Location does matter when measuring asset allocation

Discuss all general (i.e. non-personal) investing questions and issues, investing news, and theory.
Post Reply
JimInIllinois
Posts: 349
Joined: Wed Dec 22, 2010 9:55 am

Location does matter when measuring asset allocation

Post by JimInIllinois » Thu Jun 19, 2014 11:57 pm

I'm working my way through William Bernstein's Rational Expectations and just got to the "Location, Location, Location" section where he very briefly dismisses the concept that you should adjust tax deferred and taxable account balances downward by their future tax exposure relative to Roth accounts when calculating asset allocation. The footnote points to http://www.efficientfrontier.com/ef/704/where.htm, where he makes a similar argument but offers http://www.efficientfrontier.com/ef/704 ... arkOpt.pdf as a counterpoint. I find Bernstein's argument incorrect, but the Reichenstein article does not actually refute it. I will attempt to make a proper case below.

Consider an investor with $50K of stocks in a Roth IRA and $50K of bonds in a traditional IRA who will be in a 30% tax bracket in retirement. Bernstein argues that the allocation is 50:50 because the stocks can be sold to buy bonds and vice-versa. Reichenstein argues that only 70% of the value of the bonds in the traditional IRA are spendable, so the allocation is 50:(0.7*50) = 59:41. Both arguments fail because they ignore risk (variability of asset returns), the control of which is the purpose of an asset allocation.

Consider a market crash in which stocks drop 50% while bonds increase 10%. A 50:50 portfolio should experience a drop of 20% (the average of -50% and +10%) in value, where the only value that matters is the amount available to spend after taxes.

Our example of $50K stocks in Roth and $50K bonds in traditional originally had an after-tax spendable value of $50K stocks + 0.7 * $50K bonds = $85K, but now has a spendable value of only 0.5 * $50K stocks + 0.7 * 1.1 * $50K bonds = $63.5K, a loss of over 25%, exactly the loss expected for Reichenstein's 59:41 tax-adjusted allocation. Clearly this is not a 50:50 portfolio as the investor has experienced a greater loss than a 50:50 allocation in a single account type would have.

I would further argue that the value of assets in taxable accounts should also be reduced for asset allocation purposes by the expected marginal tax rate as if they had zero basis, since any capital loss can be offset against other gains or ordinary income. The math is a bit tricky, but it works out that $100K in stocks with no gains in a taxable account at a 20% marginal rate is equivalent to $80K in a Roth account plus $20K in a mattress because liquidating the account after a 50% market drop would yield $50K from the stock sale plus a $50K capital loss worth $10K in avoided taxes, for a total value of $60K, which is equal to 0.5 * $80K + $20K.

I hope I've explained this clearly. I love Bernstein's books, but this one issue really bugs me. Questions or comments?

steve_14
Posts: 1507
Joined: Wed Jun 20, 2012 12:05 am

Re: Location does matter when measuring asset allocation

Post by steve_14 » Fri Jun 20, 2014 1:15 am

Hmm battle of the -steins. I think Bernstein's argument is wrong, as you point out. What you have is a 59:41 portfolio - are you saying this isn't correct?

User avatar
market timer
Posts: 5944
Joined: Tue Aug 21, 2007 1:42 am

Re: Location does matter when measuring asset allocation

Post by market timer » Fri Jun 20, 2014 4:09 am

Location matters because we are not indifferent between $1 in a Roth and $1 in a traditional IRA. The Roth dollar is more valuable, and should be weighted as such.

JimInIllinois
Posts: 349
Joined: Wed Dec 22, 2010 9:55 am

Re: Location does matter when measuring asset allocation

Post by JimInIllinois » Fri Jun 20, 2014 7:56 am

steve_14 wrote:Hmm battle of the -steins. I think Bernstein's argument is wrong, as you point out. What you have is a 59:41 portfolio - are you saying this isn't correct?
59:41 is the right answer, but Reichenstein's reasoning doesn't consider fluctuations.

Asset allocation is about controlling the response of the spendable part of your portfolio to fluctuation in the underlying asset classes, so it should be measured with this in mind.

Reichenstein also treats a taxable account with no gains the same as a Roth because you can spend 100% of either. Here I disagree because in a taxable account you don't get to spend 100% of the gains or losses since the government takes a cut of the gains (or refunds a cut of the losses). You do eventually get to spend your entire basis, but only experience (1 - taxrate) of the fluctuations, so the (taxrate) fraction of your basis in a taxable account is best either ignored or treated as a kind of illiquid cash.

stan1
Posts: 5918
Joined: Mon Oct 08, 2007 4:35 pm

Re: Location does matter when measuring asset allocation

Post by stan1 » Fri Jun 20, 2014 8:13 am

Jim, have you read this page on the wiki?
http://www.bogleheads.org/wiki/Tax-adju ... allocation

Johm221122
Posts: 5072
Joined: Fri May 13, 2011 6:27 pm

Re: Location does matter when measuring asset allocation

Post by Johm221122 » Fri Jun 20, 2014 8:18 am

stan1 wrote:Jim, have you read this page on the wiki?
http://www.bogleheads.org/wiki/Tax-adju ... allocation
We also had a poll

http://www.bogleheads.org/forum/viewtop ... 0&t=131310
I'm trying to pay 0%(or close to zero) so I don't

John

User avatar
Boglenaut
Posts: 3090
Joined: Mon Mar 23, 2009 7:41 pm

Re: Location does matter when measuring asset allocation

Post by Boglenaut » Fri Jun 20, 2014 8:31 am

From a theoretical viewpoint it makes sense to account for location; from a practical viewpoint I found it is not so easy.

