Asymmetric risk in the Roth vs Traditional question

Discuss all general (i.e. non-personal) investing questions and issues, investing news, and theory.
Topic Author
Northern Flicker
Posts: 5856
Joined: Fri Apr 10, 2015 12:29 am

Re: Asymmetric risk in the Roth vs Traditional question

Post by Northern Flicker » Mon Jun 29, 2020 5:47 pm

TomatoTomahto wrote:
Mon Jun 29, 2020 3:00 pm
It sometimes happens, but am I the only one who has lost the plot?
Summary of the original basis for my starting the thread.

1. Marginal tax rate in retirement is not a single, fixed constant, but a varying sequence of yet to be determined tax rates based on many variables (withdrawal rate, investment return, marital status, future tax laws, country and state of residence).

2. The Roth vs trad decision process should state an objective to optimize as a probabilistic model, e.g. minimize the probability of having a shortfall, maximize the expected value of a bequest, maximize expected value of sustainable withdrawals, etc.

3. Roth assets amplify risk (whether measured as after-tax asset volatility, after-tax max drawdown, or risk of an asset shortfall) and would require a somewhat more conservative allocation to hold risk constant vs trad assets to hold risk constant.

4. Because of the intractability of the complex probabilistic models, heuristics will be needed to make the decision as effectively as possible. The decision procedure in the wiki and the one in a referenced Vanguard article are reasonable first cut tries at a heuristic, but I believe are biased to the goal of maximizing one's expected value of sustainable withdrawal or size of bequest at the expense of taking more risk. This is akin to a risk-reward tradeoff in investing. But it is not presently clear what the best heuristics are for any particular optimization objectives. My own experience has suggested that the decision method in the wiki biases toward Roth decisions, and may be suboptimal by doing so.

It was not my intent to advocate for trad or Roth as a solution, but to advocate for a better decision process. The level of trad or Roth that is best is a function of individual goals and individual situation and net worth.
Risk is not a guarantor of return.

ruud
Posts: 295
Joined: Sat Mar 03, 2007 1:28 pm
Location: san francisco bay area

Re: Asymmetric risk in the Roth vs Traditional question

Post by ruud » Mon Jun 29, 2020 6:13 pm

Northern Flicker wrote:
Mon Jun 29, 2020 5:47 pm
3. Roth assets amplify risk (whether measured as after-tax asset volatility, after-tax max drawdown, or risk of an asset shortfall) and would require a somewhat more conservative allocation to hold risk constant vs trad assets to hold risk constant.
Isn't this just a consequence of a Roth dollar being worth more than a Traditional dollar?
.

User avatar
grabiner
Advisory Board
Posts: 27175
Joined: Tue Feb 20, 2007 11:58 pm
Location: Columbia, MD

Re: Asymmetric risk in the Roth vs Traditional question

Post by grabiner » Mon Jun 29, 2020 6:28 pm

ruud wrote:
Mon Jun 29, 2020 6:13 pm
Northern Flicker wrote:
Mon Jun 29, 2020 5:47 pm
3. Roth assets amplify risk (whether measured as after-tax asset volatility, after-tax max drawdown, or risk of an asset shortfall) and would require a somewhat more conservative allocation to hold risk constant vs trad assets to hold risk constant.
Isn't this just a consequence of a Roth dollar being worth more than a Traditional dollar?
Yes. If you will be withdrawing at a 25% marginal tax rate, $4000 in a traditional account and $3000 in a Roth account are equivalent, as both will give you the same after-tax spending if allocated the same way.

But one issue which came up in this thread is that if you are forced to retire early, you may retire in a lower tax bracket; now $4000 in a traditional account is equivalent to $3400 in a Roth account if your marginal tax rate is 15%. The larger your Roth, the less you benefited from the lower tax bracket.
Wiki David Grabiner

User avatar
firebirdparts
Posts: 1581
Joined: Thu Jun 13, 2019 4:21 pm

Re: Asymmetric risk in the Roth vs Traditional question

Post by firebirdparts » Mon Jun 29, 2020 7:10 pm

ruud wrote:
Mon Jun 29, 2020 6:13 pm
Northern Flicker wrote:
Mon Jun 29, 2020 5:47 pm
3. Roth assets amplify risk (whether measured as after-tax asset volatility, after-tax max drawdown, or risk of an asset shortfall) and would require a somewhat more conservative allocation to hold risk constant vs trad assets to hold risk constant.
Isn't this just a consequence of a Roth dollar being worth more than a Traditional dollar?
Part of that is a consequence of a Roth dollar costing more than a traditional dollar. It might not be worth more (you worry about that for yourself), but it definitely cost me more.
A fool and your money are soon partners

Topic Author
Northern Flicker
Posts: 5856
Joined: Fri Apr 10, 2015 12:29 am

Re: Asymmetric risk in the Roth vs Traditional question

Post by Northern Flicker » Mon Jun 29, 2020 8:13 pm

ruud wrote:
Mon Jun 29, 2020 6:13 pm
Northern Flicker wrote:
Mon Jun 29, 2020 5:47 pm
3. Roth assets amplify risk (whether measured as after-tax asset volatility, after-tax max drawdown, or risk of an asset shortfall) and would require a somewhat more conservative allocation to hold risk constant vs trad assets to hold risk constant.
Isn't this just a consequence of a Roth dollar being worth more than a Traditional dollar?
It is more subtle than that. It is because the extent of how much more valuable Roth space is varies commensurately with asset and spending level. As an example, consider the limiting case. If your spending and withdrawals fall into the zero bracket, then $1 in Roth has the same value as $1 in trad and you paid tax on the remaining Roth assets unnecessarily.
Risk is not a guarantor of return.

