Asset Location: Do Bonds Really Belong in Retirement Accounts? – SWR Series Part 35

Welcome back to another installment of the Safe Withdrawal Rate Series. This one is about taxes. Amazing, how after 30+ installments in the series, I have written conspicuously little about taxes. Sure, I’ve done some Case Studies where, among many other issues, I delved into the tax planning, most recently in the Case Study for Becky and Stephen. But I’ve never written much about taxes and tax planning in the context of the Series.

There are two reasons why I kept the tax discussion on such a low burner: First, my background: If I had an accounting Ph.D. and CPA instead of an economics Ph.D. and a CFA charter, I would have written a whole lot more about taxes! Second, pinning down the Safe Withdrawal strategy and the safe withdrawal rate is my main concern. Most (early) retirees will have extremely low tax liabilities as I outlined in a post last year. You’d have to try pretty hard to pay more than a 5% federal effective tax rate in retirement. So, as long as you stay away from anything clearly irresponsible on the tax planning side, you’re fine. Don’t stress out over taxes in retirement unless you have a really, really large nest egg and taxable income deep into the six-figures during retirement.

But you don’t want to leave any money on the table either. So, I still want to write about taxes if I encounter something that captures my attention. And I came across a topic that’s most definitely interesting from a withdrawal strategy perspective: Asset Location (as opposed to Asset Allocation).

Imagine you target a particular asset allocation, say 60% stocks and 40% bonds. Or 70/30, or 80/20, or whatever suits your needs the best. How should we allocate that across the different account types? If we put all the different accounts into three major buckets…

  1. Taxable, i.e., your standard taxable brokerage account: Interest, dividends and realized capital gains are taxable every year they show up on your 1099 tax form. But you don’t have to pay taxes on capital gains until you realize them.
  2. Tax-deferred, i.e., your 401(k) or your Traditional IRA. Your account grows tax-free until you actually withdraw the money (or roll it over to a Roth). So, so can realize as much in interest income, dividends, capital gains along the way, as long as you keep the money inside the account.
  3. Tax-free, i.e., your Roth IRA or your HSA. The money grows tax-free and you can withdraw tax-free as well.

… then where do we put our bonds and where do we put our stocks? It would be easy, though likely not optimal to simply keep that same asset allocation in all three types of accounts. But is there a better way to allocate your stock vs. bond allocation?

Sure, there is! One of the oldest pieces of “conventional wisdom” investment advice I can remember is this:

“Keep stocks in a taxable account and bonds in your tax-advantaged accounts.”

Or more generally:

“Keep the relatively tax-efficient investments in a taxable account and relatively tax-inefficient investments in a tax-advantaged account.”

Most stocks would be considered more tax-efficient than bonds because a) dividends and capital gains are taxed at a lower rate than interest income and b) you can defer capital gains until you actually withdraw your money, which is a huge tax-advantage (more on that later).

So, it appears that we should ideally load up the taxable account with stocks and the tax-advantaged accounts with bonds. Hmmm, but that doesn’t sound quite right, does it? Why would I want to “waste” the limited shelf space I have in my tax-advantaged accounts with low-return bonds while I expose my high-return stocks to dividend and capital gains taxes? So, it would be completely rational to be skeptical about this common-sense advice!

So who’s right? Conventional wisdom or the skeptics? Long story short: they’re both wrong! You can easily construct examples where either conventional wisdom or the skeptics prevail. So neither side should claim that their recommendation is universally applicable. The asset location decision depends on…

  1. Your expected rates of return,
  2. Your expected tax rates,
  3. Your investment horizon. Yup, you heard that right, it’s possible that you want to go either one way or the other depending on the horizon. Though, this is not really a separate case but really only a result of asset allocation drift. Accounting for that, we’re back to the two cases, but more on that later!

Let’s look at the details…

Before we even jump into weeds here, let’s just cover a few Tax 101 and Tax 102 topics:

Tax 101: Tax-efficient vs tax-inefficient assets

Tax-efficient assets:

  • Stocks, Mutual Funds, ETFs (especially with low dividend yield). Ideally, index funds with very little turnover and zero or very small capital gains distributions
  • Municipal Bonds and mutual funds and ETFs investing in them
  • Real Estate (most of the time)
  • Section 1256 contracts: options, futures, options on futures. (caveat: they are still tax-inefficient in absolute terms but tax-efficient relative to non-1256 derivatives contracts)

Tax-inefficient assets:

  • REITs, because their dividends are taxed as ordinary income
  • Taxable bonds, because interest is taxed at the personal income tax rate
  • Stocks with a high dividend yield (incl., high-dividend yield stock ETFs, replicating, for example, the Dividend Aristocrats Strategy, etc.). This is mostly an inefficiency if you plan to reinvest the dividends in the accumulation phase and less of a problem in the withdrawal phase when you intend to live off your dividends.
  • Mutual funds and ETFs that are actively managed, creating a lot of turnover and taxable capital gains along the way. Too much turnover and short holding periods may also create a higher portion of your dividend income falling into the non-qualified category.
  • Target Date Funds, even if they are based on index funds (e.g., the Fidelity Freedom Index or Vanguard class of TDFs). Not only will they hold tax-inefficient asset classes (bonds, sometimes REITs) but the target weights of the assets change over the years (by definition) and even in the short-term, the funds are regularly rebalanced back to their target weights. Turnover creates taxable events and tax-inefficiency.
  • Single-stock options trading strategies (covered-calls, naked puts, short straddles/strangles, Iron Condor, etc.)

And to mention it again, notice that efficiency vs. inefficiency is never a 0/1 thing. This is about relative efficiency. Again, the prime example is that in absolute terms, Section 1256 contracts are still tax-inefficient. But they are more efficient than the single stock options. Also, notice that there’s some overlap: Some Target Date Funds also use actively-managed funds (Fidelity Freedom, i.e., without the “Index” in the name), so please don’t interpret this list as mutually exclusive (nor exhaustive!!!).

Tax 102: The power of deferring capital gains!

Stocks are a (relatively) tax-efficient asset class because dividends and capital gains are taxed at lower rates than interest income. But stocks also have an additional, equally important benefit. You pay taxes on the capital gains only when you realize them. (A caveat: there are now some proposals in Washington, D.C. to change that and force you to pay taxes on gains every year. Let’s hope that this nonsense never passes!)

So, this is not exactly tax-avoidance, but tax timing. Paying capital gains taxes only once in the end rather than over time has a crucial benefit. Let’s go through a really simple example to illustrate this. Imagine you have an initial investment and you achieve a 50% return in year 1 and another 50% return in year 2. Yeah, yeah, I know that seems excessive but I like to show this tax advantage through a visual aid and in order to see this better in the diagrams, let’s take large round numbers!

Also, assume that the tax rate on capital gains is 50% (a large round number again for the same reason). So, deferring capital gains works just like in the diagram below:

  • You first grow your portfolio by 50% in year 1, so you now own this rectangle comprised of the blue square plus the green rectangle on the right.
  • The year 2 returns are added along the vertical axis. Again, you grow your rectangle by 50%, going up.
  • After year 2, you pay the capital gains taxes and you’re left with the blue square plus the two (smaller) green bars on the right. Notice that you don’t owe any money on your cost basis, so you only lose half of the green area to the tax-man (or woman).
How to calculate taxes and after-tax returns if you’re deferring capital gains taxes.