I tried early on to weigh my allocations but in reality there are too many unknowns... what is my future tax rate, etc.

So, I do not weigh them, but do mentally account that I know the traditional 401k and IRA will be taxed.

User avatar
Doc
Posts: 8618
Joined: Sat Feb 24, 2007 1:10 pm
Location: Two left turns from Larry

Re: Location does matter when measuring asset allocation

Post by Doc » Fri Jun 20, 2014 8:33 am

JimInIllinois wrote:59:41 is the right answer, but Reichenstein's reasoning doesn't consider fluctuations.
Fluctuations don't really matter because we rebalance periodically to control risk and that rebalancing gets rid of the fluctuations for us.

I think that Reichenstein in some cases has defined risk as the risk of not meeting one's retirement goal instead of the more usual metric of standard deviation. You get a somewhat different tax adjustment between the two but either is better than not doing any adjustment at all.

David Grabiner is the Reichenstein guru around here. Maybe he will jump in.
A scientist looks for THE answer to a problem, an engineer looks for AN answer and lawyers ONLY have opinions. Investing is not a science.

JimInIllinois
Posts: 349
Joined: Wed Dec 22, 2010 9:55 am

Re: Location does matter when measuring asset allocation

Post by JimInIllinois » Fri Jun 20, 2014 8:51 am

stan1 wrote:Jim, have you read this page on the wiki?
http://www.bogleheads.org/wiki/Tax-adju ... allocation
Somehow I missed that last night. :oops:

The actual adjustment described in the wiki is 100% correct, including reducing taxable accounts by the marginal tax rate.

However, the top-line summary that "you should base your asset allocation on the after-tax value of the accounts" is misleading for taxable accounts, which are actually reduced below their after-tax value.

User avatar
Doc
Posts: 8618
Joined: Sat Feb 24, 2007 1:10 pm
Location: Two left turns from Larry

Re: Location does matter when measuring asset allocation

Post by Doc » Fri Jun 20, 2014 9:39 am

Wiki wrote:You should value your Roth at its full value, and reduce the value of your traditional IRA or 401(k) by your expected tax bracket at retirement (including state taxes if your state taxes these withdrawals, which varies by state).
That's the Reichenstein approach. I suggest that the CAPM (variance) approach would use the current marginal tax rate. But it doesn't matter enough to make a difference whether you discount your tIRA by 25% or 28%. Just make sure it's not zero unless you really never ever are going to pay any taxes at all on your investments. (Vegas investing excepted.)

Taxable accounts may be expected to have more of a difference between the two approaches. Basically you want to adjust for taxes on unrealized capital gains which could be quite different currently and at "retirement". But again unless your taxable account is very large and you never tax harvest along the way the difference is not likely to affect your AA that much.
A scientist looks for THE answer to a problem, an engineer looks for AN answer and lawyers ONLY have opinions. Investing is not a science.

User avatar
ThePrune
Posts: 897
Joined: Wed Nov 10, 2010 9:38 am
Location: Midland, MI

Re: Location does matter when measuring asset allocation

Post by ThePrune » Fri Jun 20, 2014 2:58 pm

I track my asset allocations on both a before- and an after-tax basis. But my IPS allocation targets (and their rebalance bands) are after-tax.

For a long time I was "sitting on the fence" with respect to this discussion. Then I started building a ladder of TIPS bonds as a form of insurance against potential future inflation spikes. The realization then struck me - my "insurance" was only as good as the after-tax value of my TIPS as they matured! So I took two actions, (1) placed all my TIPS in Roth IRAs, and (2) adjusted my IPS asset allocation targets so they were on an after-tax basis.
Investment skill is often just luck in sheep's clothing.

larryswedroe
Posts: 15711
Joined: Thu Feb 22, 2007 8:28 am
Location: St Louis MO

Re: Location does matter when measuring asset allocation

Post by larryswedroe » Sat Jun 21, 2014 1:08 am

absolutely location matters and it should be accounted for
Larry

letsgobobby
Posts: 11594
Joined: Fri Sep 18, 2009 1:10 am

Re: Location does matter when measuring asset allocation

Post by letsgobobby » Sat Jun 21, 2014 6:23 am

JimInIllinois wrote:
Consider a market crash in which stocks drop 50% while bonds increase 10%. A 50:50 portfolio should experience a drop of 20% (the average of -50% and +10%) in value, where the only value that matters is the amount available to spend after taxes.

Our example of $50K stocks in Roth and $50K bonds in traditional originally had an after-tax spendable value of $50K stocks + 0.7 * $50K bonds = $85K, but now has a spendable value of only 0.5 * $50K stocks + 0.7 * 1.1 * $50K bonds = $63.5K, a loss of over 25%, exactly the loss expected for Reichenstein's 59:41 tax-adjusted allocation. Clearly this is not a 50:50 portfolio as the investor has experienced a greater loss than a 50:50 allocation in a single account type would have.
Some tax adjustment should occur but by how much? Unless you are close to retirement it's hard to predict what your personal tax rates will be 20-40 years in the future, even before taking into account possible changes in tax policy. For instance ending up in the 15% bracket is vastly different than ending up in the 25% bracket. Or ending up with AGI > $250k is quite different than ending up with AGI < $250k.