MrDrinkingWater
Posts: 166
Joined: Tue Dec 23, 2014 11:30 am

Re: Asymmetric risk in the Roth vs Traditional question

Post by MrDrinkingWater » Mon Jun 29, 2020 8:41 pm

grabiner wrote:
Mon Jun 29, 2020 9:34 am
whereskyle wrote:
Mon Jun 29, 2020 9:04 am
Is there not a viable argument that the Roth reliably reduces risks in virtually every case where contributors/converters are married with children, such that the marginal tax rate forecast becomes the most unreliable aspect of the prediction? In contrast to predicting future marginal tax rates (which seems to me a bit silly for Bogleheads. We don't predict the market but we will predict tax rates?), no matter what, there will be no child tax credit in retirement (in the majority of cases), and, no matter what, one of the spouses will die during retirement, putting the survivor in a higher tax bracket with likely similar expenses and need for similar annual income. Given that these two facts are virtual certainties (excluding simultaneous death and later in life child-rearing), while future marginal tax rates are anyone's guess, doesn't the Roth more likely than not win on the probabilities for the married with children investor?
The child tax credit is a fixed-dollar amount, and the pre-2018 exemptions were fixed reductions in income. Unless you are in the phase-out of the credit, or the personal exemptions cause you to cross a tax bracket boundary, neither one affects your maginal tax rate.

What happens after one spouse dies does favor having some Roth money, to reduce the risk that RMDs will be more than the other spouse needs and will push the survivor into a higher tax bracket. The surviving spouse can then withdraw from the traditional account up to the top of the appropriate tax bracket, then take all other expenses from a Roth (or a taxable account where capital gains are taxed at a nearly-fixed 15%). This issue may also be better resolved by converting traditional to Roth in retirement (which also reduces RMDs), rather than contributing to Roth in a higher tax bracket while working.
FiveK wrote:
Sat Jun 27, 2020 8:45 pm
<SNIP>
AFAIK, no such tool exists publicly. Several good tools handle pieces, e.g., opensocialsecurity, RPM, financial toolbox, and I-ORP, but nothing does it all. And that's for the "simple" case of specified life expectancies, investment returns, tax law, etc., let alone a probabilistic approach.

<SNIP>
Another common issue is MFJ vs. single filing status. That one is probabilistic (assuming the change occurs due to unforeseen death).
I'm late to this thread, so I apologize for that.

I'd just like to report that I ran a test case using RPM where DW is 4 years older than DH. In a scenario, where DH lives to be 92 and DW lives to be 96, they stay in the MFJ filing status. In this scenario both DH and DW are retired at 56 and 60 years of age, respectively. I tried to keep effective tax rate pretty consistently in the 17% effective tax rate range, doing Roth conversions that were sized to remain in that effective tax rate, and was able to a keep marginal tax rates for the hypothetical couple mostly in the 22% marginal rate bracket, with some tax years moving into the 24% marginal rate bracket in the middle-end portion of the scenario, before dropping in the final years where they are mostly spending their Roth conversion account last.

Then, I made adjustments to that scenario in the Setup page, including social security payout adjustments, where DH lives to age 84 and DW lives to 96. When the DW changes to Single Filing status, the tax rates for DW jump into the 32% marginal rate bracket, and even has one year in the 37% marginal rate bracket. I made some adjustments to this scenario to make the effective tax rate pretty consistently in the 19% effective tax rate range, while DH and DW are both still kicking, by doing larger Roth conversions that were sized to remain in that 19% effective tax rate range. This reduced the size of the RMDs for the DW, as a widow, later. After making these adjustments, when DW changes to having a single filing status, the effective tax rate remains in the 19% effective tax rate range, the marginal rates do not spike into the 32% marginal rate bracket at all. DW is making a larger draw from the Roth accounts sooner.

This isn't a lot of testing, nor is it exhaustive in any way. It does suggest that a middle course of action might be best. Put retirement savings in 401(k) and Traditional IRA accounts, and if you have enough runway to do Roth conversions between age 60 and age 72, do the Roth conversions that will keep the surviving spouse out of the higher marginal tax rate brackets later, when he or she becomes a single filer. You might pay more in income taxes for the first ten years or so of retirement, but the surviving spouse will pay far less income tax as a single filer. This looks like it results in more spendable money and probably paying the lowest legal amount of income tax. We may not know how many years we have on this earth, but we can use the statistics that we have to help guide us in our decision-making.

User avatar
teen persuasion
Posts: 1155
Joined: Sun Oct 25, 2015 1:43 pm

Re: Asymmetric risk in the Roth vs Traditional question

Post by teen persuasion » Mon Jun 29, 2020 8:50 pm

Northern Flicker wrote:
Mon Jun 29, 2020 8:13 pm
ruud wrote:
Mon Jun 29, 2020 6:13 pm
Northern Flicker wrote:
Mon Jun 29, 2020 5:47 pm
3. Roth assets amplify risk (whether measured as after-tax asset volatility, after-tax max drawdown, or risk of an asset shortfall) and would require a somewhat more conservative allocation to hold risk constant vs trad assets to hold risk constant.
Isn't this just a consequence of a Roth dollar being worth more than a Traditional dollar?
It is more subtle than that. It is because the extent of how much more valuable Roth space is varies commensurately with asset and spending level. As an example, consider the limiting case. If your spending and withdrawals fall into the zero bracket, then $1 in Roth has the same value as $1 in trad and you paid tax on the remaining Roth assets unnecessarily.
Unless, of course, you paid no extra tax to contribute that $1 to Roth.

Lee_WSP
Posts: 2693
Joined: Fri Apr 19, 2019 5:15 pm
Location: Arizona

Re: Asymmetric risk in the Roth vs Traditional question

Post by Lee_WSP » Mon Jun 29, 2020 11:41 pm

grabiner wrote:
Mon Jun 29, 2020 6:28 pm
ruud wrote:
Mon Jun 29, 2020 6:13 pm
Northern Flicker wrote:
Mon Jun 29, 2020 5:47 pm
3. Roth assets amplify risk (whether measured as after-tax asset volatility, after-tax max drawdown, or risk of an asset shortfall) and would require a somewhat more conservative allocation to hold risk constant vs trad assets to hold risk constant.
Isn't this just a consequence of a Roth dollar being worth more than a Traditional dollar?
Yes. If you will be withdrawing at a 25% marginal tax rate, $4000 in a traditional account and $3000 in a Roth account are equivalent, as both will give you the same after-tax spending if allocated the same way.

But one issue which came up in this thread is that if you are forced to retire early, you may retire in a lower tax bracket; now $4000 in a traditional account is equivalent to $3400 in a Roth account if your marginal tax rate is 15%. The larger your Roth, the less you benefited from the lower tax bracket.
Help me understand the nuance trying to be advocated by OP.

Is this only a problem if you have only Roth assets, or is it still a problem if you have a blend? If so, are there deciles or quin-tiles or 50/50 probabilistic scenarios where having the "too much Roth" is going to be a problem?