(a side note: If you’re so very troubled by the 50% return you can also interpret this as the first period lasting 5 years with an average compound return of 8.45% for a total of 50% and the second period as the same length again)

How about if we realize capital gains along the way. Let’s look at the diagram below.

  • After year 1, we pay our taxes and then reinvest the balance.
  • In year 2, we add capital gains again by extending the rectangle along the vertical axis.
  • And we pay capital gains taxes on that new gain. But not on the year 1 gain because that’s already after-tax and added to the cost basis!
How to calculate taxes and after-tax returns if you’re realizing capital gains along the way.

At first glance, the after-tax return looks exactly identical to the deferral case. But there is one difference. You lose that one little square marked in the chart above. And that’s because taxes in the first year reduce the amount you have invested in the market. And this loss, this drag from taxes if you get taxed every year along the way will compound over the years. It could make a significant difference over the years!

Compounding is a beautiful thing. Compounding losses due to the tax drag isn’t!

So, the lesson so far: It’s tax-efficient to buy-and-hold stocks in taxable accounts. There is a (small) drag from having to pay taxes on dividends every year but a) the dividend yield is often relatively small compared to capital gains and b) the tax on dividends is also smaller than the ordinary income tax rate.

But notice how a bond portfolio in a taxable account gets hit by this compounding of taxes. That’s probably the main rationale for that conventional wisdom “stocks in taxable, bonds in retirement accounts,” which brings us to the main topic for today…

Asset Location: taxable vs. tax-advantaged accounts

Let’s calculate the pros and cons of the different asset location strategies.  Consider an investor with the following parameters:

  • A taxable account and a Roth IRA, each with a $100 account balance (scale it to any value that you like!)
  • The retiree has a 50% Stock, 50% Bond target overall.
  • In the taxable account, dividends and interest will be taxed every year. Both after-tax-dividends and after-tax interest are reinvested in the same asset class that it came from.
  • If stocks are in the taxable account, we keep track of the cost basis (initial investment plus all the after-tax-dividend reinvestments over time) and we pay capital gains taxes at the end of the horizon.

(Side note: what happens if you have a Traditional IRA or 401k account instead of the Roth IRA? In this simple example with a flat ordinary income tax, the pros and cons for investing in taxable vs. retirement accounts are the same if you simply re-interpret your 401k account valued at $X as a Roth IRA worth $X*(1-T) where T is the tax rate on ordinary income. Without loss of generality, I just assume that the investor has only one taxable and one Roth account.)

To calculate all the pertinent numbers here’s a Google Sheet

—> Link to Asset Location Google Sheet <—

As always: This is a clean copy posted on the web for everyone’s reference. Hence, I cannot give permission to edit this sheet. Please save your own copy if you like to enter your own parameters: Menu: File->Make a copy

Please save your own copy if you like to edit this!

You can enter your own parameters in the orange fields. All other fields are calculated!


Let’s look at the three different cases

Case 1: Conventional wisdom prevails – Stocks belong in the taxable account!

  • Stock expected returns are 8% p.a., of which 2% come from dividends and 6% from capital gains
  • Bond returns are 5% p.a. (I know this seems high, but there are some bond funds with a yield that high! I’ve seen some Preferred Shares with dividend yields around 8%, paid as ordinary income/interest. Also, we may eventually leave the current low-yield environment and return to more reasonable bond yields, even for government and investment-grade corporate bonds).
  • Tax rates are 5% for dividends and capital gains and 17% for ordinary income. This would be an individual who’s in the second federal tax bracket (currently 12%), still within the 0% bracket for long-term capital gains and qualified dividends but also faces a 5% flat tax on the state level.
  • A horizon of 20 years.

Let’s look at the results, in the table below.

  • Loading up on stocks in the taxable account gets you $709.29 after 20 years. That’s a bit more than the $691.62 from investing the Roth in stocks. The after-tax IRR is 6.53% vs. 6.40%. It doesn’t look like a huge difference, but 0.13% (13 basis points) is nothing to sneeze at. Considering how much hoopla people make out of few basis points between this index fund and that index fund…
  • The reason for the outperformance of the “stocks in taxable” is very easy to pin down. Let’s look at the table right under the parameters: The After-Tax Value of $100 account value today. Stocks obviously outperform bonds in either account. But if we look at the ratio S/B, the outperformance is larger in the taxable account (1.9686 vs. 1.7567). Thus, stocks belong in the Taxable account!    (Note: to make this easier to spot the differences, I color-coded the relevant measures in red vs. green to mark the smaller vs. larger number!)
  • We can also look at the opposite ratio: the ratio of investment outcomes Roth divided by Taxable. For stocks, you get a ratio of 1.049854 and for bonds 1.176523. So, by that measure, bonds belong in the Roth. Same result!
  • The line of reasoning in the last two bullet points should sound familiar. If you’ve ever taken microeconomics and/or international economics: this is the concept of comparative advantage, i.e., the Ricardian model of international trade. Actually, not just international trade, but any trade. And here’s an example where one asset has an absolute advantage over the other in both accounts ($443.96>$225.52 and $466.10>$265.33), but stocks have the comparative advantage over bonds in the taxable account, and bonds have a comparative advantage over stocks in the Roth IRA. Sorry, this may sound like jibberish, but as an economist, I couldn’t resist pointing that out! 🙂
  • A similar route of showing the comparative advantage is to look at the after-tax IRRs. In the taxable account, stocks have a roughly 3.6% return advantage but only a 3.0% advantage in the Roth. Likewise, bonds get a 0.85% return boost in the Roth but stocks only 0.26%! Same story here: Stocks belong in the taxable account!
Case 1: Conventional Wisdom prevails: Stocks belong in the taxable account.

Also, I like to quickly confirm that this result is really independent of the time horizon. Let’s look at the chart that plots the time series of the “stocks in Roth” advantage over “stocks in taxable” and no matter what your horizon is whether 1 year or 30, Stocks in the Roth will always be worse than Stocks in the Taxable. And the effect increases over time. (but note an important caveat, please see below)

Over a 30-year horizon, the disadvantage of “Stocks in Roth” gets worse.

So, summary thus far: conventional wisdom prevails in this example. The rationale of this has to do with the Tax 102 lesson above. You want to keep the compounding of taxes low. Because stock dividends are taxed a rate of only 5% you keep the compounding of the tax drag relatively small. Much smaller than a 17% tax on 5% interest every year. Even the fact that capital gains are taxed once at the end at 5% cannot reverse this math.

Case 2: Conventional wisdom fails – Stocks belong in the tax-advantaged account!

And, you guessed it, here’s an example where bonds should be in the taxable account. Let’s assume the same parameters as above, but we lower the bond yield to 2% and raise the tax rates to 15% for dividends and capital gains and 33% for ordinary income.