In these discussions it seems to me there is also a tendency to assume your future tax rates are fixed regardless of what happens with your actual portfolio. Clearly this is not the case although the example above provides the exception that proves the rule more than anything else. Most scenarios are far less black and white, and far less dire, since one or more of the following will occur:

- the investor will hold a mix of stocks and bonds in various account types;
- the investor will have a taxable account;
- the investor will rebalance across account types at various points during the decline;
- the decrease in the investor's pretax accounts will lower RMDs and thus lower the tax bracket;
- the investor does not need to withdraw all the money at a single point in time, meaning there is no 'locking in' of the apparent losses and apparent changes in allocation percentages.
- as money remains in various accounts over a many decades period, values will recover, allocation percentages will shift, and tax brackets will also change.
- very large tax deductions, specifically for health care expenses late in life, are likely to provide substantial opportunity to withdraw money from pretax accounts at very favorable rates; likewise, offer the opportunity to pay 0% LTCG rates on taxable investments.
- charitable giving from IRAs and taxable accounts with appreciated shares will further decrease taxes paid.
- etc.

In my opinion while it is useful to make some adjustments I think 30% adjustments off the pretax accounts are very extreme even if one expects to end up in that tax bracket early in retirement.

Call_Me_Op
Posts: 7029
Joined: Mon Sep 07, 2009 2:57 pm
Location: Milky Way

Re: Location does matter when measuring asset allocation

Post by Call_Me_Op » Sat Jun 21, 2014 6:35 am

This is just one of the reasons that I maintain a separate taxable AA.
Best regards, -Op | | "In the middle of difficulty lies opportunity." Einstein

User avatar
Doc
Posts: 8618
Joined: Sat Feb 24, 2007 1:10 pm
Location: Two left turns from Larry

Re: Location does matter when measuring asset allocation

Post by Doc » Sat Jun 21, 2014 8:11 am

letsgobobby wrote:Some tax adjustment should occur but by how much? Unless you are close to retirement it's hard to predict what your personal tax rates will be 20-40 years in the future, even before taking into account possible changes in tax policy. For instance ending up in the 15% bracket is vastly different than ending up in the 25% bracket. Or ending up with AGI > $250k is quite different than ending up with AGI < $250k
Bobby, it doesn't make a lot of difference. It's not the difference between 15% and 25% that counts but rather the difference between 85% and 75% and that difference is within many people's bands. So just use 80% (20% tax rate) and go for it. As long as you make a reasonable assumption and be consistent you will be OK. Nobody really knows whether a 60/40 AA gives any meaningful different results than a 55/45 on a risk adjusted basis anyway. Do you use age in bonds or age +10 in bonds or maybe age -10 in bonds? What lets you sleep at night may have me walking the halls.

The important thing is to use a reasonable tax rate and zero is not reasonable.

And remember you are probably rebalancing every few years and if your tax assumptions change dramatically you just rebalance a little more or a little less. The process is self correcting.
A scientist looks for THE answer to a problem, an engineer looks for AN answer and lawyers ONLY have opinions. Investing is not a science.

letsgobobby
Posts: 11594
Joined: Fri Sep 18, 2009 1:10 am

Re: Location does matter when measuring asset allocation

Post by letsgobobby » Sat Jun 21, 2014 8:16 am

I admit I've always been flummoxed by this entire issue and do not tax adjust.

I can see this mattering for someone in or near retirement, when a market decline could lead to a larger than expected loss in spendable income. For someone who is accumulating, I don't see that it makes much difference.

Right now I am 60/40. I own some I bonds, otherwise all my fixed income are in pretax accounts. My Roths and taxable (outside of I bonds) are stocks. How does tax-adjusting help me?

User avatar
Doc
Posts: 8618
Joined: Sat Feb 24, 2007 1:10 pm
Location: Two left turns from Larry

Re: Location does matter when measuring asset allocation

Post by Doc » Sat Jun 21, 2014 8:32 am

letsgobobby wrote:I can see this mattering for someone in or near retirement, when a market decline could lead to a larger than expected loss in spendable income. For someone who is accumulating, I don't see that it makes much difference
You are looking at the future value question. Nobody knows with any kind of precision what any AA is going to give you 30 years in the future. All you can do now is make sure that your risk profile is what you intend it to be. If you have only bonds in your tIRA/401k account and don't tax adjust your portfolio is riskier than you think. And as your ratio of taxable to tax advantaged accounts changes over time your risk changes in some unknown way if your don't tax adjust.

You also don't need to be doing a lot of calculations. If tax adjustment means your desired 70/30 AA is actually 75/25 just adjust your unadjusted target down by 5% to 65/35 and use that. Repeat the calculation every five years or so or whenever your inputs change drastically.
A scientist looks for THE answer to a problem, an engineer looks for AN answer and lawyers ONLY have opinions. Investing is not a science.

letsgobobby
Posts: 11594
Joined: Fri Sep 18, 2009 1:10 am

Re: Location does matter when measuring asset allocation

Post by letsgobobby » Sat Jun 21, 2014 9:05 am

Doc wrote:
letsgobobby wrote:I can see this mattering for someone in or near retirement, when a market decline could lead to a larger than expected loss in spendable income. For someone who is accumulating, I don't see that it makes much difference
You are looking at the future value question. Nobody knows with any kind of precision what any AA is going to give you 30 years in the future. All you can do now is make sure that your risk profile is what you intend it to be. If you have only bonds in your tIRA/401k account and don't tax adjust your portfolio is riskier than you think. And as your ratio of taxable to tax advantaged accounts changes over time your risk changes in some unknown way if your don't tax adjust.

