How do you model Roth conversions without planning software?

[silent cal]

Hi all,

I’ve recently done a deep dive into financial planning software like New Retirement versus tools like VPW.

My (admittedly amateur) understanding is that the main critique of Monte Carlo-based tools versus amortization-based methods like VPW is that Monte Carlo relies on far more assumptions, which affect the range of probabilistic outcomes—essentially, “garbage in, garbage out.” Tools like VPW are simpler to understand and depend on fewer assumptions.

For the most part, I see how VPW can assist with retirement timing, withdrawal strategies, Social Security/pension claiming strategies, etc. However, one area I don’t quite understand how to model without planning software is Roth conversions.

For those of you who don’t use planning software, how do you model these? Do you have an Excel document?

[bombcar]

Excel is the king, queen, prime minister, and all of parliament.

Nothing else even comes close.

However; it all comes down to assumptions about future growths (you can negate this assumption via equalization and not trying to tax advantage WHERE investments are but instead have the same mix in every type of account) and “bird in hand” questions about paying tax now vs paying tax in the future.

Also remember that you’re making assumptions even if you make no assumptions - the USA could switch to a VAT style tax tomorrow and suddenly “tax free Roth withdrawals” would be taxed on spending (but that would affect other withdrawals also).

With Excel you can focus on the things you care about and you can see what you did each year as you move forward and adjust assumptions and activities based on it.

[tibbitts]

I completely disagree with there being any distinction between "planning software" and an Excel (well, it's Bogleheads, so maybe Google Sheets to save money) spreadsheet, regardless of who develops the spreadsheet. It's all just software and effectively a spreadsheet is a programming language. Didn't some of the commercial software start out as just that: a spreadsheet someone developed for their own financial planning? I would say that only a small minority of (yes, even) Bogleheads are capable/determined enough to develop software (using a spreadsheet or otherwise) that's comparable to - and particularly as well tested as - most of the commercial software, but I'm sure some (many?) believe theirs is better in various respects.

[1moreyr]

I used excel in my accumulation phase and now in retirement. I guess you are on a DIY page so i am not surprised you get excel answers.. if you are not interested in all the puts and takes and constantly learning about investments, you may be better off with a software that helps.

but frankly, if you are doing basic stuff. There is the current annual tax bracket, your income, less the standard deduction (and 401K etc) vs the percentage rate in your tax bracket that you owe vs what you paid in.

That's all i do to figure out where i am and I can model how much i can roll over to roth and what it's going to cost me.
I do have a tax accountant and I have been pretty close to his number without doing much more than this. (my reasons for even using a tax accountant are not because I actually need one, but it's a firm I support for personal reasons).

[Tom_T]

It is difficult to model Roth conversions in a spreadsheet without taking all other factors into account. Other sources of income including SS, their taxability, future expenses (which determine income needs), RMDs, and so on. It can be very complex because there are a number of variables that determine your AGI, and therefore the amount of room you have in your tax bracket for Roth conversions. Any Excel spreadsheet that does this would certainly be quite detailed - but I don't put it past Bogleheads to do it!

Me, I use Pralana, even though I'm good with Excel. There are just too many factors for me to track. Plus, Pralana (and other software) allows me to create multiple scenarios, so I can compare different conversion strategies side-by-side and see the results.

For the less adventurous, Excel might be a good DIY option if evaluating a Roth conversion for a single year, as you probably have a good idea of what your finances look like for the next year.

[Jack FFR1846]

To get the best results, don't be set on doing everything using a single tool. To determine my "retirement spending", I use a simple list. For "life planning" into the future, I use Excel, taking that retirement spending number and putting it into the spending block. To determine income, I use another simple list and it goes into the life planning income block. For Roth conversions, I use Excel mainly to see what my tIRA balance will become. It includes social security payments for me and for DW, total interest, total dividends, RMDs as they start and the annual Roth conversions. The reason for this is to see if we can avoid IRMAA once our total income is "through the roof" from SS and RMDs. I've looked at many of the software tools available and find many of them too complicated to understand. Building my own simple tools has worked well for me. I've utilized a single Excel sheet to do all of this. Pretty simple to make a "retirement spending" list and then link the total into my sheet for other things.

