, while my tax rate is at its highest, rather than before SS and RMDs start or after earned income ends. This struck me as backwards, and as best I can tell Welch agrees. He says the program is not aimed at people who plan to do this. It is designed for people who want to maximize spending in retirement and want to retire as soon as they can.

The program also assumes that all tax deferred funds must start RMDs at 70. For those of us whose plans permit, we can delay the start of RMDs until we actually stop working, even if it is later than 70 1/2. Thus, the program would need an option to specify which tax deferred assets (IRAs) would start RMD at 70 1/2 and which would not begin until earned income ended. He agrees, but says that no one has asked about that before, so it does not appear important to enough people to add it as a feature.

Welch says the program is intended to help people maximize lifetime spending, not maximize estates. We did not discuss it, but I assume that is why he does not update the federal estate tax exclusion amount- he says he has not seen plans with anticipated estate tax problems. (Note: it may be that the program returns results in real dollars, after inflation. If the exclusion amount increases at the same rate as the inflation rate used in the calculations, then it is appropriate to fix the exclusion amount. We did not discuss this.)

It seems ORP, although an impressive achievement, does not work if the goal is estate maximization, let alone maximization of the after tax value of an estate assuming the heirs will stretch the retirement assets.

Does anyone know of a calculator that attempts to do this? Combine maximizing the after tax value of an estate (accounting for income and estate taxes) with Roth conversion strategy?
We don't know how to beat the market on a risk-adjusted basis, and we don't know anyone that does know either | --Swedroe | We assume that markets are efficient, that prices are right | --Fama

User avatar
Peter Foley
Posts: 4100
Joined: Fri Nov 23, 2007 10:34 am
Location: Lake Wobegon

Re: Understanding I-orp Logic

Post by Peter Foley » Fri Aug 19, 2016 5:24 pm

I've been modeling some Roth conversion options using BigFoot48s' software. His Retiree Model Portfolio is a downloadable spreadsheet.

What I like about Bigfoots' version is that it has a base model (a do nothing approach) and offers a comparison of the base approach to the "Full" approach that you set up using various inputs and parameters.

Two downsides to Bigfoots' software: It takes awhile to really understand everything that is there, because there is a lot of "there" there - perhaps this is a good thing; and it models out to age 95 to provide some of the base case to full case results (I wish it would stop at age 90).

The problem with modeling too far into the future is that tax free assets have too much time to compound. As with any compound interest table comparison, this first 20 years show little difference, next 5 or so some difference, and from 25 years to 35 years a great deal of difference. This tends to favor more Roth conversions than most folks would likely see. RMD's and the resulting taxes also increase a great deal once you get to age 90.

There is information about BigFoot48's software on the Wiki under Roth conversions: https://www.bogleheads.org/wiki/Retiree_Portfolio_Model

Here is a brief description from the Wiki: Forum member BigFoot48 has created a spreadsheet for use by retirees, or those nearing retirement, which will estimate the financial impact on your portfolio, including income taxes and RMDs, from doing Roth conversions. Use this spreadsheet to determine if Roth IRA conversions may be worthwhile for your personal situation.

Yearly results are calculated and provided (such as income, expenses, taxes, inheritances, and asset sales over a selectable 1 to 40 year period) for both doing conversions, and not doing conversions, so an easy comparison can be made.

User avatar
TimeRunner
Posts: 1194
Joined: Sat Dec 29, 2012 9:23 pm

Re: Understanding I-orp Logic

Post by TimeRunner » Fri Aug 19, 2016 6:38 pm

The big difference between Bigfoot48's RPM and Welch's ORP is that the former is a modeler and the latter an optimizer. In RPM you have to decide how much to pull from after-tax (money subject to Cap Gains, etc), tax-deferred, or Roth IRA accounts, and how much you want to spend. IORP will optimize across the account classes for each year, trying to maximize your total spending in retirement. (If you set inflation to 0 it's easier to see in IORP.) ESPlanner seems closer to RPM because you tell it the order to spend assets, typically depleting employee tax-deferred retirement accounts first, employer tax-deferred retirement accounts second, and Roth IRA assets last because Roth IRA assets aren’t subject to minimum distribution requirements. It does that regardless of anticipated earnings for each class of accounts. (You can change the order if you like, but still it depletes the first before moving to the second, etc unless you do special withdrawals, in which case the program's smoothing code is bypassed.) ESPlanner will also spend non-retirement assets (checking, savings, etc) before dipping into retirement funds. Ditto ORP I believe.

One way perhaps to approach this is to start with ORP and try to understand the pattern of the optimizing, and then use that in modeling with RPM. I think you could also use it to model in ESPlanner using special withdrawals, but you would really be fighting what that one is trying to do.

If I didn't have a full-time job and many other obligations, it would be fun to play with this, but I will have to retire before I really have time. :D

P.S. I could be wrong in my understanding too! (more likely than not on finer points)
One cannot enlighten the unconscious. | Endurance athletes are the Bogleheads of sports. | "I like people - I just don't want to be around 'em." - Russell Gordy

wanderer
Posts: 60
Joined: Sat Aug 16, 2014 4:09 pm
Location: Houston, Texas, USA

Re: Understanding I-orp Logic

Post by wanderer » Fri Aug 19, 2016 7:45 pm

Based on this thread I went back and ran the ORP model using my assets and a fixed initial spend rate. The 15% bracket failed, The 25% and 28% brackets provided results. The "No Partial Conversions" and "Unlimited Conversions" gave the same results each time! Looking at the last pages of the output where the input data is reflected, I note that the index for ira2Roth is the same for the latter 2 cases, but appears to correctly change for the individual brackets. This was consistent after resetting the model, changing other parameters, etc. I've sent a note to Mr. Welsh and will post the feedback.