You also don't need to be doing a lot of calculations. If tax adjustment means your desired 70/30 AA is actually 75/25 just adjust your unadjusted target down by 5% to 65/35 and use that. Repeat the calculation every five years or so or whenever your inputs change drastically.
However as you pointed out above there is no precision in asset allocation ratios to begin with. So say you are right and my "risk profile" is actually 65/35 instead of 60/40. What does that mean? What impact does it have? How will it affect me either practically or psychologically? What would I do differently other than sell a few stocks and even then, what difference does that make given the inherent imprecision in the system?

User avatar
Doc
Posts: 8618
Joined: Sat Feb 24, 2007 1:10 pm
Location: Two left turns from Larry

Re: Location does matter when measuring asset allocation

Post by Doc » Sat Jun 21, 2014 9:19 am

letsgobobby wrote:However as you pointed out above there is no precision in asset allocation ratios to begin with. So say you are right and my "risk profile" is actually 65/35 instead of 60/40. What does that mean? What impact does it have? How will it affect me either practically or psychologically? What would I do differently other than sell a few stocks and even then, what difference does that make given the inherent imprecision in the system?
But what if its 90/10. Would you like to know? Would you do something differently? I know I would if it was off by that much.

You decided on your AA based on some data or opinions you saw somewhere. Undoubtedly whatever you saw was not tax adjusted. So you are not doing what you wanted to do. Maybe it's just off by a little. Fine then do nothing if you choose. But playing ostrich isn't useful.
A scientist looks for THE answer to a problem, an engineer looks for AN answer and lawyers ONLY have opinions. Investing is not a science.

letsgobobby
Posts: 11594
Joined: Fri Sep 18, 2009 1:10 am

Re: Location does matter when measuring asset allocation

Post by letsgobobby » Sat Jun 21, 2014 9:27 am

but 90/10 would only be relevant if it were 90% taxable, 10% tax-deferred. The more extreme the asset allocation, the less likely the extreme scenario in which 100% of stocks are in taxable and 100% of fixed income are in pretax simply because most people have most of their assets in tax-sheltered accounts.

And again, I can agree in principle that there should be some adjustment but no one can say with any precision what my adjustment should be, and I am very sure it is much closer to 0% than 30%. It's probably something like 5% on the low end and 10% max over my lifetime, considering all possibilities around early retirement, leaving unused assets to heirs, large charitable and medical deductions during retirement, etc. How much difference does a 10% reduction in my fixed income portfolio make in a 60/40 portfolio? Not very much. Essentially a rounding error. In a million dollar portfolio, set up just so to maximize the effects of this whole exercise, we would have $600k of stocks and $360k bonds rather than $600k of stocks and $400k bonds, for an allocation of... 62.5/37.5.

That's a small enough difference to just roll off my back, much more duck than ostrich thank you very much.

cowboysFan
Posts: 290
Joined: Sat Jan 25, 2014 1:46 pm

Re: Location does matter when measuring asset allocation

Post by cowboysFan » Sat Jun 21, 2014 9:52 am

JimInIllinois wrote:Here I disagree because in a taxable account you don't get to spend 100% of the gains or losses since the government takes a cut of the gains (or refunds a cut of the losses). You do eventually get to spend your entire basis, but only experience (1 - taxrate) of the fluctuations, so the (taxrate) fraction of your basis in a taxable account is best either ignored or treated as a kind of illiquid cash.
First, I think most people have unrealized capital gains. Second, this is really a question of how much is TLH worth and how much it helps you. It's probably very useful in some situations and less so in others. Here's a counter-example: a retired couple has SPIAs/TIPS/pension/SS to cover all their living expenses and $2M in stocks that they expect to go to their heirs. Suppose, the market drops 50% and they harvest $1M in capital losses. Since they don't need to sell their stocks and their heirs get the step up in basis anyway, they never realize capital gains to use up their capital losses. They do get the $3k/year deduction against ordinary income, but even if they live to be 100, they still won't come close to using up the $1M in capital losses. The $3k/year isn't indexed to inflation, so getting a $3k deduction in 30 years is worth a lot less than $3k today. It's not as if the government just refunds them their tax rate*their losses.

User avatar
market timer
Posts: 5944
Joined: Tue Aug 21, 2007 1:42 am

Re: Location does matter when measuring asset allocation

Post by market timer » Sun Jun 22, 2014 9:22 am

letsgobobby wrote:And again, I can agree in principle that there should be some adjustment but no one can say with any precision what my adjustment should be, and I am very sure it is much closer to 0% than 30%. It's probably something like 5% on the low end and 10% max over my lifetime, considering all possibilities around early retirement, leaving unused assets to heirs, large charitable and medical deductions during retirement, etc. How much difference does a 10% reduction in my fixed income portfolio make in a 60/40 portfolio? Not very much. Essentially a rounding error. In a million dollar portfolio, set up just so to maximize the effects of this whole exercise, we would have $600k of stocks and $360k bonds rather than $600k of stocks and $400k bonds, for an allocation of... 62.5/37.5.
Are you adjusting by expected marginal tax rate during IRA distribution phase? My guess is the median retired Boglehead is above the 25% marginal tax rate--probably 25% Federal plus some state income tax. While the average tax rate on IRA distributions may be lower than 25%, one should adjust by the expected marginal tax rate for asset allocation purposes.