[rkhusky]

I don’t model Roth conversions, I just convert to the top of the 12% bracket and hope that’s good enough. I did do some calculations to show that this is reasonable, given our balances, spending, and years to SS. But it might not be “optimal”.

However, I did download this spreadsheet model, but never got around to using it:
viewtopic.php?t=97352

[tibbitts]

It didn't occur to me to ask earlier, but what's the reason(s) for not wanting to use planning software?

[mhc]

I'm almost 20 years away from RMDs. I assume a lot will change over the next 20 years. I do not believe one can accurately plan what things will be 10+ years out. At some point, it becomes a guessing game. I do believe in making long term plans, but I do not think it is worth my time to put in a lot of effort in things that will occur decades out. I spend most of my planning on the near future.

For the current year, I do a dummy TurboTax return to see how much to convert. My taxes are complex because there are a lot of tax credits that come into play.

We have all our investments at Fidelity. I run the Fidelity retirement tool. It can show the amounts of RMDs by year under different market conditions. I look at the numbers to see how our income will look relative to our current year. I am trying to keep our income/taxes flat over the years. I think this approach is close enough.

[avalpert1]

A model built in Excel is planning software - so the way the question and too many of the response are framed here is a red herring.

The challenge is whether the model you bought from someone else or built yourself is providing additional value for Roth conversion decisions beyond some naive simple heuristics - and I find that questionable. The issue with most of the models I've seen on this is they leave an impression of false precision that paper over the large uncertainty bars in the many assumptions that go into it (its bad enough when you are only dealing with errors in returns and inflation but when you add in assumptions about tax rates, RMD tables, Medicare costs, etc. it really is out of hand).

[FiveK]

For those of you who don’t use planning software, how do you model these? Do you have an Excel document?

At the very least you should be doing some compound growth calculations so you can estimate RMDs, so unless you do those with your trusty slide rule or your favorite HP or TI calculator, a spreadsheet is probably the minimum to get in the game.

But that minimum may in fact be sufficient for a very good estimate. See Estimating future marginal tax rate.

Even the most sophisticated software can't guarantee accurate predictions of future tax law, investment returns, and your personal situation. Nothing wrong with using such tools if that is your preference, but some decent back of the envelope estimates may be just as useful. Of course, that can lead to a discussion about the definition of "decent"....

Just noticed that this is very similar to what avalpert1 said in the previous post.

[Exchme]

What's so bad about using planning software? Roth Conversions depend on the details of tax laws, life expectancy, returns, other income, spending, etc. Aside from the inherent uncertainty that poses, it means you need a very detailed model to even understand the tradeoffs. I would not want to undertake programming enough of the tax code to capture the relevant issues when others have done a good job for you. If you are allergic to spending money, the Retiree Portfolio Model at the wiki is good and free.

An even more powerful tool is Pralana Gold (paid spreadsheet, the online version is in limited release, hopefully full rollout soon). It has all kinds of income and expense options, withdrawal strategies, historical and Monte Carlo options in addition to static projections, various optimizers and more.

Use a bit of money so you don't have to work as hard as the programmers.

[N.Y.Cab]

I use Fidelity retirement planner to see the range of income in normal market condition in today’s dollar. My goal is to flatten out the income slope caused by RMDs. There are constraints like NIIT and IRMAA cliffs that I don’t want to cross over so no need to over analyze this.

[Parkinglotracer]

I have tried some modeling software but ended up with two excel spreadsheets.

One that models our federal and state taxes for annual tax planning

One whose goal it is to see how much our portfolio (including taxable IRAs) will grow and hence how much our RMDs will be in 12 years when we turn 75.

After seeing the results for what our RMDs will be and our estimate of how much our taxable income / taxes will be with RMDs, or taxes will be if one of us were to die early, then I used the “That Looks About Right” TLAR method. We decided to convert to the top of the 22% tax bracket and stay below the first IRMMA shelf. I think paying top end 22% tax bracket is a fair deal today.

I will continue to sample planning software but don’t think it is really worthwhile to try to nail jello to the wall with an exact conversion number. My wild guess is market returns will average 5-6% annually in next 15 years and tax brackets for married filing jointly in the 200K AGI range will be about 25% federal and 5% state.