All the results are reversed now:

  • After 20 years, you’ll get to $596.60 with stocks in the Roth and $549.63 with stocks in the taxable account (all after-tax). That’s a very large impact in the IRR (5.62% vs. 5.18%).
  • And that result is perfectly intuitive because stocks do have a comparative advantage in the Roth (3.136691 vs. 3.072983).
  • And you can also quickly confirm that this result is robust over different time horizons one through 30 years (chart not displayed here in the post).
Conventional wisdom fails! Stocks belong in the Roth!

So, what’s going on here. It’s pretty obvious, the bond yield is so low that even with the significant tax rate (33%) the drag on the bond portfolio in the taxable account is relatively low (0.66%). You’d actually have more tax drag from stocks in the taxable portfolio (0.8089%) than from bonds.

So, hands-down here’s an example where conventional wisdom fails you!

Case 3: Depends on the horizon!

Next, let’s look at the one really peculiar case. Let’s take the same parameters as above in case 2 but make two changes: 1) increase the bond yield to 5% again and raise the investment horizon to 30 years. The results are in the screenshot below. And they are really confusing!!!

  • The comparative advantage calculations point to placing stocks in the taxable portfolio: And it’s not even close! 3.085604 vs. 2.328272 in the S/B ratio for the taxable vs. Roth IRA, respectively.
  • And yet, after 30 years, putting stocks in the Roth slightly outperforms the Stocks in taxable case!
What’s going on here? The comparative advantage calculations give you the wrong advice! After 30 years, you’re better off with stocks in the Roth!!!

If that’s not confusing yet, let’s look at the time series chart for the “Stocks in Roth advantage” – see below. The comparative advantage calculations are indeed consistent with the final portfolio values if your horizon is between one and 25 years. But after that, you cross the zero line and stocks in the Roth are better again. What’s going on here???

For shorter horizons, you’d be better off putting stocks in the taxable account. For horizons of 26 years and beyond, the Roth is the better place for stocks!

Well, here’s an explanation for this phenomenon: Drift in the asset allocation! You see, here’s the final after-tax net worth that came from stocks:

  • Stocks in taxable, Bonds in Roth: 65.74%
  • Stocks in Roth, Bonds in taxable: 78.92%

So, it’s no wonder you outperform loading up on stocks in the Roth: You had much higher (after-tax) exposure to stocks and thus also much higher (after-tax) stock risk along the way. So, if we compare 100% Stocks in taxable plus 100% in the Roth with 100% Stocks in the Roth plus 100% bonds in the taxable, we’re really comparing apples and oranges. The two portfolios drift apart over time if measured by their (after-tax) asset allocation and that explains why you get these counterintuitive results.

So, how do we account for the asset allocation drift? Two potential methods come to mind:

First, when we start with stocks in the Roth, I could periodically adjust the Roth allocation to keep the true after-tax allocation in-line with the other setup (stocks in the taxable). This would mean that we slowly shift from stocks to bonds in the Roth, which we can do without tax consequences!

Second, and this is the method I will use here in the Google Sheet because it feels a lot more intuitive and it’s easier to calculate as well. The results for this method are in the Google Sheet in a table to the right of the main results. It goes through the following thought experiment:

  • Let’s start with a 50/50 asset allocation in both accounts and assume that we don’t rebalance the portfolio, so dividends and interest are reinvested (after paying taxes) in that same asset class. Notice that due to the different average returns, you still get the drift toward a stock portfolio and end up with 72.37% from stocks in the final value ($917.73 out of $1,268.19). Not a bad idea, because remember, I showed that a glidepath shifting out of bonds into stocks (relatively) is a partial hedge against Sequence Risk (see SWR Series Part 19 and Part 20).
  • Next, we move to the middle panel, where we shift the entire taxable account into stocks, but in the Roth, we keep $8.80 in stocks and invest only $91.20 in bonds. How did I pick that number? I want to keep the final after-tax value due to stocks constant at $917.73. Since the taxable portfolio growth is hampered by the tax drag on the dividends, I have to set aside a small stock share in the Roth to compensate!
  • We could also do the opposite, i.e., move to the panel on the right where I go to 100% bonds in the taxable account. If I do so, I want to invest slightly less than $100 in stocks in the Roth, again to exactly target the $917.73 in final net wroth coming from stocks. That’s because the Roth IRA with its tax advantage will generate much faster growth, so to account for the asset allocation drift, I’d have to invest 91.20%/8.80% in stocks/bonds.
  • Also, I included the outcomes of the 100/0/0/100 and 0/100/100/0 approach again in the bottom panels in the middle and on the right, for comparison.
  • The outcome in this particular example is that it’s optimal to go 100% in the taxable account and 8.8%/91.2% in the Roth: $1,311.90 final after-tax portfolio value. That even beats the 100% stocks in the Roth plus 100% bonds in the taxable setup the optimal strategy when looking only at the corner solutions with 100% or 0% allocations. And you beat the 0%/100%/100%/0% even though you have a lower share of final net worth coming from stocks!
Accounting for the asset allocation drift: Stocks belong in taxable! Plus a small allocation to stocks in the Roth. This will beat all the other asset allocation options!

So, long story short: accounting for asset allocation drift, it’s optimal to invest the taxable account in stocks, if you also allocate a small share of the Roth to stocks to compensate for the tax drag in the taxable account!

Revisiting Case 2: How much of the “Stocks in Roth” advantage is really due to asset allocation drift?

Let’s revisit Case 2 (where stocks belong in the Roth IRA). That scenario generated a $596.60 final value, much better than the stocks in the taxable case: $549.63. So, the “right” asset location will increase the final portfolio value by a pretty impressive 8.55% over 20 years. But how much of that is due to the better tax efficiency and how much comes from the asset allocation drift? Let’s answer that question with the same table as in case 3, see below. Changing the allocation to either 100%/0% or 0%/100% in the taxable account, while fixing the final equity value at $433.56 (the same it would have been in a 50/50 portfolio) through the “correct” allocation in the Roth will produce only a tiny difference. Sure, stocks still belong in the Roth but the advantage is tiny: $574.44 vs. $571.79, a gain of 0.46% over 20 years. So, before the proponents of the “stocks in the Roth” approach spike the football (sorry couldn’t help but use the football reference – congrats to the KC Chiefs for an awesome game last Sunday), keep in mind that out of the 8.55% advantage, really only 0.46% comes from the tax efficiency and the rest from the asset allocation drift.

Accounting for asset allocation drift. Stocks certainly still belong in the Roth. But the gains are tiny compared to keeping stocks in the taxable account if we account for the drift!


How does this differ from the White Coat Investor analysis?

I’m not the first person to write about this topic. Most inspiring and informative was the exhaustive discussion on Dr. Dahle’s White Coat Investor blog:

WCI and I clearly agree on the following points:

1: “Stocks in taxable, bonds in retirement account” is not a universally optimal strategy.

2: If assets have very similar expected returns then sure, the tax-efficient asset goes in the taxable account and the tax-inefficient asset belongs in the Roth.

3: If you have a high-return, low-tax-efficiency asset, and a low-return, high-tax-efficiency asset then it’s a no-brainer: Put the former in the retirement account and latter in the taxable account.