BillyG
Posts: 398
Joined: Sat Nov 17, 2012 9:02 pm
Location: Maryland, USA

Re: Location does matter when measuring asset allocation

Post by BillyG » Sun Jun 22, 2014 9:42 am

When I saw the thread title I thought "location" in terms of where you live when you retire... which affects the state tax rate on certain things like IRA withdrawals. This adds another variable and another unknown in terms of future tax laws, and for some who have not decided where they are going to live or whose plans may change due to death, divorce, or wanting to be near children who may move in their jobs.

Billy

stlutz
Posts: 4752
Joined: Fri Jan 02, 2009 1:08 am

Re: Location does matter when measuring asset allocation

Post by stlutz » Sun Jun 22, 2014 10:07 am

This discussion shows why the oft repeated refrain, "It's easier to think of your assets as a single portfolio" is wrong. If you replicate your asset allocation (at least at the stock/bond level) within each of your accounts (401K, IRA, taxable), you can add this to the list of questions you no longer need to worry about.

letsgobobby
Posts: 11594
Joined: Fri Sep 18, 2009 1:10 am

Re: Location does matter when measuring asset allocation

Post by letsgobobby » Sun Jun 22, 2014 10:54 am

market timer wrote:
letsgobobby wrote:And again, I can agree in principle that there should be some adjustment but no one can say with any precision what my adjustment should be, and I am very sure it is much closer to 0% than 30%. It's probably something like 5% on the low end and 10% max over my lifetime, considering all possibilities around early retirement, leaving unused assets to heirs, large charitable and medical deductions during retirement, etc. How much difference does a 10% reduction in my fixed income portfolio make in a 60/40 portfolio? Not very much. Essentially a rounding error. In a million dollar portfolio, set up just so to maximize the effects of this whole exercise, we would have $600k of stocks and $360k bonds rather than $600k of stocks and $400k bonds, for an allocation of... 62.5/37.5.
Are you adjusting by expected marginal tax rate during IRA distribution phase? My guess is the median retired Boglehead is above the 25% marginal tax rate--probably 25% Federal plus some state income tax. While the average tax rate on IRA distributions may be lower than 25%, one should adjust by the expected marginal tax rate for asset allocation purposes.
The 25% federal bracket requires approximately $72k of taxable income for a married couple, so including even the standard deduction and 2 senior exemptions and charitable donations and excluding tax-free municipal income, we're talking about a couple with easily over $100,000 of actual income. I doubt even most Bogleheads are in that situation- I may be from time to time but do not expect to be most of retirement. (I also live in a state with 0% income tax, and plan to stay here.). I haven't even considered the possibility of early retirement, carried forward tax losses to use against realized gains, etc.

It would be very interesting to see what marginal tax rate the average retired Boglehead pays. I'm guessing there are many in the 15% or below. (excluding state taxes).

User avatar
LadyGeek
Site Admin
Posts: 48097
Joined: Sat Dec 20, 2008 5:34 pm
Location: Philadelphia
Contact:

Re: Location does matter when measuring asset allocation

Post by LadyGeek » Sun Jun 22, 2014 10:59 am

We've extensively discussed this before. In summary, see the two banners at the top of Principles of tax-efficient fund placement:
Image Determination of your asset allocation (% stocks / % bonds), which sets your portfolio's level of acceptable risk, is the single most influential decision you can make on your portfolio's performance. Only consider taxes after you have configured your total portfolio.

Image Tax regulations can be complex and contain subtle details that may escape inexperienced investors. If this article seems overly complicated, then just remember a few key points:
  • Set your asset allocation first, taxes come second. If you don't have any funds which can be put in a location to reduce your tax bill, then stop here. You've done the best you can.
  • Tax rates and brackets change frequently. What was a logical tax location one year may turn out to be a poor choice a few years later. Consider if it's worth the effort (added complexity) to take this approach.
Wiki To some, the glass is half full. To others, the glass is half empty. To an engineer, it's twice the size it needs to be.

columbia
Posts: 913
Joined: Tue Aug 27, 2013 5:30 am

Re: Location does matter when measuring asset allocation

Post by columbia » Sun Jun 22, 2014 11:38 am

stlutz wrote:This discussion shows why the oft repeated refrain, "It's easier to think of your assets as a single portfolio" is wrong. If you replicate your asset allocation (at least at the stock/bond level) within each of your accounts (401K, IRA, taxable), you can add this to the list of questions you no longer need to worry about.
Yes and I learned this idea from posters like yourself.
It's the easiest approach and will give me one less thing to worry about during retirement.

User avatar
LadyGeek
Site Admin
Posts: 48097
Joined: Sat Dec 20, 2008 5:34 pm
Location: Philadelphia
Contact:

Re: Location does matter when measuring asset allocation

Post by LadyGeek » Sun Jun 22, 2014 11:48 am

Here's the wiki article which discusses this in-depth: Tax-adjusted asset allocation
Wiki To some, the glass is half full. To others, the glass is half empty. To an engineer, it's twice the size it needs to be.