[tonyclifton]

You can use tax prep software figure out the taxes paid on the conversion.

[steadyosmosis]

I do "what if" tax returns periodically throughout the year, to determine my optimal Roth-conversion amounts.

[BigFoot48]

For the most part, I see how VPW can assist with retirement timing, withdrawal strategies, Social Security/pension claiming strategies, etc. However, one area I don’t quite understand how to model without planning software is Roth conversions.

For those of you who don’t use planning software, how do you model these? Do you have an Excel document?

You might want to try the Bogleheads hosted Excel Retiree Portfolio Model which was designed to analyze the benefit of doing Roth conversions and over the years has evolved to modeling many other aspects of retirement financial planning: viewtopic.php?t=97352

[Peter Foley]

BigFoot48 is being a bit modest in his post. The RPM spreadsheet he designed is an excellent way to model Roth conversions. It is a downloadable, complex, excel spreadsheet.

The Minnesota Bogleheads group did a review of a number of retirement calculators and spreadsheets a few years ago and the RPM was very highly rated. Firecalc was the other top rated planning tool but it purpose is more for planning for retirement, rather than Roth conversions and withdrawal scenarios.

If you do not want to use software, the Numbers Unlimited annual publication is an excellent reference.

[dcabler]

I looked at a bunch of different software packages, including those that used Excel and those that didn't. They all either came with assumptions that didn't map to my situation (or which I disagree with) or had such a steep learning curve that I might as well just do things myself. As others have noted, Excel is how this gets done.

In my case several spreadsheets are involved.
- A spreadsheet that uses optimization and constraints to work through a plan for conversions and their potential to improve after-tax spending
- A withdrawal spreadsheet I have that uses amortization and the conversion plan for my withdrawals
- A spreadsheet where I enter all of my quarterly information as input to Personal Finance toolbox in order to calculate how much quarterly tax I owe using the annualized income method.
- Personal Finance Toolbox, available from a link on the Bogleheads wiki, which is an excel spreadsheet that I use to help calculate quarterly tax obligations

In my case, what I have found is that overall after-tax spending improvements for doing Roth conversions are modest, at best, and that a large range of conversion strategies end up with nearly the same level of improvements. YMMV. There are of course other reasons to do conversions.

Cheers.

[FiveK]

...a large range of conversion strategies end up with nearly the same level of improvements.

+1 (at least, that's what we see also)

[dcabler]

...a large range of conversion strategies end up with nearly the same level of improvements.

+1 (at least, that's what we see also)

I suspect it's pretty common. Once you see that, then you can start to think about other reasons to choose one strategy over another. One I'm leaning towards right now is one that results in smaller conversions over a slightly longer time period, giving me the option of "pulling the plug" on the whole thing if there's a good reason to...

Cheers.

[Parkinglotracer]

viewtopic.php?p=8025873#p8025873

Sure makes one think twice about how worthwhile Roth conversions are …

[diy60]

...a large range of conversion strategies end up with nearly the same level of improvements.

+1 (at least, that's what we see also)

I suspect it's pretty common. Once you see that, then you can start to think about other reasons to choose one strategy over another. One I'm leaning towards right now is one that results in smaller conversions over a slightly longer time period, giving me the option of "pulling the plug" on the whole thing if there's a good reason to...

Cheers.

I spread my Roth conversions over 8 years. This allowed me to adjust along the way. One year I reduced the conversion amount to maximize the Covid stimulus amount. My final year I pulled the plug to account for a couple of international vacations.

Also, I built large spreadsheet with 28 tabs and OpenSolver plugin to run scenarios. Early in the process I abandoned that spreadsheet for a paper and pencil approach.

[jeffyscott]

A model built in Excel is planning software - so the way the question and too many of the response are framed here is a red herring.