In other words, I'm still having a hard time with the practical implications. I can understand how it can be a major problem if one has all their savings in Roth vehicles. However, if one has even 50% in traditional vehicles, then one will eventually be facing RMD's and/or will be drawing down the trad assets and thus incurring income taxes. Incurring income taxes would quite quickly bump one up the income ladders.

Not only that, but if one was in the position of having too much Roth, but one had trad assets, you could shift your drawdown schedule to take advantage of the lower tax rates "today" and withold the Roth assets for later years; possibly avoiding RMD's and controlling one's tax schedule.

Topic Author
Northern Flicker
Posts: 5856
Joined: Fri Apr 10, 2015 12:29 am

Re: Asymmetric risk in the Roth vs Traditional question

Post by Northern Flicker » Tue Jun 30, 2020 12:32 am

teen persuasion wrote:
Mon Jun 29, 2020 8:50 pm
Northern Flicker wrote:
Mon Jun 29, 2020 8:13 pm
ruud wrote:
Mon Jun 29, 2020 6:13 pm
Northern Flicker wrote:
Mon Jun 29, 2020 5:47 pm
3. Roth assets amplify risk (whether measured as after-tax asset volatility, after-tax max drawdown, or risk of an asset shortfall) and would require a somewhat more conservative allocation to hold risk constant vs trad assets to hold risk constant.
Isn't this just a consequence of a Roth dollar being worth more than a Traditional dollar?
It is more subtle than that. It is because the extent of how much more valuable Roth space is varies commensurately with asset and spending level. As an example, consider the limiting case. If your spending and withdrawals fall into the zero bracket, then $1 in Roth has the same value as $1 in trad and you paid tax on the remaining Roth assets unnecessarily.
Unless, of course, you paid no extra tax to contribute that $1 to Roth.
Yes, I could have been more rigorous in defining the assumption if non-zero tax on the Roth $1 when describing the explanatory example.
Risk is not a guarantor of return.

Topic Author
Northern Flicker
Posts: 5856
Joined: Fri Apr 10, 2015 12:29 am

Re: Asymmetric risk in the Roth vs Traditional question

Post by Northern Flicker » Tue Jun 30, 2020 12:44 am

Lee_WSP wrote:
Mon Jun 29, 2020 11:41 pm
grabiner wrote:
Mon Jun 29, 2020 6:28 pm
ruud wrote:
Mon Jun 29, 2020 6:13 pm
Northern Flicker wrote:
Mon Jun 29, 2020 5:47 pm
3. Roth assets amplify risk (whether measured as after-tax asset volatility, after-tax max drawdown, or risk of an asset shortfall) and would require a somewhat more conservative allocation to hold risk constant vs trad assets to hold risk constant.
Isn't this just a consequence of a Roth dollar being worth more than a Traditional dollar?
Yes. If you will be withdrawing at a 25% marginal tax rate, $4000 in a traditional account and $3000 in a Roth account are equivalent, as both will give you the same after-tax spending if allocated the same way.

But one issue which came up in this thread is that if you are forced to retire early, you may retire in a lower tax bracket; now $4000 in a traditional account is equivalent to $3400 in a Roth account if your marginal tax rate is 15%. The larger your Roth, the less you benefited from the lower tax bracket.
Help me understand the nuance trying to be advocated by OP.

Is this only a problem if you have only Roth assets, or is it still a problem if you have a blend? If so, are there deciles or quin-tiles or 50/50 probabilistic scenarios where having the "too much Roth" is going to be a problem?

In other words, I'm still having a hard time with the practical implications. I can understand how it can be a major problem if one has all their savings in Roth vehicles. However, if one has even 50% in traditional vehicles, then one will eventually be facing RMD's and/or will be drawing down the trad assets and thus incurring income taxes. Incurring income taxes would quite quickly bump one up the income ladders.

Not only that, but if one was in the position of having too much Roth, but one had trad assets, you could shift your drawdown schedule to take advantage of the lower tax rates "today" and withold the Roth assets for later years; possibly avoiding RMD's and controlling one's tax schedule.
The problem with this thread is that I'm advocating for a different Roth vs trad decision process and many respondents are replying to an imagined position of advocating for not using Roth assets, which is not my position.

My point is not that any particular plan or configuration is bad for anyone in particular, but that the optimal decision process is probabilistic in nature. In fact, a couple of posters have brought up the scenario of one spouse of a couple dying well before the other leading to an jbcrease in marginal tax rate, a probabilistic risk for trad assets. This is tempered by a single person needing less to live on, but it is unclear whether after-tax sustainable withdrawal decreases by mire or by less than expenses.
Risk is not a guarantor of return.

Lee_WSP
Posts: 2693
Joined: Fri Apr 19, 2019 5:15 pm
Location: Arizona

Re: Asymmetric risk in the Roth vs Traditional question

Post by Lee_WSP » Tue Jun 30, 2020 12:47 am

As far as I'm concerned, as long as you dont go out of your way to do a Roth 401k or Roth conversions (that aren't backdoor) i think you'll be fine no matter what. Ie, you'd need a good reason to convert from trad to Roth.

rossington
Posts: 669
Joined: Fri Jun 07, 2019 2:00 am
Location: Florida

Re: Asymmetric risk in the Roth vs Traditional question

Post by rossington » Tue Jun 30, 2020 4:16 am

Northern Flicker wrote:
Tue Jun 30, 2020 12:44 am
The problem with this thread is that I'm advocating for a different Roth vs trad decision process and many respondents are replying to an imagined position of advocating for not using Roth assets, which is not my position.