User avatar
ogd
Posts: 4854
Joined: Thu Jun 14, 2012 11:43 pm

Re: Location does matter when measuring asset allocation

Post by ogd » Sun Jun 22, 2014 11:58 am

columbia wrote:
stlutz wrote:This discussion shows why the oft repeated refrain, "It's easier to think of your assets as a single portfolio" is wrong. If you replicate your asset allocation (at least at the stock/bond level) within each of your accounts (401K, IRA, taxable), you can add this to the list of questions you no longer need to worry about.
Yes and I learned this idea from posters like yourself.
It's the easiest approach and will give me one less thing to worry about during retirement.
Well, you always have to weigh the allocation precision against the actual tax savings, i.e. money in hand.

That said, I think this style of allocation costs surprisingly little at the present low yields, when we take all effects into account; the faster growth of tax-advantaged space, for example. It's only in the highest tax brackets that it might be too much. And it eliminates an obstacle for using all-in-one funds (like LifeStrategy), which I think would serve many or most investors very well.

User avatar
thursdaysd
Posts: 66
Joined: Tue May 13, 2014 10:55 am
Location: NC
Contact:

Re: Location does matter when measuring asset allocation

Post by thursdaysd » Sun Jun 22, 2014 12:00 pm

The 25% federal bracket requires approximately $72k of taxable income for a married couple, so including even the standard deduction and 2 senior exemptions and charitable donations and excluding tax-free municipal income, we're talking about a couple with easily over $100,000 of actual income. I doubt even most Bogleheads are in that situation
Not all posters here are married! For single filers the 25% rate starts at $36,250, I submit that many single posters would be over that, I certainly am. And once I take full SS at 70 and start RMD at 70 1/2, in addition to my pension, I would not be at all surprised to find myself at the 28% marginal rate. :(
Thursday's child has far to go

User avatar
Doc
Posts: 8618
Joined: Sat Feb 24, 2007 1:10 pm
Location: Two left turns from Larry

Re: Location does matter when measuring asset allocation

Post by Doc » Sun Jun 22, 2014 12:50 pm

thursdaysd wrote: Not all posters here are married! For single filers the 25% rate starts at $36,250, I submit that many single posters would be over that, I certainly am. And once I take full SS at 70 and start RMD at 70 1/2, in addition to my pension, I would not be at all surprised to find myself at the 28% marginal rate. :(
I have not looked at the numbers for your situation but remember that the SS phaseout can increase your marginal rate by as much as 85% of the statutory rate if you are (un)fortunate enough to be in the phaseout band. I know that it can incur in the 15% bracket and Grabiner has mentioned that it can also occur in the 25% bracket IIRC. In either case marginal rates in the high 20's are not unusual if you are taking SS.
A scientist looks for THE answer to a problem, an engineer looks for AN answer and lawyers ONLY have opinions. Investing is not a science.

User avatar
Doc
Posts: 8618
Joined: Sat Feb 24, 2007 1:10 pm
Location: Two left turns from Larry

Re: Location does matter when measuring asset allocation

Post by Doc » Sun Jun 22, 2014 1:14 pm

LadyGeek wrote:Here's the wiki article which discusses this in-depth: Tax-adjusted asset allocation
Excellent point.

I think there is often some confusion between asset allocation and asset location. They are not the same and not really related.

Asset allocation in it's basic form is your ratio of stocks to bonds. It is done to control the risk of your overall portfolio. It has nothing to do with returns. (Other than the obvious that if you want lower risk you should expect lower returns.)

Asset location is all about taxes. Once you have decided on an asset allocation you then have to decide which assets you put in tax advantaged to get the most tax efficient location. Ideally you want to locate without "screwing up" your allocation.

The complication is that the taxes affect the amount of each account that is actually yours and that balance should not include accrued taxes which belongs not to you but to the guv in the form of accrued or deferred taxes. If you have both taxable and tax advantaged accounts you have to make a correction to the balances so that only what is actually yours figures into the asset allocation.

Whether you use current marginal rates or future marginal rates to make your tax correction is subject to debate but it is very seldom zero and you should not ignore the ideas of asset allocation or asset location simply because there is some uncertainty in your tax estimate.
A scientist looks for THE answer to a problem, an engineer looks for AN answer and lawyers ONLY have opinions. Investing is not a science.

User avatar
Christine_NM
Posts: 2635
Joined: Tue Feb 20, 2007 1:13 am
Location: New Mexico

Re: Location does matter when measuring asset allocation

Post by Christine_NM » Sun Jun 22, 2014 1:46 pm

Until this year I would have agreed that a traditional IRA should be discounted for taxes due.

However, now that I have taken my first RMD, I see in practice that Bernstein is correct, money is completely fungible. 60% of my RMD paid my federal income tax due for this year, 15% paid state tax. The remainder went into my Vanguard taxable account. The RMD had been in Prime money market for 2 years so did not affect my asset allocation.

This is manna from heaven. No longer do I have to spend either divendends, savings, or pension money on income tax. I had rather dreaded the RMD for the increase in taxes, but if you have left an IRA alone till age 70 1/2, then you suddenly have a big pool of money to cover your biggest expense, income taxes.