So if DIY spreadsheets are out, I'd start by looking at the new paper from McQ:
viewtopic.php?t=438720
(that's the same discussion that parkinglottracer linked to above)

And then look at his "Easy No-Bake Recipe":

[This isn’t quite the bounded solution Dave requested, but it is in that spirit]

McQ’s Easy No-Bake Recipe for Roth Conversions

1. If the first digit of the tax rate on the conversion will be 0 (0.0x%), DO convert.

2. If the tax rate applicable to the conversion begins with ‘1’ AND total AGI in the first year of RMDs will exceed (in real dollars) $125,000 (MFJ), then DO convert; ELSE, do not.

3. If the tax rate applicable to the conversion begins with ‘2’, don’t bother converting UNLESS you anticipate SS and pension income north of $225,000 (real, MFJ) AND a TDA balance when RMDs begin north of $6 million in 2023 dollars.

4. If the first digit in the conversion tax rate will be ‘3’, please retain bsteiner or another advisor accustomed to working with High Net Worth individuals—there may be estate planning considerations that shape your decision to convert.

The above assumes that you are a “Kahneman and Tversky" type decision maker: that you would never voluntarily spend / give up a certain $1.00 today to save an uncertain $1.20 tomorrow, or an uncertain $2.00 ten years hence.

But you'd definitely wager $1.00 under good odds now with a prospect of earning $10.00 soon.

viewtopic.php?p=7382677#p7382677

Note that where it says "tax rate" I would interpret that to mean whatever the true effective cost of the conversion is, including impacts on ACA cost, tax credits, etc.

[billthecat]

Some time ago I became acutely aware of true marginal tax rate as opposed to the simple income marginal tax rate. In other words, because my qualified dividends straddle the line between taxed and not taxed, any ordinary income I add (such as a Roth conversion) would not only be subject to ordinary income tax but also would push additional qualified dividends into being taxed. Thus, the true marginal tax rate on that ordinary income is the combination of the two rates. NewRetirement does account for that, apparently, but overall I didn't like the way it handled modeling Roth conversions.

So I made my own DIY spreadsheet (in Numbers, FYI), with a focus on producing a chart where I could eyeball the true (combined) marginal rate. It's terribly complicated but I'm trying to weave a path around new rates starting in 2026 (based on current legislation), as well as NIIT and IRMAA, by factoring not only Roth conversions, but also Social Security, I bond interest, RMDs, and qualified and non-qualified dividends and interest.

The resulting numbers (e.g., how much to convert) only are there to build the strategy - the actual numbers depend on so many things that will develop in the future, such as tax rates, investment returns, inflation, and IRMAA thresholds. Most of my pre-tax holdings (97%) are in total bond, so returns aren't a huge factor. And because IRMAA is a retroactive tax, there's no way to get it exactly right, so I have to leave margin or I risk stepping over the line.

[ThankYouJack]

And then look at his "Easy No-Bake Recipe":

[This isn’t quite the bounded solution Dave requested, but it is in that spirit]

McQ’s Easy No-Bake Recipe for Roth Conversions

...
3. If the tax rate applicable to the conversion begins with ‘2’, don’t bother converting UNLESS you anticipate SS and pension income north of $225,000 (real, MFJ) AND a TDA balance when RMDs begin north of $6 million in 2023 dollars.
...

Does anyone know how McQ came up with this rule of thumb? Seems off to me

[FiveK]

And then look at his "Easy No-Bake Recipe":

[This isn’t quite the bounded solution Dave requested, but it is in that spirit]

McQ’s Easy No-Bake Recipe for Roth Conversions

...
3. If the tax rate applicable to the conversion begins with ‘2’, don’t bother converting UNLESS you anticipate SS and pension income north of $225,000 (real, MFJ) AND a TDA balance when RMDs begin north of $6 million in 2023 dollars.
...

Does anyone know how McQ came up with this rule of thumb? Seems off to me

A good rule of thumb for traditional vs. Roth questions is to ignore most rules of thumb that attempt to simplify the issue.

[Tom_T]

And then look at his "Easy No-Bake Recipe":

[This isn’t quite the bounded solution Dave requested, but it is in that spirit]

McQ’s Easy No-Bake Recipe for Roth Conversions

...
3. If the tax rate applicable to the conversion begins with ‘2’, don’t bother converting UNLESS you anticipate SS and pension income north of $225,000 (real, MFJ) AND a TDA balance when RMDs begin north of $6 million in 2023 dollars.
...