4: If you are faced with a high-return, high-tax-efficient asset, and a low-return, low-tax-efficiency asset it gets more complicated. The pros and cons crucially depend on tax rates and – very importantly – on the expected returns!

5: Careful with asset allocation across different account types: $100 in stocks in a Roth and $100 in bonds in a 401k may give you a higher expected final after-tax net worth than $100 bonds in a Roth and $100 stocks in the 401k. But there’s no free lunch here! You have to factor in the tax implications! Your $100 in a 401k is worth only $80 if you have a 20% tax rate. So, the different final results come entirely from a different (after-tax) asset allocation, not at all from tax-arbitrage!      (WCI Pet Peeve #2)

And here’s where we disagree:

1: Muni bonds don’t add much to the discussion if you do a proper apples-to-apples comparison, i.e., Munis vs. taxable bonds with similar duration, default and prepayment risks. More on that below.

2: While I agree that different expected returns impact the asset location decision, we have to separate the effects of pure tax arbitrage and asset allocation drift. Not doing so will give you nonsensical results (see case 3 above, identical to one of the WCI case studies!). In that sense, WCI commits the exact mistake he laments in his Pet Peeve #2 (“Thinking That Locating Higher Expected Return Asset Classes into Tax-free Accounts is A Free Lunch”). If you notice, every single one of WCI’s simulations is of the form either 100% or 0% of one asset in the Roth and the opposite in the taxable account. But that’s ignoring the fact the $100 of assets in a taxable account are a very different animal from $100 in a Roth.

Admittedly, the “conversion” of nominal asset value into “true” tax-adjusted asset value is a little bit harder than in the Roth vs. 401k comparison, where you just apply a haircut equal to the marginal ordinary tax rate to the tax-deferred account. But there is a haircut to be applied to the taxable account and it depends not just on tax rates but also expected returns and the horizon.

But just because the haircut is more complicated to calculate doesn’t mean we can all pretend it’s not there! In my Google Sheet, I calculate that conversion factor. Here’s the calculation for Case 3, please see the chart and table below.

  • As you can see, $100 of assets in a 401k/Traditional IRA are worth only $67 due to the 33% marginal tax rate. That’s simple! It’s independent of the asset class, expected returns or the time horizon. Everybody should be on board with that one!
  • We can do the same conversion for the taxable account. This is calculated by scaling the after-tax final portfolio values to Roth=$100. Notice that $100 in stocks in a taxable account is worth only about $82.40 after accounting for the tax drag. That’s why we can’t compare a 100/0/0/100 allocation with a 0/100/100/0 portfolio. We can compare a 100% stocks in taxable plus 8.8% stocks in and 91.2% bonds in the Roth with a 100% bonds in the taxable plus 91.2% stocks and 8.8% bonds in the Roth, because they both have the same after-tax exposure to stocks, using the conversion factors below.
  • And again, really rubbing it in here, notice how the haircut for stocks in the taxable account is less than for bonds. That means stocks belong in the taxable. And 100% stocks in taxable plus 8.8%/91.2% allocation in the Roth gives you the highest final after-tax value.
Why the 100/0 and 0/100 portfolio comparisons are meaningless: $100 of stocks in a taxable account is very different from $100 of stocks in a Roth. (calculated for Case 3 parameters, see above)

How about Muni Bonds? A Red Herring!

Another potential reason to put bonds in the taxable portfolio: Municipal Bonds (“Munis”) offer tax-free income. Of course, there’s no free lunch: the tax-efficiency is already baked into the muni bond price (and hence the yield). So, if your taxable bond fund has a yield of x% then expect your tax-free bond fund to yield roughly x*(1-t)% and the implicit tax rate t would be somewhere around the tax rate of the typical marginal muni bond investor, which is someone in the top or at least top-two federal income tax brackets (35-37%).

For example, White Coat Investor runs his calculations once with a 2.69% taxable yield and 2.16% Muni yield and once with a 5% taxable yield and 4% Muni yield. Uhm, unfortunately, that’s unrealistic. Each case would imply only a 20% implicit tax rate. I’ve talked to some of my buddies who still work in the industry and specifically in the Muni bond sector and they confirmed that this is impossible. A 4% Muni bond would likely have risk characteristics (duration risk, default risk, prepayment risk, etc.) of a 6+% taxable bond. Comparing a 4% Muni bond in the taxable account with a 5% taxable bond in the Roth IRA is like comparing apples and oranges. Markets are not that wildly inefficient! If there is a non-arbitrage condition in the bond market and if you’re in the highest tax bracket you’re roughly indifferent between the two instruments. And if you happen to be a tax bracket below the top one then the Muni route would even worsen the result for the “bonds in taxable” route!

Still not convinced? Let’s put some data out here: On you can get Treasury and Muni yields, please see the table below. Over 1 to 5-year maturities, you can indeed see that Munis have a roughly 40% marginal tax rate built-in. Much higher than the 20% used by WCI and much more in line with the top tax rates. That implicit tax rate goes down when you look at longer horizons but that’s easily explained by noting that long-term Munis have a much lower credit quality than Treasury Bonds. Over relatively short horizons, nobody really expects Muni defaults now, but over longer horizons, that’s definitely in the cards! Looking at you, Illinois and New Jersey!

Treasury vs. Munin Bond yields. 1 to 30-year horizons plus the implicit marginal tax rate. Source Bloomberg, as of 2/3/2020

So, people can always come up with whacky examples like, hey, the Vanguard Muni Bond ETF (VTEB) yield of 2.3% is almost as high as the Total Bond Market ETF (BND) at 2.72%, but that’s comparing apples and oranges! Look at the maturities (much longer in the VTEB) and the credit ratings (lower in the VTEB) and that explains why the yield is so high. I’m only waiting for someone to point out that Munis actually have higher yields than taxable bonds: Look at the Nuveen NMZ: 4.9% distribution rate, vs. the 1.5% yield in Treasuries! Unfortunately, the NMZ is also a highly speculative, leveraged closed-end fund.

So, long story short: Muni bonds are a complete red herring in the context of these calculations here!

Some additional caveats

Just for the record:

  • With volatile asset returns, one could make an additional case for stocks in the taxable accounts not properly accounted for in my calculations here: Tax-Loss Harvesting.
  • Most retirees will already have large equity positions in their taxable account(s) with large built-in gains. Even if they wanted to hold bonds in the taxable account and stocks in the Roth, it would be too costly to first liquidate all the stock holdings in the taxable accounts. Another reason to keep stocks in the taxable account. Defer those gains in taxable accounts, see lesson 102 above!
  • There’s another advantage of keeping stocks in the taxable account. If you die before running out of money you can leave your highly appreciated stock portfolio to your heirs and they get the so-called “step-up basis,” so they can walk up the cost basis of the inherited asset (not just stocks but also real estate) to the value at the time of your passing. It’s effectively setting the cap gains rate to zero. You still have the drag from the dividend taxation before you die but the step-up basis is a hugely powerful estate planning tool. I was reminded of that tax hack when my buddy Justin from started an interesting Twitter discussion over the weekend on the step-up basis. But caution: that step-up tax hack might go away if greedy politicians have their way. I wouldn’t want to rely on this being available in 30, 40 or even 50 years!
  • In the calculations for case 1 (where stocks go in the taxable account)  I noticed that for a long enough horizon (51 years and longer), you get the same kind of reversal as in case 3. So, as long as your stock capital gains are taxed at a positive rate, it’s only a matter of time before the asset allocation drift will swing the pendulum in the favor of “Stocks in the Roth IRA” simply because the high-return assets grow the most there over time. But again, if you account for the asset allocation drift, the conventional wisdom prevails: Case 1 calls for stocks in the taxable account!