In real life Dr. B. is correct to point out that you should care only about your total tax liability.
17% cash 47% stock 36% bond. Retired, w/d rate 2.85%

pkcrafter
Posts: 13078
Joined: Sun Mar 04, 2007 12:19 pm
Location: CA
Contact:

Re: Location does matter when measuring asset allocation

Post by pkcrafter » Sun Jun 22, 2014 2:07 pm

Christine wrote:

Code: Select all

Until this year I would have agreed that a traditional IRA should be discounted for taxes due
This raises another point. When comparing TIRA to Roth, you have to start before taxes with the Roth contributions. $10,000 to invest, TIRA $10,000 invested. Roth $7500 invested, $2,500 paid in tax in 25% bracket. If the contributions are identical and taxes the same, after X years the TIRA will have 25% more in it. Withdraw the funds from each account and you end up with the same amount of money from each.

Paul
When times are good, investors tend to forget about risk and focus on opportunity. When times are bad, investors tend to forget about opportunity and focus on risk.

User avatar
Christine_NM
Posts: 2635
Joined: Tue Feb 20, 2007 1:13 am
Location: New Mexico

Re: Location does matter when measuring asset allocation

Post by Christine_NM » Sun Jun 22, 2014 2:42 pm

Thanks for raising that point, Paul. $1 in a Roth is more valuable (as stated earlier) only because it is post-tax.
17% cash 47% stock 36% bond. Retired, w/d rate 2.85%

sambb
Posts: 2145
Joined: Sun Mar 10, 2013 3:31 pm

Re: Location does matter when measuring asset allocation

Post by sambb » Sun Jun 22, 2014 8:23 pm

Taxes do matter, and that affects one's allocation. I am constantly somewhat surprised that people invest in "bonds in tax deferred", as that mantra is commonly held here. It all depends. Im in my 40s. I have no idea what income tax rates and capital gains tax rates will be in 20 years. So, i don't understand how people can predict that stock funds in taxable side is better than bonds. Then, of course, in higher tax brackets, munis are attractive. There are too many variables. I still don't understand how people have accurately predicted the future capital gains tax rate.
I think it is a perfectly reasonable idea to have an equal allocation in taxable and in tax deferred accounts (401), and in tax exempt accounts (roths)

There was a long thread previously on this with a link to an academic paper, and it was a good thread i thought. Of course opinions vary, and everyone has to do what they feel fit to do. For estate planning, i like the idea of roths growing more than the others, as there can be advantages to one's heirs as well.

User avatar
Kevin M
Posts: 10110
Joined: Mon Jun 29, 2009 3:24 pm
Contact:

Re: Location does matter when measuring asset allocation

Post by Kevin M » Sun Jun 22, 2014 9:35 pm

stlutz wrote:This discussion shows why the oft repeated refrain, "It's easier to think of your assets as a single portfolio" is wrong. If you replicate your asset allocation (at least at the stock/bond level) within each of your accounts (401K, IRA, taxable), you can add this to the list of questions you no longer need to worry about.
It may not be easier to think of your investments as a single portfolio, but it could be more effective. A big advantage for some people is that they may have a decent fund in their 401k/403b plan for one asset class, but not for another. The most typical example is access to an S&P 500 index fund in the 401k/403b for exposure to large-cap US blend stocks, but all other funds in the 401k/403b are high-cost, actively-managed funds. By viewing your investments as a single portfolio, you can use better funds in IRA and taxable accounts for the other asset classes.

I too don't really get Bill's argument on this one, and from a theoretical perspective think adjusting AA for taxes makes sense. However I have never bothered to do so, other than realizing that my portfolio might be somewhat more or less risky than the non-tax adjusted AA indicates.

Kevin
Wiki ||.......|| Suggested format for Asking Portfolio Questions (edit original post)

User avatar
Doc
Posts: 8618
Joined: Sat Feb 24, 2007 1:10 pm
Location: Two left turns from Larry

Re: Location does matter when measuring asset allocation

Post by Doc » Mon Jun 23, 2014 8:46 am

sambb wrote:I have no idea what income tax rates and capital gains tax rates will be in 20 years. So, i don't understand how people can predict that stock funds in taxable side is better than bonds.
1) Since you rebalance every few years you don't need to know what happens in 20 years. You need only to know what is happening now. The process is self correcting.

2) Short term bonds are more tax efficient than an S&P 500 fund now for many people. So a lot of those people that are making that "better than" prediction are wrong now. And if interest rates go up in 10 or 20 years you simply sell bonds and buy stocks in taxable. See 1).