My point is not that any particular plan or configuration is bad for anyone in particular, but that the optimal decision process is probabilistic in nature. In fact, a couple of posters have brought up the scenario of one spouse of a couple dying well before the other leading to an jbcrease in marginal tax rate, a probabilistic risk for trad assets. This is tempered by a single person needing less to live on, but it is unclear whether after-tax sustainable withdrawal decreases by mire or by less than expenses.
So what is your conclusion? Is it a one size fits all approach to allocating between Roth and Traditional?
I would think not.
Rather as you have alluded to:
The end result is ultimately dependent on how much one can save and how one allocates those savings over time.
Thus the retirement time frame (which is completely variable ) can only be applied on an individual level because everyone has different circumstances.
So tax rates and expenses can be predicted....but nothing more.
One has to make allocation decisions within their retirement accounts over their life as they see fit.
"Success is going from failure to failure without loss of enthusiasm." Winston Churchill.

conservativeinvestor
Posts: 45
Joined: Tue Feb 05, 2019 12:14 pm

Re: Asymmetric risk in the Roth vs Traditional question

Post by conservativeinvestor » Tue Jun 30, 2020 5:17 am

rossington wrote:
Tue Jun 30, 2020 4:16 am
Northern Flicker wrote:
Tue Jun 30, 2020 12:44 am
The problem with this thread is that I'm advocating for a different Roth vs trad decision process and many respondents are replying to an imagined position of advocating for not using Roth assets, which is not my position.

My point is not that any particular plan or configuration is bad for anyone in particular, but that the optimal decision process is probabilistic in nature. In fact, a couple of posters have brought up the scenario of one spouse of a couple dying well before the other leading to an jbcrease in marginal tax rate, a probabilistic risk for trad assets. This is tempered by a single person needing less to live on, but it is unclear whether after-tax sustainable withdrawal decreases by mire or by less than expenses.
So what is your conclusion? Is it a one size fits all approach to allocating between Roth and Traditional?
I would think not.
Rather as you have alluded to:
The end result is ultimately dependent on how much one can save and how one allocates those savings over time.
Thus the retirement time frame (which is completely variable ) can only be applied on an individual level because everyone has different circumstances.
So tax rates and expenses can be predicted....but nothing more.
One has to make allocation decisions within their retirement accounts over their life as they see fit.
After reading through all of this I get the impression that the OP doesn't actually have a conclusion and is using a lot of big words to say "nobody knows nothing". I'm more confused now about what exactly the problem being discussed is than before 6 pages of discussion occurred trying to solve it.

Topic Author
Northern Flicker
Posts: 5856
Joined: Fri Apr 10, 2015 12:29 am

Re: Asymmetric risk in the Roth vs Traditional question

Post by Northern Flicker » Tue Jun 30, 2020 2:24 pm

conservativeinvestor wrote:
Tue Jun 30, 2020 5:17 am
rossington wrote:
Tue Jun 30, 2020 4:16 am
Northern Flicker wrote:
Tue Jun 30, 2020 12:44 am
The problem with this thread is that I'm advocating for a different Roth vs trad decision process and many respondents are replying to an imagined position of advocating for not using Roth assets, which is not my position.

My point is not that any particular plan or configuration is bad for anyone in particular, but that the optimal decision process is probabilistic in nature. In fact, a couple of posters have brought up the scenario of one spouse of a couple dying well before the other leading to an jbcrease in marginal tax rate, a probabilistic risk for trad assets. This is tempered by a single person needing less to live on, but it is unclear whether after-tax sustainable withdrawal decreases by mire or by less than expenses.
So what is your conclusion? Is it a one size fits all approach to allocating between Roth and Traditional?
I would think not.
Rather as you have alluded to:
The end result is ultimately dependent on how much one can save and how one allocates those savings over time.
Thus the retirement time frame (which is completely variable ) can only be applied on an individual level because everyone has different circumstances.
So tax rates and expenses can be predicted....but nothing more.
One has to make allocation decisions within their retirement accounts over their life as they see fit.
After reading through all of this I get the impression that the OP doesn't actually have a conclusion and is using a lot of big words to say "nobody knows nothing". I'm more confused now about what exactly the problem being discussed is than before 6 pages of discussion occurred trying to solve it.
The question is: what is a good procedure for making a decision on how much of a retirement contribution should be trad or Roth, or whether to do Roth conversions and by how much. My posting was not trying to provide a solution, but pointing out that existing solutions could be improved by treating the problem probabilistically.
Risk is not a guarantor of return.

Lee_WSP
Posts: 2693
Joined: Fri Apr 19, 2019 5:15 pm
Location: Arizona

Re: Asymmetric risk in the Roth vs Traditional question

Post by Lee_WSP » Tue Jun 30, 2020 3:41 pm

Northern Flicker wrote:
Tue Jun 30, 2020 2:24 pm
conservativeinvestor wrote:
Tue Jun 30, 2020 5:17 am
rossington wrote:
Tue Jun 30, 2020 4:16 am
Northern Flicker wrote:
Tue Jun 30, 2020 12:44 am
The problem with this thread is that I'm advocating for a different Roth vs trad decision process and many respondents are replying to an imagined position of advocating for not using Roth assets, which is not my position.

My point is not that any particular plan or configuration is bad for anyone in particular, but that the optimal decision process is probabilistic in nature. In fact, a couple of posters have brought up the scenario of one spouse of a couple dying well before the other leading to an jbcrease in marginal tax rate, a probabilistic risk for trad assets. This is tempered by a single person needing less to live on, but it is unclear whether after-tax sustainable withdrawal decreases by mire or by less than expenses.
So what is your conclusion? Is it a one size fits all approach to allocating between Roth and Traditional?
I would think not.
Rather as you have alluded to:
The end result is ultimately dependent on how much one can save and how one allocates those savings over time.
Thus the retirement time frame (which is completely variable ) can only be applied on an individual level because everyone has different circumstances.
So tax rates and expenses can be predicted....but nothing more.
One has to make allocation decisions within their retirement accounts over their life as they see fit.
After reading through all of this I get the impression that the OP doesn't actually have a conclusion and is using a lot of big words to say "nobody knows nothing". I'm more confused now about what exactly the problem being discussed is than before 6 pages of discussion occurred trying to solve it.
The question is: what is a good procedure for making a decision on how much of a retirement contribution should be trad or Roth, or whether to do Roth conversions and by how much. My posting was not trying to provide a solution, but pointing out that existing solutions could be improved by treating the problem probabilistically.
You can't know because the future is unknown. Just as we can't know whether the stock market will end the year on an up or down note on January 1st. All we know is that X/Y of the time such and such has historically happened.