I almost feel sorry for writing such a long post to establish the really trivial insight that no asset location rule of thumb, neither the conventional wisdom (“stocks in taxable!”) nor skeptics claim (“stocks in Roth!”) can ever be universally optimal. You just look at the two extreme cases:

  1. Stock returns are tax-free because you’re in the 0% bracket for dividends and capital gains on your federal return and you live in one of the tax-haven states without an income tax. Then, you’d obviously put bonds in the Roth and stocks in the taxable account!
  2. Bond returns are very low or zero or even negative! Hey, it happened in Germany, Switzerland, Japan, etc. and might come to the U.S. eventually. Then you’d have to be crazy to put bonds in the Roth IRA!

And then all cases in between have to be carefully examined. The asset location advice has to be a lot more nuanced especially for the intra-asset-class location with very different expected returns. With today’s low yields, one could certainly make the case that bonds belong in the taxable account!

But I learned something new: Looking at the 100/0/0/100 and 0/100/100/0 corner solutions isn’t very informative in this context. I have provided a few more intuitive and more consistent calculations to weigh the tradeoff. And it all goes back to comparative advantage, how amazing is that???

And how much money can you squeeze out of this tax arbitrage? Not very much if you account for the asset allocation drift. That’s why I write so little about tax issues. But this one was certainly an interesting topic!

Thanks for stopping by today! Please leave your comments and suggestions below! Also, make sure you check out the other parts of the series, see here for a guide to the different parts so far!

Title Picture Source:


62 thoughts on “Asset Location: Do Bonds Really Belong in Retirement Accounts? – SWR Series Part 35

  1. Thanks for this! This is another brick in the wall for my theory that you can describe asset location in a way that is easy to understand! Next, you need to work on asset location when you transition from 100/0 to 60/40 in your pre-retirement glidepath to prevent SORR!

  2. Ern,

    You use the 10 year treasury bond in the comparison above and in many of your simulations. What does that look like if you are going to execute this as a dynamic percentage of your portfolio? (in an equity glidepath for example) Are these purchased directly from Treasury or is there a fund that roughly tracks treasury bonds with this duration?

    Big fan of your work! Thank you and keep the articles coming

    1. Great question!
      The returns I use are for the “10-year benchmark Treasury”
      So, there’s some turnover in that one because you sell the bond when it gets too low below a 10-year maturity and then replace it with a new bond that has a 10-year maturity.
      The closest you would get to this is an intermediate government bond fund, e.g. iSHares IEF, 7 to 10-year Treausuries.

  3. I just came to say thanks for the shout out at the end. 🙂

    And to say thanks for the food for thought. I’ve always treated that Roth space as magical for high growth less-efficient assets and fortunately I have plenty of traditional 401k/IRA space to dump those inefficient bonds in. Please tell me stocks in taxable, bonds in traditional IRA is a universally accepted truth!

    With dividend yields roughly equal to short-intermediate term high quality bond yields (aka 2%) I feel like it must be true (stocks–> taxable; bonds–> trad IRA/401k). Dividends will be favorably taxed and cap gains will always be taxed lower in taxable than if withdrawn from IRA (After they’ve appreciated).

    1. I could see a rationaly for “stocks in Roth” and “bonds in the 401k” if you face very large RMDs and you try to limit the growth of your 401/T-IRA while doing the Roth conversions prior to claiming SocSec.

      The allocation taxable vs. retirement all depends in the tax rates. If you can stay in the 0% bracket for dividends, you’re a great candidate for stocks->taxable.

  4. Ern: You are a master at drilling down to the core of financial knowledge, reminds me of the diamond drill that the father of Howard Hughes invented to drill for oil. Anxiously awaiting the publication of your book. Keep it coming!

  5. ERN, mid/late 50s retired couple. Goals to (1) build SAFE bridge to social security (62 / 70 claiming strategy); and (2) fully subsidized ACA premiums.

    Taxable income consists of dinky pension plus meager CD interest. For us, building a self liquidating CD ladder in taxable account accomplished our two goals. Total Stock Market Index in taxable account might lose money (possibly a lot of money) or alternatively create a lot of realized capital gain when liquidated.

    So our CD ladder in taxable account is perhaps inelegant (perhaps) but it works for us. To paraphrase Bill Bernstein, “Nothing helps me sleep better at night than a warm CD ladder.”

      1. I just think you need to take the premium tax credit into account in your calculations. A couple in their 50s could easily pay an unbelievable $18,000 in annual health insurance premiums. And this is paid with after tax money. So for example someone in the 15% federal plus state tax bracket would need to withdrawal $21,176 to have $18,000 left to pay premiums. This number can essentially be reduced to zero, in some cases, with some forethought on asset location. For someone living on $50K per year this increase required withdrawal rate by 40% or more depending on location of other assets! Just pointing out with my post that no conversation on asset location is complete without taking into account the comprehensive plan. Something for “late FIRE” readers to ponder when making decision on asset location.

  6. Ern: Thank you so much for your detailed post on this.

    I’m interested in a scenario to consider roth conversion after retiring at ~50 with a large tax deferred account.

    It seems that during the working years at the top marginal tax rate (and in a high tax state), one would want to have enough munis in their taxable account that would allow her to liquidate to pay for the conversion taxes and living expenses instead of liquidating equities because of SRR — and this feeds into a glide path strategy as well. A couple thing s I’d love to dive into are 1) does it make sense for her to convert her munis to total bond fund after “r-day” or should she not do this to leave more “room” in the lower tax rates for a faster/more efficient conversion and 2) would it likely return more to just keep stocks in taxable and liquidate them to pay for roth conversion taxes/living expenses even considering the SRR and keep bond allocation in tax deferred. It’s hard to find info on the roth conversion strategy for tax optimization during early retirement…its seems to be an under-discussed topic. Would love a post on it! 🙂


  7. Another advantage of stock in a taxable account is that for international stocks you will be able to take advantage of the foreign tax credit. International stock in an IRA/401k will lose the advantage of that credit.

  8. ERN. Excellent post. Your conclusion of the right strategy is “it depends” is spot on. One can make assumptions on returns and tax rates but not know whether any strategy was correct until many years down the road when once withdrawals the funds in retirement.