Caveat: Unfortunately many people have a lot of equities in taxable with large unrealized gains so switching equities into tax advantaged is not as easy as the reverse. Beware of rules of thumb that only really apply to some other guy in a different time and place.
A scientist looks for THE answer to a problem, an engineer looks for AN answer and lawyers ONLY have opinions. Investing is not a science.

letsgobobby
Posts: 11594
Joined: Fri Sep 18, 2009 1:10 am

Re: Location does matter when measuring asset allocation

Post by letsgobobby » Mon Jun 23, 2014 12:51 pm

But the paradox of 'equal location' is this: the typical rebalancing scenario, which involves selling stocks to buy bonds, means selling and realizing capital gains. So while all-stock-in-taxable folks may have lots of unrealized capital gains making it hard to switch back to an equal location approach, that is only because equal location folks have already been paying taxes on those realized gains along the way.

columbia
Posts: 913
Joined: Tue Aug 27, 2013 5:30 am

Re: Location does matter when measuring asset allocation

Post by columbia » Mon Jun 23, 2014 6:54 pm

letsgobobby wrote:But the paradox of 'equal location' is this: the typical rebalancing scenario, which involves selling stocks to buy bonds, means selling and realizing capital gains. So while all-stock-in-taxable folks may have lots of unrealized capital gains making it hard to switch back to an equal location approach, that is only because equal location folks have already been paying taxes on those realized gains along the way.
Yes, there's no perfect solution, but through reading this forum, I found the one for me:

My taxable account is - and always will be a - small fraction of my overall portfolio. I'm not a high income earner and there's no free lunch coming 'round the bend to bolster it.
I deposit what I can and when I can. I even use the (seemingly) heretical route of VASGX (Life Strategy Growth), because it's easy and negligible tax burden on the bond side. I'll eventually reduce my stock holdings in my IRA and won't really worry about the taxable account allocation. What sometimes gets lost here is that some investors are doing everything they can save up enough to not run out of money in retirement....that's goal, plain and simple.
Fussing over grand strategies across five accounts, including municipal bonds to avoid hight tax rates isn't in the cards for some.

User avatar
fourwedge
Posts: 319
Joined: Sun Sep 22, 2013 10:10 pm

Re: Location does matter when measuring asset allocation

Post by fourwedge » Mon Jun 23, 2014 7:28 pm

Great, now you are telling me I need to adjust my allocation based on where it's held..... Can't we over think things here guys? Not gonna participate in this one.
Max out your tax sheltered retirement accounts with inexpensive, well diversified, index funds and you will beat 90% of all investors.

dbr
Posts: 27207
Joined: Sun Mar 04, 2007 9:50 am

Re: Location does matter when measuring asset allocation

Post by dbr » Tue Jun 24, 2014 8:18 am

fourwedge wrote:Great, now you are telling me I need to adjust my allocation based on where it's held..... Can't we over think things here guys? Not gonna participate in this one.
+1

Bob
Posts: 192
Joined: Mon Feb 26, 2007 5:15 pm

Re: Location does matter when measuring asset allocation

Post by Bob » Tue Jun 24, 2014 8:53 am

To calculate the location-adjusted AA, what tax rate would one use for a taxable IRA from which one must take RMDs in a couple years? I have figured out what my taxes will likely be based on total income (RMDs, Social Security, Dividends, Interest, etc.) and total tax rate (Federal plus State) are "X" while incremental tax rate is much higher at "Y". Which would be the correct one to apply? Or is it another number that should be used?

User avatar
Doc
Posts: 8618
Joined: Sat Feb 24, 2007 1:10 pm
Location: Two left turns from Larry

Re: Location does matter when measuring asset allocation

Post by Doc » Tue Jun 24, 2014 9:21 am

Bob wrote:To calculate the location-adjusted AA, what tax rate would one use for a taxable IRA from which one must take RMDs in a couple years?
It doesn't matter enough to worry about. If you don't know if it's going to be 15% or 25% pick 20%. You then multiply the account balance by 0.80. Every five year or so do the calculation again and if you then think the tax rate is going to be a little higher use maybe 0.75 as the multiplier. Then do your regular rebalancing.

If you are like columbia "My taxable account is - and always will be a - small fraction of my overall portfolio" it won't matter one iota because all your bonds and all your stocks are going to be multiplied by the same number and the number itself will have no effect on your AA at all.

You guys are worried about the lack of precision in your calculation. Think like an engineer not a scientist. If the bridge doesn't fall down its close enough. Investing is not a science.
A scientist looks for THE answer to a problem, an engineer looks for AN answer and lawyers ONLY have opinions. Investing is not a science.

dbr
Posts: 27207
Joined: Sun Mar 04, 2007 9:50 am

Re: Location does matter when measuring asset allocation

Post by dbr » Tue Jun 24, 2014 9:24 am

Doc wrote: You guys are worried about the lack of precision in your calculation. Think like an engineer not a scientist. If the bridge doesn't fall down its close enough. Investing is not a science.
The bridge will not fall down due to failing to tax adjust your asset allocation.

This is not the same thing as imagining that you will pay no taxes.

User avatar
market timer
Posts: 5944
Joined: Tue Aug 21, 2007 1:42 am

Re: Location does matter when measuring asset allocation

Post by market timer » Tue Jun 24, 2014 9:38 am

Bob wrote:To calculate the location-adjusted AA, what tax rate would one use for a taxable IRA from which one must take RMDs in a couple years? I have figured out what my taxes will likely be based on total income (RMDs, Social Security, Dividends, Interest, etc.) and total tax rate (Federal plus State) are "X" while incremental tax rate is much higher at "Y". Which would be the correct one to apply? Or is it another number that should be used?
I use the higher marginal tax rate to discount my IRA relative to my Roth, not the average tax rate. I think the key question to ask yourself is this: What tax rate would you pay today to move $1 from your IRA to your Roth? You should be willing to pay something close to your expected marginal tax rate in the future to transfer $1 from your IRA to your Roth. Note that this number is not static. If your wealth begins to exceed your expectations, that expected marginal tax rate should rise.

Post Reply