Does anyone know how McQ came up with this rule of thumb? Seems off to me

A good rule of thumb for traditional vs. Roth questions is to ignore most rules of thumb that attempt to simplify the issue.

Agreed. Haven't we learned by now that one's individual circumstances carry far more weight than rules of thumb? There are enough modeling tools out there to let us know if a conversion is worth considering.

[avalpert1]

And then look at his "Easy No-Bake Recipe":

[This isn’t quite the bounded solution Dave requested, but it is in that spirit]

McQ’s Easy No-Bake Recipe for Roth Conversions

...
3. If the tax rate applicable to the conversion begins with ‘2’, don’t bother converting UNLESS you anticipate SS and pension income north of $225,000 (real, MFJ) AND a TDA balance when RMDs begin north of $6 million in 2023 dollars.
...

Does anyone know how McQ came up with this rule of thumb? Seems off to me

A good rule of thumb for traditional vs. Roth questions is to ignore most rules of thumb that attempt to simplify the issue.

I actually disagree with that - this is one of those few decisions where I think rules of thumbs can be better than the false precision of modeling decades of future returns, inflations, tax rates, and numerous other tax policy decisions.

You know your current marginal rate with certainty and can narrow down to a small number of likely future situations that may serve as exceptions (large underlying retirement income being the most common). That beats pretending to know what your RMD will be in 30 years.

[jeffyscott]

And then look at his "Easy No-Bake Recipe":

[This isn’t quite the bounded solution Dave requested, but it is in that spirit]

McQ’s Easy No-Bake Recipe for Roth Conversions

...
3. If the tax rate applicable to the conversion begins with ‘2’, don’t bother converting UNLESS you anticipate SS and pension income north of $225,000 (real, MFJ) AND a TDA balance when RMDs begin north of $6 million in 2023 dollars.
...

Does anyone know how McQ came up with this rule of thumb? Seems off to me

I'm guessing the idea was don't convert at 22 or 24%, unless you project that you would otherwise be in the 32% bracket. But the numbers don't seem to match that.

In any case, the new paper is mostly about the risks and benefits conversions at those rates that begin with a 2.

[diy60]

And then look at his "Easy No-Bake Recipe":

[This isn’t quite the bounded solution Dave requested, but it is in that spirit]

McQ’s Easy No-Bake Recipe for Roth Conversions

...
3. If the tax rate applicable to the conversion begins with ‘2’, don’t bother converting UNLESS you anticipate SS and pension income north of $225,000 (real, MFJ) AND a TDA balance when RMDs begin north of $6 million in 2023 dollars.
...

Does anyone know how McQ came up with this rule of thumb? Seems off to me

A good rule of thumb for traditional vs. Roth questions is to ignore most rules of thumb that attempt to simplify the issue.

I actually disagree with that - this is one of those few decisions where I think rules of thumbs can be better than the false precision of modeling decades of future returns, inflations, tax rates, and numerous other tax policy decisions.

You know your current marginal rate with certainty and can narrow down to a small number of likely future situations that may serve as exceptions (large underlying retirement income being the most common). That beats pretending to know what your RMD will be in 30 years.

I sorta agree with this, but instead of rules of thumb just do a quick 5 minute calculation with paper and pencil. Large ticket costs in the numerator (income tax, IRMAA, loss of credits, loss of subsidies) divided by the denominator (delta income for said costs). Do this for no Roth conversion and with Roth conversion. Compare same to the future period. Rinse and repeat each year. Nearly everyone will fall into the "it doesn't matter group".

[FiveK]

A good rule of thumb for traditional vs. Roth questions is to ignore most rules of thumb that attempt to simplify the issue.

I actually disagree with that - this is one of those few decisions where I think rules of thumbs can be better than the false precision of modeling decades of future returns, inflations, tax rates, and numerous other tax policy decisions.

You know your current marginal rate with certainty and can narrow down to a small number of likely future situations that may serve as exceptions (large underlying retirement income being the most common). That beats pretending to know what your RMD will be in 30 years.