    However, one issue I have is your comment on muni bonds. There are arbitrage opportunities that affect the analysis for persons in high tax brackets. Unfortunately, it is impossible to have a truly apples-to-apples comparisons but one can come close. For example, one can compare the after-tax yields of Vanguard Intermediate Corporate Bond ETF (VCIT) and Vanguard High Yield Municipal Bond Fund (VWALX). Both funds hold primarily investment grade funds of intermediate duration. VCIT has a duration of 6.1 years and VWALX has a duration of 6.2 years.

    The credit ratings are less directly comparable as VCIT is primarily concentrated in the lowest rung of investment grade BBB (56.1%) with only 7.1% of bonds in the highest rungs of AAA and AA. In contrast, VWALX has more highly rated bonds (30.1% in AAA and AA) but it also has ~10% below investment grade bonds. Overall however, the funds appear to be comparable in credit quality.

    The SEC yield is currently 2.08% for VWALX and 2.53% for VCIT. The after-tax yield of VCIT for someone in the 40.8% federal tax bracket (37% + 3.8% NIIT) 1.50%, or 0.58% lower than VWALX. That difference is material in the analysis of whether to put bonds in taxable or tax deferred/free. At current rates, most investors in the highest tax brackets are likely to be better offer with muni bonds in taxable.

    1. Thanks!
      The VWALX and VCIT are two additional examples of bond funds that should not be compared. Never, ever trust the “duration” figures posted on Vanguard. They are most likely miscalculated for Munis.
      In your exmaple, the Muni Bond fund has an average maturity of 17+ years and a duration of only 6 years, which is mathematically impossible:

      The taxable bond has reasonable maturity and duration numbers:
      7.4y duration, 6.1y duration

      So, in a nutshell, you are comparing a Muni bond fund with 17 years maturity and taxable bond fund with a maturity of 7 years, and that alone explains the difference.

      1. Not that this makes the funds comparable, but I suspect the reason for the discrepancy between the maturity and the duration of the muni fund is because of embedded optionality in the bonds. Probably the maturity is just computed using the weighted average of the dates printed on the bonds, but the duration is calculated based on the earliest callable date.

        1. ERN: You are correct that the average stated maturity is materially different between VWALX and VCIT so it is not quite an apples-to-apples comparison. For that matter, as discussed below, SEC yield is not equivalent between them either. I was simply trying to simplify the comparison.

          Taliesin: You are correct that the discrepancy between the average maturity among two funds with similar duration is due to the higher call risk for the muni bonds. There is a higher percentage of muni bonds that are callable in VWALX than corporate bonds in VCIT. Vanguard (and most other fund and ETF providers) adjust the duration for the expected call date. This is prevalent when interest rates have declined as much as they have. A state or municipality is likely to call a bond paying 4% when they can refinance at 2%. This factor is called convexity and longer term bonds have slightly negative convexity. The risk is that when interest rates increase, the duration of muni funds will increase also as the bonds are less likely to get called.

          However, holders of muni funds (like holders of GNMA and other mortgage bonds who have a higher negative convexity) are compensated for this risk. I used the SEC yield above as a proxy but the more relevant factor for tax comparisons is the distribution yield. The SEC yield is based on what the yield would be if the bonds were held to maturity. However most bond funds do not hold bonds to maturity. Rather, they sell bonds as they move out of the target duration and maturity date for the fund. If there is an upward sloping yield curve, the SEC yield is lower than the actual yield will be.

          In contrast, the distribution yield is the yield actually paid by the fund. For VWALX, the distribution yield is 3.23% (after-tax yield of 3.23%) For VCIT, the distribution yield is 3.31% (after-tax yield of someone in the 40.8% tax bracket is 1.96%). The difference between the distribution yields of the two funds is 1.27% greater than the difference of the SEC yields of 0.58%. For convenience purposes in the original post, I simply used the SEC yield as a proxy for the call/negative convexity risk of muni bonds.

          The above analysis simply affirms ERN’s original post that the decision of whether to put bonds in taxable or tax-advantaged accounts is highly specific to the investor and depends on a number of assumptions that may or may not prove to be true. However, I do believe that for investors in high tax brackets, muni funds should be part of the discussion.

    2. In addition to ERN’s point I will add this: credit ratings agencies are not perfect and they just because two portfolios of bonds get similar ratings doesn’t make them equally risky. Investors make their own judgements as well.

  9. Not sure if I am misinterpreting this, but are you just setting the allocation of taxable vs. tax advantaged at the beginning and letting it ride? It seems to me that you would re-balance each year to your after-tax allocation.

    Also, thanks for reaffirming my choice to keep my stock futures in my tax advantaged accounts (high return, low tax efficiency). The only downside I see is that if we have a massive downturn, it would be more difficult to contribute more to meet margin requirements.

    1. I’ve played with the rebalancing approach as well (with similar results). For now, in this post, I assume that you set-and-forget your allocation and let it drift. That’s what most other people analyzing this tradeoff do (WCI) and I stuck with that tradition.

      1. Great article as always!

        One question I have on tax planning is for someone planning early retirement with access to a Roth mega backdoor, which can allow one to effectively put up to $57k into a Roth IRA per year.


        Does it always make sense to max out the contributions (57k of tax deferred per year), or are there situations where you may get into cash flow issues later on into retirement (e.g., running into RMDs or being forced to pay the tax penalty for withdrawing the capital gains portion of the Roth IRA)? I’ve done some basic analysis on the topic but I’d be curious to hear your thoughts.

        1. I never had a constraint like that. I always had enough money in taxable accounts. But I can see how some folks can get into a cash flow squeeze when too much money is tied up in tax-deferred accounts and you don’t have enough near-term. Always keep enough money handy to use before your first Roth conversions from 5 years ago become accessible.

      2. Having bonds in a tax advantaged account lets one rebalance during market swings without worrying about generating capital gains (or worse, if they’re ‘short term’, and subject to regular income tax). This alone makes it worth it for me. I haven’t paid capital gains on anything in over a decade of investing, and I’ve re-balanced at least once or twice a year.

  10. Great post, but a bigger benefit would be how to transition a “sub-optimally” allocated portfolio, which nearly everyone has, to a better tax efficient one, accounting for the after tax AA and the cost of the transition.

    Also, seems that always using the marginal tax rate on distribution is incorrect – unlike when saving, distributing should use the effective rate to get the correct blended influence of distributions across the tax/income spectrum. Plus, with various benefit cliffs in healthcare and whatnot, having tax-free Roth money is of greater importance for most retirees who are in that income/wealth zone.

    Excellent post, as always!

    1. Most people should already start with equities in the taxable account.
      To do a transition, you’d face some of the same tradeoffs I pointed out in a post last year:

      Disagree with your approach about the effective vs. marginal rate. When you don’t have a flat tax you’d do the tradeoff in stages:
      First, you’d fill up the 0% “bracket” (standard deduction) with retirement income using a 0% marginal tax rate. Then the next using a 10% bracket, then the 12% bracket, etc.
      You can’t just mush together all the tax rates into an effective/average tax rate. You’d get wildly incorrect recommendations!

  11. Great article as always!

    One question I have on tax planning is for someone planning early retirement with access to a Roth mega backdoor, which can allow one to effectively put up to $57k into a Roth IRA per year.