User avatar
ray.james
Posts: 1324
Joined: Tue Jul 19, 2011 4:08 am

Re: Asymmetric risk in the Roth vs Traditional question

Post by ray.james » Tue Jun 30, 2020 4:11 pm

Lee_WSP wrote:
Tue Jun 30, 2020 12:47 am
As far as I'm concerned, as long as you dont go out of your way to do a Roth 401k or Roth conversions (that aren't backdoor) i think you'll be fine no matter what. Ie, you'd need a good reason to convert from trad to Roth.
I guess you are more in agreement with OP than not. OP is simply trying to understand/explore the reasons for such(highlighted above) outcome for most people. It is an approach of applying probability to events and measuring expected outcomes. ( your X/Y of the time).

This is a good explanation of what is happening in this thread. One cannot simply state it is unknown. Of course, it is unknown. But we can bet based on expected value. Rather than,
https://math.stackexchange.com/question ... tself-bein
When in doubt, http://www.bogleheads.org/forum/viewtopic.php?f=1&t=79939

petulant
Posts: 1325
Joined: Thu Sep 22, 2016 1:09 pm

Re: Asymmetric risk in the Roth vs Traditional question

Post by petulant » Tue Jun 30, 2020 6:25 pm

privateID wrote:
Mon Jun 29, 2020 7:24 am
This discussion reminds me of the often stated advice to put riskier assets into Roth accounts. Although I understood the rationale, a part of me never totally agreed with it. I viewed it as 2 scenarios:

1) You put the riskiest assets in a Roth (safer assets in non-Roth), they perform as expected, when you withdraw the money all is well.

2) You put the riskiest assets in a Roth, the risk shows up, they don't perform as expected, when you withdraw the money you have alot less than expected.

My thinking years ago was to mitigate the risk of under-performance by putting some risky assets in our Roth accounts but also putting some non-risky assets in there as well. If the risky assets perform well, I won't have as much tax-free (because of the non-risky assets), but in that case I am still doing well since their was no under-performance. But if the risk showed up and the risky assets under-performed, I would not be as bad because of those non-risky assets. So, I avoided the worst case scenario of under-performance and the Roth filled will all risky assets. I therefore put all the risky assets in one Roth IRA and was much more conservative in our other Roth IRA (put more TIPS and the like in there).

Fast forward many years later and we have one Roth IRA much larger than the other. Of course, this meant I put more risky assets in our traditional IRA and that balance is higher than it would have been otherwise. Not a major problem, but now trying to balance that all out with retirement not too far away and the threat of large RMDs potentially causing problems. Of course, if the balances in those Roth IRAs were switched (higher risky assets under-performed), then I'd feel alot smarter today.
You bring up an underappreciated risk with traditional accounts that often goes without discussion. An investor really can save so much in traditional accounts that they can end up with tax management problems involving breaking into higher tax brackets, unwanted RMDs, and taxability of SS. I'm not saying that to discourage people from using traditional accounts or to disagree with the OP per se. What I have a point about is asset location.

You mentioned you made the conscious choice to put some safe assets in the Roth account and some riskier, high-return assets in the traditional account. The conventional wisdom is to put bonds in traditional accounts first and stocks in Roth accounts first. Really, having bonds in the traditional account would be much less volatile when thinking a few years down the road for tax management purposes. It is much more reliable and precise to plan for Roth conversions, IRMAA, RMDs, and SS taxability when the account is heavy on bonds.

Roth IRAs do not have income taxes or RMDs. Withdrawals do not impact SS taxability, and inheritance of a Roth account does not create tax headaches for heirs. Overall, there are many fewer tax management issues with Roth accounts. Placing assets with the most uncertainty in Roth accounts makes sense.

I think the optimum strategy for most people is to follow the conventional wisdom and to include an aggressive glidepath toward bonds as one approaches their 60s, with those bonds located in traditional accounts. Beyond all the things we say about sequence of returns risk, this has the added benefit of making tax strategies much more predictable and manageable.

privateID
Posts: 162
Joined: Sat Oct 18, 2014 4:59 pm

Re: Asymmetric risk in the Roth vs Traditional question

Post by privateID » Tue Jun 30, 2020 6:55 pm

petulant wrote:
Tue Jun 30, 2020 6:25 pm
You bring up an underappreciated risk with traditional accounts that often goes without discussion. An investor really can save so much in traditional accounts that they can end up with tax management problems involving breaking into higher tax brackets, unwanted RMDs, and taxability of SS. I'm not saying that to discourage people from using traditional accounts or to disagree with the OP per se. What I have a point about is asset location.

You mentioned you made the conscious choice to put some safe assets in the Roth account and some riskier, high-return assets in the traditional account. The conventional wisdom is to put bonds in traditional accounts first and stocks in Roth accounts first. Really, having bonds in the traditional account would be much less volatile when thinking a few years down the road for tax management purposes. It is much more reliable and precise to plan for Roth conversions, IRMAA, RMDs, and SS taxability when the account is heavy on bonds.

Roth IRAs do not have income taxes or RMDs. Withdrawals do not impact SS taxability, and inheritance of a Roth account does not create tax headaches for heirs. Overall, there are many fewer tax management issues with Roth accounts. Placing assets with the most uncertainty in Roth accounts makes sense.

I think the optimum strategy for most people is to follow the conventional wisdom and to include an aggressive glidepath toward bonds as one approaches their 60s, with those bonds located in traditional accounts. Beyond all the things we say about sequence of returns risk, this has the added benefit of making tax strategies much more predictable and manageable.
In retrospect, I totally agree. Of course, I wasn't thinking about Roth conversions, IRMAA, RMDs, and SS taxability when I made those decisions. BTW - another reason I put things like TIPS in my Roth was due to the fact that I had an opportunity to buy individual TIPS there but not in my 401K. Having said all that, I had a golden opportunity to correct that error a couple of months ago. Unfortunately, I still wasn't thinking Roth conversions, IRMAA, RMDs, and SS taxability in March. I am now thanks to this forum. I am still holding out for another dip to move things around now that I see the light. It may never come.

Lee_WSP
Posts: 2693
Joined: Fri Apr 19, 2019 5:15 pm
Location: Arizona

Re: Asymmetric risk in the Roth vs Traditional question

Post by Lee_WSP » Tue Jun 30, 2020 9:34 pm

ray.james wrote:
Tue Jun 30, 2020 4:11 pm
Lee_WSP wrote:
Tue Jun 30, 2020 12:47 am
As far as I'm concerned, as long as you dont go out of your way to do a Roth 401k or Roth conversions (that aren't backdoor) i think you'll be fine no matter what. Ie, you'd need a good reason to convert from trad to Roth.
I guess you are more in agreement with OP than not. OP is simply trying to understand/explore the reasons for such(highlighted above) outcome for most people. It is an approach of applying probability to events and measuring expected outcomes. ( your X/Y of the time).