We may be in more agreement than disagreement, depending on how we define "rule of thumb". E.g., most such rules don't acknowledge marginal rates and use "bracket" or even "income" instead. Thus we agree on starting with "current marginal rate".

Given a current marginal rate, one still needs a estimated future marginal rate for comparison. And yes, precision is not accuracy, so I think we agree that excruciating detail isn't a prerequisite for such estimates.

[FiveK]

...depending on how we define "rule of thumb".

...instead of rules of thumb just do a quick 5 minute calculation with paper and pencil.

I might use a calculator or Excel , but yes this is what I was getting at in the reply to avalpert1 - thanks!

[heyyou]

Being spreadsheet illiterate, we just converted to the top of our tax bracket with annual advice from our tax accountant, for my decade of early retirement starting in 2005 when I was 55.

With the growth in our all stock RIRAs, now I wish that we had steadily converted a little more each year, paying the higher tax on a portion of each conversion. For a reference point, my highest job income was only $65K before interest and dividends, and we were living on my almost $41K pension in early retirement, so YMMV. Soon, there is some dreaded reversion to previous tax rates.

We were saving so much in pre-retirement that we were inadvertently practicing at living on my future pension income, in our late work years. I had been saving my pay raises for a decade prior to retiring, using pay-roll deductions, so we never saw the money in our checking account.

[VanGar+Goyle]

Haven't we learned by now that one's individual circumstances carry far more weight than rules of thumb? There are enough modeling tools out there to let us know if a conversion is worth considering.

At the risk of comparing investors to lemmings, all you need is one perfect ( or close enough ) modeling tool.
Having many wrong tools, or many inconsistent tools is not as helpful.
I prefer to model one year at a time. Tools that model out 30 years with precision to the cent are laughable, and probably inaccurate.

[ThankYouJack]

...depending on how we define "rule of thumb".

...instead of rules of thumb just do a quick 5 minute calculation with paper and pencil.

I might use a calculator or Excel , but yes this is what I was getting at in the reply to avalpert1 - thanks!

+2.

Although I wouldn't even use a pencil / paper or Excel Instead I take a minute to use something like FireCalc to estimate the range my tax-deferred accounts will grow until retirement and realize I (or my heirs) won't be avoiding Uncle Sam's cut.

[tibbitts]

Haven't we learned by now that one's individual circumstances carry far more weight than rules of thumb? There are enough modeling tools out there to let us know if a conversion is worth considering.

At the risk of comparing investors to lemmings, all you need is one perfect ( or close enough ) modeling tool.
Having many wrong tools, or many inconsistent tools is not as helpful.
I prefer to model one year at a time. Tools that model out 30 years with precision to the cent are laughable, and probably inaccurate.

Although I understand the point about the relative futility of "precision" long-term planning, you have to incorporate considerations for years/decades into the future because that might be where the benefits show up. So I'm not seeing where "modeling one year at a time" is sufficient. With no potential benefit in the maybe-far-off future there would be no way to justify any kind of sacrifice now. Of course there are exceptions including obvious, essentially "free" conversions where for example someone can convert and stay in essentially the zero bracket for ... everything: in those cases one year might be all you need.

[Tom_T]

Haven't we learned by now that one's individual circumstances carry far more weight than rules of thumb? There are enough modeling tools out there to let us know if a conversion is worth considering.

At the risk of comparing investors to lemmings, all you need is one perfect ( or close enough ) modeling tool.
Having many wrong tools, or many inconsistent tools is not as helpful.
I prefer to model one year at a time. Tools that model out 30 years with precision to the cent are laughable, and probably inaccurate.

How do you model one year at a time? The benefit doesn't come until years later.

[jeffyscott]

I don't know if this would count as modeling one year at a time, but my initial analysis was to take the current balance in the TIRA accounts and determine the (at the time) age 72 RMD based on that balance. To that I added SS (it is virtually certain that 85% of SS will be taxed for us, under current law) and pension income and determined that meant the RMDs would be in the 22% tax bracket. Therefore converting year at 12% Federal seemed like a good bet.

That analysis could be repeated each year. While effectively assuming a real growth rate of 0% real may be low, the accounts were 70% bonds and stocks would be converted first. Also with a higher growth rate, we might pay more tax but would still have more after tax money than expected. An additional factor was that while pension could and probably increase enough to keep up with inflation, there is no guaranteed COLA.