    Does it always make sense to max out the contributions (57k of tax deferred per year), or are there situations where you may get into cash flow issues later on into retirement (e.g., running into RMDs or being forced to pay the tax penalty for withdrawing the capital gains portion of the Roth IRA)? I’ve done some basic analysis on the topic but I’d be curious to hear your thoughts.

  12. I think you are understating the case for why bonds should go in tax advantaged accounts over stocks. You allude to some of these points but I think you underestimate them.

    On average, stocks will outperform bonds, but with much higher variance. The IRS not only taxes part your gains, but they will also reimburse part of your losses (through tax loss harvesting) reducing the amount of risk that you are effectively bearing yourself. You can thus increase your holdings of stocks while maintaining the same level of risk/ ‘tax-adjusted asset allocation.’

    not only that, you actually get some positive convexity from holding stocks in a taxable account:
    – you can strategically harvest losses in high tax years and gains in low tax years (the first 3k of losses are especially valuable)
    – you can harvest losses quickly, and defer most of your gains indefinitely far into the future
    – you can avoid around paying cap gains taxes all together by either donating appreciated stock to charity or through the step up in tax basis on inheritance.

    I think you could construct cases (high income, moderate portfolio) where you would prefer to keep stocks in a taxable account even with 0% interest rates because the value of potential tax loss harvesting is so large.

    1. Yes, agree. Hence, I included the first bullet point in the “caveats” section.
      But also keep in mind: TLH becomes less useful in retirement. If you stay in the lowest 2 tax brackets, TLH becomes completely useless.

      1. I saw that comment, but I think that these points are really central to the argument for why bonds belong in tax advantaged accounts and deserve more than a minor footnote in a long article on that subject. From the tone of your article, I strongly suspect that while you do recognize that some of my points matter you haven’t appreciated how important they are. Bonds in taxable will almost always be the right answer for reasons I outlined, and I think most of your readers will not understand how strong the case is from your post.

        You get less of the positive convexity in retirement, but the risk sharing argument still applies as long as you expect to face a non-zero cap gains rate. That is adequate to think that there is more value from putting bonds in a taxable in most cases. If you expect to pay 0% cap gains then stocks should clearly still go in your taxable account.

        I think the only cases where I would consider preferentially putting stocks into a tax-advantaged account are ones where you experience negative convexity in your expected tax rate. This is likely if you have a large portfolio (such that you will likely end up paying substantial cap gains taxes) with few unrealized capital gains (thus might not be able to offset future losses against unrealized gains from prior years). This would most easily apply to a retiree who recently received a large inheritance but who otherwise had modest savings outside of retirement accounts. It would be less likely to apply to someone who has been aggressively saving over the past decade. In this situation, you *might* prefer to put stocks in a tax-exempt account but should probably avoid doing so in tax-deferred account because you will also experience negative convexity there (ie, large gains will push you into a higher tax bracket, but losses will do little to reduce your tax burden as you will end up in a lower tax bracket)

        1. From the tone of your article, I strongly suspect that while you do recognize that some of my points matter you haven’t appreciated how important they are.

          From the tone of your comment, I strongly suspect that you should never make assumptions about what other people should or should not appreciate. TLH might make a difference at the margin (hence the comment in the “caveats” section). I have done extensive research on TLH here on the blog. In rare cases, TLH can add substantially to the after-tax expected return of an equity portfolio. But the effect can also be small, in the neighborhood of maybe 5-20bps, where the TLH effect would have an impact only if the investor is about indifferent between the two asset location scenarios. Everyone has to find out for themselves. Here are my classic posts on the topic so far, in case you haven’t noticed:

          I also did some calculations on TLH specifically for this asset location blog post, but I decided that it wasn’t worthwhile to go into more detail after getting close to 6,000 words already. Maybe I will include those calculations in a future post.
          So, in a nutshell: the first-order effect is from the tax-arbitrage, as outlined with my comparative advantage effect. You can certainly slap on some additional assumed TLH return to the taxable stock portion, but I’d caution, both from my personal investing experience over 20+ years and my research on the blog that the impact is not that great.

          As for the rest of your comments: I’m glad your intuitive approach yields roughly the same results as my numerical approach. At the end of the day, I have to put some numbers into the formulas I supplied to showcase where some of the bloggers got their numbers wrong. Hard numbers make the case much easier. Mumbling about convexity and risk-sharing is a lot less convincing. I just don’t like blanket statements.

          But just to be sure: If you think that my blogs posts are so lacking maybe you start your own blog… 🙂

          1. I wasn’t trying to insult your blog, overall I think it is one of the most interesting sources for personal financial analysis I have found. However, that doesn’t mean that you understand everything (no one does). In this case you are so confident that you understand everything that you don’t seem to be considering my comment carefully. I believe that you have largely misunderstood my point.

            I have read all of those posts in the past and have reread the second post you link to just and parts of the third to make sure I remember it correctly. Part of the problem is that you define tax loss harvesting more narrowly than I do and most of the benefit that I am describing is not captured in your model for TLH. Only some of the benefits I describe in my first comment, and none of my second comment require a particular strategy for harvesting tax losses. It is likely that my first comment would have been more clear if I had never used the words ‘tax loss harvesting.’ That was probably careless of me and is my bad.

            Maybe I will try again:
            I agree that it’s possible that, in a low interest rate environment, placing higher variance/return assets in a taxable account will result in a larger average tax drag (depending on your tax bracket). However, by placing higher variance assets in a taxable account, you increase the correlation between your tax bill and your returns. This correlation reduces your tax-adjusted exposure to the market (or allows you to increase your equity allocation for comparable risk) by an amount that will (in most circumstances) more than compensate you for the higher *average* taxes owed. This would even be true under a model where you don’t TLH but instead liquidate your entire portfolio after a fixed number of years.

            One could certainly model that more explicitly. I am confident that if you considered the impact that different asset location choices have on the variance of your portfolio instead of only considering the mean (or if you considered the 5th percentile worst outcome) that you would see that stocks in taxable/ bonds in tax-advantaged do significantly better than the reverse even under your assumptions in ‘(revisiting) case 2’

            TLH (as you define it) is an additional benefit of holding stocks in a taxable account. I did mention it in my first comment, but explicitly did not mention it in my second comment (since you mentioned that it is less valuable in retirement). The benefits that you describe in your previous posts are not the primary benefit that I was describing.

            In addition, I do think that TLH is likely more valuable than you do. That would likely require a much longer discussion to sort out.

            1. Gotcha. Sorry, must have misread your other comments. I think I get what you’re after. You raise an important point. stochastic returns combined with the convexity of the U.S. tax code create another complication in the calculation.
              But as I said before, I already had 6000 words and I mostly tried to stay within the framework of what others had used (WCI, etc.) to make my case. But as you point out, this is not the final word either. In a follow-up post, I should certainly look at the effect that you pointed out. So many things to do!!!
              By the way, the larger impact may be on the Roth vs. 401k asset location. With a flat tax rate you can always argue that a 401k is simply a Roth times (1-tax). But there is a lot of convexity in the ordinary tax schedule, even more than in the capital gains schedule.
              Have to think about this more. Thanks for the suggestion! 🙂

              1. I was trying to emphasize that these are important points that don’t get the attention that they deserve in general. I shouldn’t have phrased that thought in a way that implied that your post wasn’t a step in the right direction.