This is a good explanation of what is happening in this thread. One cannot simply state it is unknown. Of course, it is unknown. But we can bet based on expected value. Rather than,
https://math.stackexchange.com/question ... tself-bein
I was only confused as to the purported impact OP was implying, but OP clarified.

Topic Author
Northern Flicker
Posts: 5856
Joined: Fri Apr 10, 2015 12:29 am

Re: Asymmetric risk in the Roth vs Traditional question

Post by Northern Flicker » Wed Jul 01, 2020 12:30 am

me wrote: The question is: what is a good procedure for making a decision on how much of a retirement contribution should be trad or Roth, or whether to do Roth conversions and by how much. My posting was not trying to provide a solution, but pointing out that existing solutions could be improved by treating the problem probabilistically.
Lee_WSP wrote: You can't know because the future is unknown. Just as we can't know whether the stock market will end the year on an up or down note on January 1st. All we know is that X/Y of the time such and such has historically happened.
But that is the whole point of the thread, that is that we cannot have a deterministic model because we cannot predict the future. We can however decide which risks are the most consequential for our situation and bias our choices to reduce those risks. And while probability distributions for equity returns are not known, life expectancies are well studied for instance to evaluate the risk of tax rate changes on a surviving spouse. And sustainable withdrawal research provides some input into the probability of exhausting assets at a given withdrawal level.

"Nobody knows nuthin" thus would be too strong of an assessment. There are things whose likelihood we cannot quantify, but that does not apply to all of the inputs to the decision.

Another interesting tidbit is that if you buy a nominal SPIA (and I don't think inflation-adjusted SPIAs exist any more) then the fact that tax bracket tables are indexed for inflation means that if we have inflation, the rising tax rate boundaries will lower the tax on a nominal SPIA in a traditional IRA but not in a Roth IRA, so using traditional assets for that mitigates some of the inflation risk (about 12% of it if you are in a 12% bracket etc.)
Risk is not a guarantor of return.

petulant
Posts: 1325
Joined: Thu Sep 22, 2016 1:09 pm

Re: Asymmetric risk in the Roth vs Traditional question

Post by petulant » Wed Jul 01, 2020 10:51 am

Northern Flicker wrote:
Wed Jul 01, 2020 12:30 am
me wrote: The question is: what is a good procedure for making a decision on how much of a retirement contribution should be trad or Roth, or whether to do Roth conversions and by how much. My posting was not trying to provide a solution, but pointing out that existing solutions could be improved by treating the problem probabilistically.
Lee_WSP wrote: You can't know because the future is unknown. Just as we can't know whether the stock market will end the year on an up or down note on January 1st. All we know is that X/Y of the time such and such has historically happened.
But that is the whole point of the thread, that is that we cannot have a deterministic model because we cannot predict the future. We can however decide which risks are the most consequential for our situation and bias our choices to reduce those risks. And while probability distributions for equity returns are not known, life expectancies are well studied for instance to evaluate the risk of tax rate changes on a surviving spouse. And sustainable withdrawal research provides some input into the probability of exhausting assets at a given withdrawal level.

"Nobody knows nuthin" thus would be too strong of an assessment. There are things whose likelihood we cannot quantify, but that does not apply to all of the inputs to the decision.

Another interesting tidbit is that if you buy a nominal SPIA (and I don't think inflation-adjusted SPIAs exist any more) then the fact that tax bracket tables are indexed for inflation means that if we have inflation, the rising tax rate boundaries will lower the tax on a nominal SPIA in a traditional IRA but not in a Roth IRA, so using traditional assets for that mitigates some of the inflation risk (about 12% of it if you are in a 12% bracket etc.)
The interaction between SPIAs and traditional/Roth accounts is a lot more complicated than that because of the phase-in for the taxability of social security benefits. Having traditional accounts in any way causes more taxable social security, and over time inflation increase this taxability since the phase-in start point is not indexed for inflation. Thus, for the average American investor with a fixed SPIA inside a traditional IRA who also has social security, they may never feel like the inflation helps them compared to the same investor with a Roth IRA.

User avatar
firebirdparts
Posts: 1581
Joined: Thu Jun 13, 2019 4:21 pm

Re: Asymmetric risk in the Roth vs Traditional question

Post by firebirdparts » Wed Jul 01, 2020 11:14 am

Northern Flicker wrote:
Tue Jun 30, 2020 2:24 pm
The question is: what is a good procedure for making a decision on how much of a retirement contribution should be trad or Roth, or whether to do Roth conversions and by how much. My posting was not trying to provide a solution, but pointing out that existing solutions could be improved by treating the problem probabilistically.
You're right about that.
A fool and your money are soon partners

whereskyle
Posts: 889
Joined: Wed Jan 29, 2020 10:29 am

Re: Asymmetric risk in the Roth vs Traditional question

Post by whereskyle » Wed Jul 01, 2020 11:47 am

Northern Flicker wrote:
Wed Jul 01, 2020 12:30 am
me wrote: The question is: what is a good procedure for making a decision on how much of a retirement contribution should be trad or Roth, or whether to do Roth conversions and by how much. My posting was not trying to provide a solution, but pointing out that existing solutions could be improved by treating the problem probabilistically.
Lee_WSP wrote: You can't know because the future is unknown. Just as we can't know whether the stock market will end the year on an up or down note on January 1st. All we know is that X/Y of the time such and such has historically happened.
But that is the whole point of the thread, that is that we cannot have a deterministic model because we cannot predict the future. We can however decide which risks are the most consequential for our situation and bias our choices to reduce those risks. And while probability distributions for equity returns are not known, life expectancies are well studied for instance to evaluate the risk of tax rate changes on a surviving spouse. And sustainable withdrawal research provides some input into the probability of exhausting assets at a given withdrawal level.

"Nobody knows nuthin" thus would be too strong of an assessment. There are things whose likelihood we cannot quantify, but that does not apply to all of the inputs to the decision.