Others may want to review that their input data is properly reflected in the model's output page.

User avatar
Peter Foley
Posts: 4100
Joined: Fri Nov 23, 2007 10:34 am
Location: Lake Wobegon

Re: Understanding I-orp Logic

Post by Peter Foley » Fri Aug 19, 2016 9:06 pm

A couple very astute observations.

afan wrote:

Welch says the program is intended to help people maximize lifetime spending, not maximize estates. . . . It seems ORP, although an impressive achievement, does not work if the goal is estate maximization, let alone maximization of the after tax value of an estate assuming the heirs will stretch the retirement assets.


TimeRunner wrote:

The big difference between Bigfoot48's RPM and Welch's ORP is that the former is a modeler and the latter an optimizer. In RPM you have to decide how much to pull from after-tax (money subject to Cap Gains, etc), tax-deferred, or Roth IRA accounts, and how much you want to spend. IORP will optimize across the account classes for each year, trying to maximize your total spending in retirement.


So decide what your goal is and choose the right tool.

Thanks to you both for your contributions. This helps a lot.

wanderer
Posts: 60
Joined: Sat Aug 16, 2014 4:09 pm
Location: Houston, Texas, USA

Re: Understanding I-orp Logic

Post by wanderer » Sat Aug 20, 2016 7:23 am

Mr. Welch was very responsive to my questions.

Here are his responses:
Today at 6:47 AM
Ah! You are setting the spend parameter, which caps spending and maximizes the estate. ORP won't turn off conversions in the max estate mode, because that mode constricts cash flow to the point of creating infeasible models which are not in fact actually infeasible. No conversions is also not allowed when ACA is being modeled.

James Welch
ORP

On Aug 19, 2016, at 8:52 PM,
Correct, the ORP parameter page does not reset to "no partial conversions". This is a known issue that I have not resolved yet. However, you can just select (mouse click on) no partial conversions from the drop down list and you should be good to go. This is the zeroth value, so to speak.

On Fri, Aug 19, 2016 at 8:29 PM <wanderer> wrote:

The model does not appear to reset to "no partial conversions". I am running through the Roth conversion options and noted that the "no partial conversion" case gave the same results as the "unlimited conversion" case. Looking at the last page of results for the input summary I see the same index (7) for the "ira2roth" parameter for both cases. The cases for specified withholding limits, 25%, 28%, etc. all appear to index properly. (2, 3, etc.). For my data, the 15% withholding case fails as expected.

Please advise how I can run a comparison case for no ira2Roth conversions.

Thanks!

afan
Posts: 3005
Joined: Sun Jul 25, 2010 4:01 pm

Re: Understanding I-orp Logic

Post by afan » Sat Aug 20, 2016 8:51 am

It seems estate maximization is not what ORP is about. For example, it really wants to spend all the money that is not committed to taxes on RMDs and Roth conversions. It is not clear what happens to money you earn in excess of your capped spending.

It does not look at estate tax effects at all and it does not compare the value of leaving tax deferred assets vs Roth IRAs after accounting for differences in income tax brackets between the current owner and the heirs.

In short, if you want to retire as soon as you can and then you want to maximize your spending throughout retirement ORP might give you a good guide on how to go about it. If you want to act like many bogleheads- set a reasonable spending amount and save the rest- this looks like the wrong tool for that job.
We don't know how to beat the market on a risk-adjusted basis, and we don't know anyone that does know either | --Swedroe | We assume that markets are efficient, that prices are right | --Fama

gneeby
Posts: 41
Joined: Wed Jul 02, 2008 1:49 pm

Re: Understanding I-orp Logic

Post by gneeby » Fri Aug 26, 2016 9:10 am

afan wrote:It seems estate maximization is not what ORP is about. For example, it really wants to spend all the money that is not committed to taxes on RMDs and Roth conversions. It is not clear what happens to money you earn in excess of your capped spending.

It does not look at estate tax effects at all and it does not compare the value of leaving tax deferred assets vs Roth IRAs after accounting for differences in income tax brackets between the current owner and the heirs.

In short, if you want to retire as soon as you can and then you want to maximize your spending throughout retirement ORP might give you a good guide on how to go about it. If you want to act like many bogleheads- set a reasonable spending amount and save the rest- this looks like the wrong tool for that job.


ORP operates in one of two modes, as this thread makes clear, 1)Maximize spending, or 2)Maximize estate.

The max estate mode is suitable for the retiree, or soon to retire, who has a good grip on what their retirement expenses look like. In this mode ORP enables conversions to relax the model and avoid infeasible solutions. In the max estate mode the RMD drives the tax-deferred balance down to a low level and the money ends up in the Roth or in some situations also in the After-tax Account. There will be a small balance in the tax-deferred account which the RMD doesn't force out of the Tax-deferred Account and ORP keeps it there to avoid paying taxes on the distributions, leaving that to the heirs.

ORP includes estate taxes in maximizing its estate. Frankly, retirees with estates in the $5M range need the assistance of a trust planner and need not be all that concerned with IRA withdrawal tax efficiency.

Maximum spending is definitely the concern of 50 year old potential early retirees who are doing feasibility studies of when to make the big jump and what will be spending level that they can budget against during their extended retirement beginning when? that's part of the study.

ORP does not model the situation when the user maximizes spending and then chooses not to spend it all. Presumable the surplus remains in savings and will be reflected in the initial balances when the model is run again next year. The annual rerun is how most serial ORP users operate.

Post Reply