                I think that interest rates would need to be quite low for most people to consider placing stocks in a tax-exempt account, though it depends on what sources of income you have, portfolio size, amount of unrealized tax gains, investment objectives, age, etc etc…

                If you start with a *very* naive model with a flat tax rate, even a 0% interest rate would leave you nearly indifferent between stocks in taxable or tax-exempt . A more correct model needs to grapple with the convexity of a progressive tax code, but there are a lot of sources of positive convexity as well. (Not only TLH as you define it, but the step up in tax basis on inheritance and the ability to donate appreciated stock to charity can be very important as well, depending on one’s investment goals).

                Completely agree that stocks shouldn’t go in a tax deferred account, I mentioned this in my second comment as well.

  13. Couple thoughts that come to mind. For people still accumulating (me), I have chosen to hold my lowest expected returning assets in my pretax accounts (that does happen to be bonds and cash), because I want to slow the rate that my pretax accounts grow to the point where that they will fill the expected 12% tax bracket for the rest of my life. This is so that i can continue to defer more income in higher tax brackets for later withdrawal at lower brackets. I realize that this means i push stocks into taxable. I’m glad you pointed out that stocks have a step up basis upon death. This becomes really important when talking about designing your estate. With the Secure Act, people only have 10 years to disburse assets from an IRA. Therefore, one should try to liquidate their pretax assets of their lifetime, especially, if your heirs are in a highish tax bracket. Furthermore, there is no longer an insane amount of value to leave Roth to heirs due to the 10 year limit, therefore, one might want to consider prioritizing appreciated stocks in taxable as the main part of the estate to leave. Hopefully, i’ll be able to empty my pretax accounts in my lifetime, and spend heavily from Roth, and hopefully not sell in taxable and only realize the income being kicks off of the account.

    The combination of these things means that as long as the tax costs for dividends in taxable are reasonable, and i need dollars to spend anyway. That isn’t too big of a burden. There is huge value in putting low appreciating securities in pretax.

    1. Very good point! For people who are afraid that RMDs will push them into the 22% bracket (or more) there’s another reason to put the lower exepcted return assets into the tax-deferred accounts! That’s something I’ve considered for my own planning. So, so the Roth conversions and invest in stocks in the Roth and keep the low-return assets in the 401k/TIRA.

  14. Please excuse me, but I have another question related to this topic that I haven’t seen discussed in depth anywhere: How is asset location strategy changed when one is nearing early retirement well before the “easy” distribution age of 59.5 for tax deferred accounts?

    Since there’s no clear winner in the tax efficiency debate, how about volatility management? When one needs to draw from taxable accounts for a decade or more prior to age 59, would it not be wise to put the more volatile investments (stocks) in the longer-term tax-deferred/tax-free accounts, with the safer portion of a balanced portfolio (bonds) in the taxable accounts?

  15. As someone who is still accumulating, I’m not sure what to make of the whole asset location debate because – as I suspect is true of most people – I have very different investments available to me in my tax-deferred accounts (workplace retirement plan) versus what I can get through a retail brokerage. For example, in my workplace plan I have access to Vanguard institutional class shares with a super-low expense ratio, which I could never buy in my taxable accounts without paying much higher fees. I have access to guaranteed contracts that pay 3%+ with no risk – again, unavailable outside the plan. So, do stocks then go in tax-deferred? Or perhaps I should use the guaranteed contracts for my fixed income allocation? I just don’t see how you can decide this without looking at what’s available in your workplace plan, and everyone’s plan is different.

    1. Yeah, good point. This is only in the stock vs. bond space. You may have your own personal options. Probably the 3% guaranteed return is a good option to have in the tax-deferred. Fill up the rest with stocks in the tax-deferred and keep all stocks in the taxable.

  16. I’ve done calculations which keep the asset mix fixed by rebalancing each year (to a 75/25 stock/bond mix, or a 50/50 stock/bond mix), and which handle tax calculations for gains that arise from the rebalancing. Using a range of parameter values, the outcomes I get are all within about 3% of each other (after tax) after 40 years. For the ranges of parameters that are most realistic, having the bonds in the *taxable account* wins, but of course it depends on the parameter values chosen. But more importantly, either way the effect is not large. I’m using Canadian taxation, but I tried a variety of tax rates and got similar results.

    A related comment: in Canada we have HBB (Horizons CDN Select Universe Bond ETF) which uses swaps to replicate a bond index as pure capital gains. With this, your stocks and bonds have the same taxation, so it definitely makes sense to put the one with lower expected growth in your taxable account. Is there something similar in the US?

    1. Thanks for confirming. In SWR Series Part 39 I calculated the differences of changing the rebalance frequency in the context of withdrawals. Monthly/Quarterly/Annual wouldn’t make a huge difference after 30 years!

      1. You wrote “Thanks for confirming”, but my main point was that my calculations gave a different conclusion about where to hold bonds. (They also showed that the choice doesn’t matter too much.)

        1. Gotcha, misunderstood you initially.

          Generally, if you want to constantly rebalance you have an additional inventive to keep stocks in the tax-advantaged account and bonds in the taxable. That’s because you want to constantly rebalance out of stocks = the asset that appreciates more (on average). That’s more efficient if done in the tax-advantaged account.

          But I provide one example for 50/50 with regular rebalancing. Do you get the same results? I found that bonds belong in the taxable account in that example. So, I don;t really see where the disagreement is.

  17. I didn’t follow the section where you take into account asset allocation in Case 3. You seem to be still allowing it to drift, but with a target final allocation. That sounds very different to me from the more realistic desire to actually control the asset allocation over time. There’s another example called “Revisiting Case 2”. Maybe that is the one where you rebalance? The prose there doesn’t mention rebalancing, so if that’s what you were doing, I didn’t realize it. (I now see that a table heading has the word “rebalance”.)

    In any case, I don’t disagree with any details, but the overall summary seems to come out in favour of bonds in the tax-sheltered account, while my person overall summary would be the opposite (especially with access to a fund like HBB). Another advantage of having the higher growing asset in a TFSA (like a Roth in the US) is that those funds can be accessed very freely, with no tax consequences, even if they are all redeemed in the same tax year. Selling a large amount of stock holdings in a taxable account in one year could bump you into a higher tax bracket.

    Thanks for a great series! I’ve learned a lot and have been motivated to improve my own stochastic simulations based on your tips.

    1. We might be talking pasty each other. I did my calculations. If there’s a mistake in the calculations, please let me know where.
      I don’t know what you calculated, so I can’t tell you what causes differences in results. (though I did venture a guess in an earlier reply).
      I did a calculation where I try to control for the asset drift, so as to eliminate the whacky intermediate case (“Case 3: Depends on the horizon!”)
      Not exactly monthly rebalancing, but due to the tax implications along the way it’s not that trivial to simulate this (especially considering that actual returns will also be volatile, not just a fixed calibrated number).

Leave a Reply to Cancel reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.