(I did more recently look at whether or not a 2% real growth rate and higher RMDs (using the age 80 percentage) made any difference and it didn't, we'd would still end up in the low end of the 22% bracket only want to convert at 12%.)

Now that we will be transitioning to the ACA, the 12% bracket will become effectively a 27% tax rate. Thus further conversions seem like an extremely poor bet and so they will come to an end.

[Tom_T]

I don't know if this would count as modeling one year at a time, but my initial analysis was to take the current balance in the TIRA accounts and determine the (at the time) age 72 RMD based on that balance. To that I added SS (it is virtually certain that 85% of SS will be taxed for us, under current law) and pension income and determined that meant the RMDs would be in the 22% tax bracket. Therefore converting year at 12% Federal seemed like a good bet.

That analysis could be repeated each year. While effectively assuming a real growth rate of 0% real may be low, the accounts were 70% bonds and stocks would be converted first. Also with a higher growth rate, we might pay more tax but would still have more after tax money than expected. An additional factor was that while pension could and probably increase enough to keep up with inflation, there is no guaranteed COLA.

(I did more recently look at whether or not a 2% real growth rate and higher RMDs (using the age 80 percentage) made any difference and it didn't, we'd would still end up in the low end of the 22% bracket only want to convert at 12%.)

Now that we will be transitioning to the ACA, the 12% bracket will become effectively a 27% tax rate. Thus further conversions seem like an extremely poor bet and so they will come to an end.

When you estimate the age 72 RMD, if you've determined that a conversion makes sense, do you convert an amount equal to the projected RMD? That gives you an apples-to-apples comparison of the tax savings for that amount, no? And then, the following year, do the same exercise for the age 73 RMD, and so on?

[jeffyscott]

I don't know if this would count as modeling one year at a time, but my initial analysis was to take the current balance in the TIRA accounts and determine the (at the time) age 72 RMD based on that balance. To that I added SS (it is virtually certain that 85% of SS will be taxed for us, under current law) and pension income and determined that meant the RMDs would be in the 22% tax bracket. Therefore converting year at 12% Federal seemed like a good bet.

That analysis could be repeated each year. While effectively assuming a real growth rate of 0% real may be low, the accounts were 70% bonds and stocks would be converted first. Also with a higher growth rate, we might pay more tax but would still have more after tax money than expected. An additional factor was that while pension could and probably increase enough to keep up with inflation, there is no guaranteed COLA.

(I did more recently look at whether or not a 2% real growth rate and higher RMDs (using the age 80 percentage) made any difference and it didn't, we'd would still end up in the low end of the 22% bracket only want to convert at 12%.)

Now that we will be transitioning to the ACA, the 12% bracket will become effectively a 27% tax rate. Thus further conversions seem like an extremely poor bet and so they will come to an end.

When you estimate the age 72 RMD, if you've determined that a conversion makes sense, do you convert an amount equal to the projected RMD?

No, assuming the RMD is in the 22% bracket, I convert as much as I can at a cost of 12% Federal (+6% state in my case). Pensions + 85% of SS are enough to ensure that we will remain in at least the 12% bracket (under current law). And even if I have overestimated, the penalty will likely only be 2% (10% bracket instead of 12%).

(I may be misremembering the RMD rules, was the first one required at 70 and then it changed to 72 before changing again? If so then I would have used age 70 initially, then 72, then 75. Always just looking at the first year RMD.)

And then, the following year, do the same exercise for the age 73 RMD, and so on?

No, suppose in 2017 I have $1 million in TIRA. If RMDs at that time began at 72, the factor was 27.3, so RMD would be about $36,600. I then add pensions and 85% of SS to that and determine what the tax is on the RMD. If it's 22% using current brackets, I convert as much as possible at 12%.

Then in 2018, suppose after the conversions and 2017 returns, I now have $980K in the TIRA. So now I calculate the first RMD based on that new balance using the same age factor of 27.3 and using the new tax brackets, etc.

I've done more analysis than that, but thought that doing just this much would be looking at 1 year at a time.