Another interesting tidbit is that if you buy a nominal SPIA (and I don't think inflation-adjusted SPIAs exist any more) then the fact that tax bracket tables are indexed for inflation means that if we have inflation, the rising tax rate boundaries will lower the tax on a nominal SPIA in a traditional IRA but not in a Roth IRA, so using traditional assets for that mitigates some of the inflation risk (about 12% of it if you are in a 12% bracket etc.)
The likelihoods of which inputs can we quantify? For many of us (I'm 31), we have no idea what social security will look like, we don't know what the standard deduction will be, and we don't know what future marginal tax rates will be.

The logic of your recommended approach requires us to assume that these factors are predictable, if not fixed, does it not?

In the end, the same old rule seems to apply: if you think your tax rate will be higher in retirement, choose the Roth.
"I am better off than he is – for he knows nothing and thinks that he knows. I neither know nor think that I know." - Socrates. "Nobody knows nothing." - Jack Bogle

KlangFool
Posts: 16652
Joined: Sat Oct 11, 2008 12:35 pm

Re: Asymmetric risk in the Roth vs Traditional question

Post by KlangFool » Wed Jul 01, 2020 11:54 am

whereskyle wrote:
Wed Jul 01, 2020 11:47 am


In the end, the same old rule seems to apply: if you think your tax rate will be higher in retirement, choose the Roth.
whereskyle,

A) All the uncertainty forces that to be not true.

B) Only if everything went well, that could be true.

C) But, in that case, it is not too bad as compared to not having enough money.

After all said and done,

1) Max up your Trad 401K

2) Put your tax savings into Roth IRA

is still the best combination.

<<if you think your tax rate will be higher in retirement, choose the Roth.>>

We had been through this with many forum members. 90+% of the time, we can prove to the person that the thinking is wrong. Usually, the first question that we asked would be

"Do you do your own taxes?"

And, for many times, the answer would be no.

KlangFool

whereskyle
Posts: 889
Joined: Wed Jan 29, 2020 10:29 am

Re: Asymmetric risk in the Roth vs Traditional question

Post by whereskyle » Wed Jul 01, 2020 12:03 pm

KlangFool wrote:
Wed Jul 01, 2020 11:54 am
whereskyle wrote:
Wed Jul 01, 2020 11:47 am


In the end, the same old rule seems to apply: if you think your tax rate will be higher in retirement, choose the Roth.
whereskyle,

A) All the uncertainty forces that to be not true.

B) Only if everything went well, that could be true.

C) But, in that case, it is not too bad as compared to not having enough money.

After all said and done,

1) Max up your Trad 401K

2) Put your tax savings into Roth IRA

is still the best combination.

<<if you think your tax rate will be higher in retirement, choose the Roth.>>

We had been through this with many forum members. 90+% of the time, we can prove to the person that the thinking is wrong. Usually, the first question that we asked would be

"Do you do your own taxes?"

And, for many times, the answer would be no.

KlangFool
If the taxes on the first $30k of income rise to 15% and there is no standard deduction, then I will be in a better position if I contribute to a Roth with funds taxed today at 12%. The only way the OP's approach makes sense for me is if I assume that the tax code will resemble its current state when I retire (hopefully) in 30 years. I do not assume that. Instead, I focus on the fact that the present-day 12% rate on my contributions to the Roth are historically low. We can debate where tax rates will go all day, but in no way do I assume that they will be the same when I retire. I think current rates are ridiculously low and unsustainable. Therefore, I think it is important to question whether OP's advice is based on the tax code staying stable.
"I am better off than he is – for he knows nothing and thinks that he knows. I neither know nor think that I know." - Socrates. "Nobody knows nothing." - Jack Bogle

Topic Author
Northern Flicker
Posts: 5856
Joined: Fri Apr 10, 2015 12:29 am

Re: Asymmetric risk in the Roth vs Traditional question

Post by Northern Flicker » Wed Jul 01, 2020 12:04 pm

wherskyle wrote: The likelihoods of which inputs can we quantify? For many of us (I'm 31), we have no idea what social security will look like, we don't know what the standard deduction will be, and we don't know what future marginal tax rates will be.

The logic of your recommended approach requires us to assume that these factors are predictable, if not fixed, does it not?
You are actually saying the same thing as I am, which is that we cannot have a deterministic model because of the future unknowns. This has not been the prevailing view, which is why I started the thread.

At age 31, the best you probably can do likely is to decide which risks you don't wish to take and bias your Roth/trad decision accordingly. It is useful to note the asymmetric risk as well.

When close to or in retirement, you may be able to model some of the unknowns somewhat more realistically.

I get that alot of people using this site want clear, definitive prescriptions for managing their finances. Those types of solutions do not always exist.
Last edited by Northern Flicker on Wed Jul 01, 2020 11:14 pm, edited 4 times in total.
Risk is not a guarantor of return.

whereskyle
Posts: 889
Joined: Wed Jan 29, 2020 10:29 am

Re: Asymmetric risk in the Roth vs Traditional question

Post by whereskyle » Wed Jul 01, 2020 12:08 pm

Northern Flicker wrote:
Wed Jul 01, 2020 12:04 pm
wherskyle wrote: The likelihoods of which inputs can we quantify? For many of us (I'm 31), we have no idea what social security will look like, we don't know what the standard deduction will be, and we don't know what future marginal tax rates will be.

The logic of your recommended approach requires us to assume that these factors are predictable, if not fixed, does it not?
You are actually saying the same thing as I am, which is that we cannot have a deterministic model because of the future unknowns. This has not been the prevailing view, which is why I started the thread.

At age 31, the best you probably can do likely is to decide which risks you don't wish to take and bias your Roth/trad decision accordingly. Is is useful to note the asymnetric risk as well.

When close to or in retirement, you may be able to model sone of the unknowns somewhat more realistically.

I get that alot of people use this site wanting clear, definitive prescriptions for managing their finances. Those types of solutions do not always exist.
I entirely agree, and I appreciate your clarifying that these are indeed logical considerations to take into account when following your recommended approach.
"I am better off than he is – for he knows nothing and thinks that he knows. I neither know nor think that I know." - Socrates. "Nobody knows nothing." - Jack Bogle

KlangFool
Posts: 16652
Joined: Sat Oct 11, 2008 12:35 pm

Re: Asymmetric risk in the Roth vs Traditional question

Post by KlangFool » Wed Jul 01, 2020 12:16 pm

whereskyle wrote:
Wed Jul 01, 2020 12:03 pm

If the taxes on the first $30k of income rise to 15% and there is no standard deduction, then I will be in a better position if I contribute to a Roth with funds taxed today at 12%. The only way the OP's approach makes sense for me is if I assume that the tax code will resemble its current state when I retire (hopefully) in 30 years. I do not assume that. Instead, I focus on the fact that the present-day 12% rate on my contributions to the Roth are historically low. We can debate where tax rates will go all day, but in no way do I assume that they will be the same when I retire. I think current rates are ridiculously low and unsustainable. Therefore, I think it is important to question whether OP's advice is based on the tax code staying stable.
whereskyle,

<<I think current rates are ridiculously low and unsustainable. >>

If you spend some time studying any other democratic countries, you would find that in order to raise more tax revenues, the government reduces the income taxes and substitute that with a national consumption/sale tax. This had been the historical trend for the last 10 to 20 years.

Just because the tax would not be the same, it does not mean it would be an increase in income taxes.

<<if I contribute to a Roth with funds taxed today at 12%.>>

Depending on your income level, why would you choose to pay 12% when you can pay negative taxes? Aka, instead of paying taxes, you are paid by IRS due to tax credits.

<<If the taxes on the first $30k of income rise to 15% and there is no standard deduction,>>

Even assuming that is true, Trad 401K may still win. Have you done the actual calculation?

In summary, if you are interested, start your own topic and go over your own numbers.

KlangFool

afan
Posts: 4919
Joined: Sun Jul 25, 2010 4:01 pm

Re: Asymmetric risk in the Roth vs Traditional question

Post by afan » Sun Jul 05, 2020 2:52 pm

For me, the numbers favor contributing to Roth or doing conversions if I assume relatively high returns on the retirement assets and a relatively high tax bracket in retirement. But these are both impossible to predict. If the returns to markets are lower, then the advantage of Roth is lower. If the retirement tax bracket is lower then the return to Roth is lower.
Even if I assume there are no further changes in tax rates from current law, my retirement tax bracket will depend on market returns, which I cannot predict.
If I assume there will be changes in tax laws, but I don't know what they will be, then it is impossible to factor those in. I have to ignore them because I don't know what they would be. I have to keep in mind how sensitive the comparison is to these unknowns.

If I go traditional then I get my benefit up front. Money in the pocket. Those could be taken away by future changes in the tax code but I have them now. If I pay the tax up front and hope to get the benefits later, then I am betting that they will still be there after who knows how many revisions to the tax code between now and retirement.
This set of uncertainties favors traditional.

One factor not much discussed is death taxes. Federal estate taxes and state inheritance or estate taxes. For people who will be subject to these, the Roth conversion or Roth contributions up front reduce the size of the taxable estate as compared to traditional. That may make Roth a better option for those whose primary concern in maximizing the after tax amounts passed on to heirs, rather than maximizing after tax spending for themselves.

Hard to see that there is a single best answer without considering individual circumstances.
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

randomguy
Posts: 9036
Joined: Wed Sep 17, 2014 9:00 am

Re: Asymmetric risk in the Roth vs Traditional question

Post by randomguy » Sun Jul 05, 2020 3:23 pm

afan wrote:
Sun Jul 05, 2020 2:52 pm

One factor not much discussed is death taxes. Federal estate taxes and state inheritance or estate taxes. For people who will be subject to these, the Roth conversion or Roth contributions up front reduce the size of the taxable estate as compared to traditional. That may make Roth a better option for those whose primary concern in maximizing the after tax amounts passed on to heirs, rather than maximizing after tax spending for themselves.

Hard to see that there is a single best answer without considering individual circumstances.
Federal estate taxes are over 10 million per person. That tends to be a very subset:) Some states have much lower numbers where it could be an issue. For most of us the more relevant part is that your ROTH's don't have RMDs. You can run into cases where it works out much better for you to spend down other assets and then pass the ROTH on to the kids. But it is definitely very situation dependant.

JBTX
Posts: 6338
Joined: Wed Jul 26, 2017 12:46 pm

Re: Asymmetric risk in the Roth vs Traditional question

Post by JBTX » Sun Jul 05, 2020 3:49 pm

KlangFool wrote:
Wed Jul 01, 2020 12:16 pm
whereskyle wrote:
Wed Jul 01, 2020 12:03 pm

If the taxes on the first $30k of income rise to 15% and there is no standard deduction, then I will be in a better position if I contribute to a Roth with funds taxed today at 12%. The only way the OP's approach makes sense for me is if I assume that the tax code will resemble its current state when I retire (hopefully) in 30 years. I do not assume that. Instead, I focus on the fact that the present-day 12% rate on my contributions to the Roth are historically low. We can debate where tax rates will go all day, but in no way do I assume that they will be the same when I retire. I think current rates are ridiculously low and unsustainable. Therefore, I think it is important to question whether OP's advice is based on the tax code staying stable.
whereskyle,

<<I think current rates are ridiculously low and unsustainable. >>

If you spend some time studying any other democratic countries, you would find that in order to raise more tax revenues, the government reduces the income taxes and substitute that with a national consumption/sale tax. This had been the historical trend for the last 10 to 20 years.

Just because the tax would not be the same, it does not mean it would be an increase in income taxes.

<<if I contribute to a Roth with funds taxed today at 12%.>>

Depending on your income level, why would you choose to pay 12% when you can pay negative taxes? Aka, instead of paying taxes, you are paid by IRS due to tax credits.

<<If the taxes on the first $30k of income rise to 15% and there is no standard deduction,>>

Even assuming that is true, Trad 401K may still win. Have you done the actual calculation?

In summary, if you are interested, start your own topic and go over your own numbers.

KlangFool
Your point is valid about potential for consumption taxes vs income taxes. However, look at these top marginal rates by country, and many of them already have consumption taxes.

https://taxfoundation.org/taxing-high-income-2019/

But for most of us top marginal rates are irrelevant. However what can happen is that in lieu of increases in statutory tax rates, there are various phase ins and phase outs that increase your marginal rates but not your tax bracket or statutory rate.

But I think the point of the OP is a valid one that isn't always mentioned. A traditional retirement account takes advantage of the progressive tax code and provides a form of insurance or hedge if future income unexpectedly drops.

Post Reply