Here’s an idea for a new ETF

Actually, not one ETF, but two! Or more! How can there be a need for a new ETF? Aren’t there enough already? Earlier this year, Motley Fool argued there are too many ETFs (1,929 at that time, probably over 2,000 by now) and they are covering pretty much every thinkable (and unthinkable) benchmark. Soon we might have more ETFs than publicly traded equities in the U.S., how crazy is that??? Why would I propose a new ETF that doesn’t already exist?

Here’s some background. I’m an index investor at heart and I like tax optimization. For so many years now, I’ve held equity index ETFs and Mutual Funds in both taxable accounts and tax-deferred accounts (both retirement and deferred compensation at work). It’s so painful to see the dividend payments in the taxable accounts getting taxed every year. Sure, it’s only about 1.9% dividend yield in the S&P500 right now but for us, that’s taxed at 15% federal, 10+% state (California!) and 3.8% Obamacare tax, for a total of almost 30% marginal tax! Isn’t there a better way? Sure! Simply put the taxable equity allocation into stocks that pay zero (or close to zero) dividends and keep the high-dividend stocks in the tax-deferred account where they can compound in peace and be taxed only once upon withdrawal rather than every year along the way! So, the two ETFs that I wish existed would exactly replicate the S&P500 if held in equal shares. But individually they’d have non-index weights and one would hold the equities with the lowest dividend yield and the other with the high-yield equities!

Notice that most folks already do this tax optimization across asset classes: Hold the tax-inefficient asset classes (bonds, REITs, etc.) in tax-deferred accounts and equities in taxable accounts. So, why not do this within the equity asset class as well for additional tax efficiency? How much extra after-tax return would we get out of this? Let’s look at the numbers…

How much variation in dividend yield is there?

A lot! I pulled some data from the web for all the stocks in the S&P500. Fidelity has a neat “Stock Screener” tool where you can pull all sorts of stock metrics, including dividend yield. Here are some fun facts (all figures as of 5/25/2018):

  • There are actually 505 stocks in the S&P500. This includes two version of Google!
  • The total market cap is almost $25 trillion (almost 1.5x annual GDP in the U.S.!) and the total dividend payments amount to about $470b a year, or about a 1.9% yield.
  • 88 companies, equal to 19% of the index market cap, pay no dividends at all! The largest five are Amazon, Facebook, Google, Berkshire Hathaway and Netflix, responsible for more than 10% of the index capitalization!
  • The highest dividend yield (by far) is almost 12%: Century Link (CTL). There’s a significant share of companies (20+% of market cap) that pay 3% or more dividend yield!
DivYieldStats Chart01
Stocks in the S&P500 sorted by dividend yield!

So, imagine I hold 50% of the overall equity portfolio in taxable accounts and another 50% in tax-deferred accounts. How low could I push the dividend yield if I were to hold the taxable account only in the lowest-yielding equities? The answer is 0.76%. That’s the average (market-cap weighted) dividend yield in the lower 50% of the spectrum. The remainder of the S&P500 equities, to be held in a tax-deferred account, yield a bit over 3%! The average of the two is, of course, just exactly 1.9% as in the S&P500. If the split between taxable and tax-deferred is 40/60 we can push the dividend yield in the lower portion all the way down to 0.45%.

DivYieldStats Table01
Yield and Sector Composition of the S&P500 and different sections of the dividend yield spectrum.

If we assume a marginal tax on dividends of 29%, we’d save (1.90%-0.76%)x0.29=0.33% in taxes every year. 0.42% for the 40/60 split. Not a bad boost in the after-tax returns for essentially doing nothing. No market timing, no stock selection, simply a smarter way of allocating between taxable and tax-deferred accounts!

Alternative ways of dividend-hacking

Until iShares, State Street, Vanguard, Schwab, Fidelity, etc. set up ETFs to replicate this, there are (at least) three alternative routes to do this dividend hack:

  1. DIY! Simply buy the appropriate equities in your taxable and tax-deferred accounts. Well, it’s not that simple. If one has to spend $5 per equity trade that would negate all the tax efficiency gains. But there is a way now: FinTech to the rescue! M1 Financial is a cool new brokerage that allows you to purchase shares, even fractional shares (!), commission-free. (Note: this is an affiliate link and if you click it and open an account with at least $1,000 initial investment I’d get a referral bonus! But rest assured that I actually really like this company and I’m working on a more detailed review, coming later this year!) I’d probably not go full-hog and buy all 500 stocks in the S&P. Probably set up two “pies” in M1 Finance with 50 stocks each, one with the high dividends and one with zero/low dividends. That would get you close enough to the index already.
  2. Value vs. Growth ETFs. One could use the IUSV and IUSG (iShares Value and Growth ETFs) with a nice low expense ratio (only 0.05%, lower than the Vanguard Value/Growth funds!) and the two will roughly aggregate up to a broad index again (in this case the S&P900, i.e., large-cap plus mid-cap). It’s a low-hassle solution but not quite as effective. The yield in the Growth ETF is still about 1.27%, so there are obviously many growth companies that pay higher than average dividends. So, you’d get only about half the tax efficiency this way!
  3. QQQ vs. HDV. Another option: pick two equity ETFs that roughly resemble the characteristics of the top/bottom 50% dividend yield stocks in the S&P500. For example, the QQQ (Nasdaq 100 index fund) has a low dividend yield of only 0.83% with a large Tech overweight: 60+%, thus, even a bit higher than the Tech share in the low-dividend yield portion of the S&P. Likewise, the HDV (iShares High Dividend Yield ETF) has a 3.62% (!) dividend yield with heavy overweights to Consumer Staples and Energy, just like the high-dividend stocks. The caveat is that 0.5x QQQ plus 0.5x HDV is still not really near the S&P500 sector weights (and even less so with the individual security weights) so there will be some mistracking vis-a-vis the S&P500.
  4. Any other suggestions? Maybe the readers want to weigh in and suggest pairs of ETFs that would do trick! 🙂
DivYieldStats Chart02
Dividend Yields. For the ETFs, I’m using the SEC Yield as of 4/30/2018.

Some caveats!

As always, there are some caveats, so just for the record:

  1. Most folks have already built their sizable index ETF/Mutual Fund portfolios. There is no point in selling your existing funds and realizing taxable capital gains. This dividend hack is most likely relevant for new investments only!
  2. For us personally, the new ETFs would come too late! 2018 will be our last year with high marginal taxes on dividends. In the future, we plan to stay within the 0% tax bracket for (qualified) dividends, stay well below the Obamacare tax cliff and reside in a zero-tax state (likely Washington State). In fact, given our new tax parameters, we might even go the exact opposite route and shift more money into high-dividend stocks/ETFs and completely forget about the low-dividend portion. That’s what retirement does to you: once the paychecks stop rolling in you get a new appreciation for dividends and income! 🙂
  3. There will be some turnover, namely when companies jump from the low-dividend to the high-dividend bucket or vice versa. The ETFs would then potentially realize taxable capital gains by reshuffling equities. Probably not enough to undo the tax advantage but it’s something to be aware of. It’s a reason to go the DIY route where you have control as to whether it’s worthwhile to shift stocks from one bucket into another when dividend yields change over time.
  4. There will be a slight style drift. If you take your VOO or IVV index ETF and reinvest the dividend payment it will be spread over the entire ETF by purchasing new shares. So, in a way, the high-yielding stocks generate dividends to be reinvested into the zero/low-dividend stocks as well. But if we keep the two components separate there will be no more “cross-subsidization,” i.e., the dividend income from the high-yield bucket will be reinvested only in the high-yield bucket. Without further adjustments, the two components may drift a little bit apart and no longer resemble the benchmark index after a few years of dividend reinvestments.
  5. What if someone has a 30/70 split between taxable and tax-deferred accounts? Or 80/20? There’d have to be customized pairs of ETFs to match the different investor preferences. Not sure the ETF providers like to issue too many different ETF pairs, but 25/75, 50/50 and 75/25 might cover enough ground for most investors!
  6. One could even improve the tax hack a little bit more: There are still a few REITs in the low-dividend portion (0.28% weight). Shift those into the tax-deferred bucket to avoid the dreaded non-qualified dividends from REITs that are taxed at the full ordinary income!

We hope you enjoyed today’s post! Looking forward to your comments and suggestions!

66 thoughts on “Here’s an idea for a new ETF

  1. Yes! How do we make this happen? Do you know anybody at Vanguard? I don’t think my Flagship rep has enough pull, but I love the idea.

    While many investors have been touting the benefits of dividends, I’ve been wanting to avoid them like the plague; a slight majority of our funds are in a taxable account and the tax drag is real, even with index funds. Last year’s dividend take added an unnecessary $8,000 or so to our state and federal income tax bills.

    I imagine there are some growth funds that perform somewhat similarly to your low-dividend fund, and there’s Berkshire Hathaway with its no-dividend policy, but I do like the idea of owning 250 blue chip, low-dividend companies in one fund.


    1. Thanks, Dr. PoF! Unfortunately, I don’t know anyone at Vanguard. Blackrock (iShares) is just two blocks down the road but they’re our mortal enemy. Have to wait until I officially resign to go even close to there! 🙂
      It’s quite amazing that there are so many high-dividend tilt funds now, but no real serious low-dividend ETF out there! So, one might be stuck with replicating the S&P with 100 or 200 stocks via M1 Finance. Maybe even do the low-dividend yield portion in M1 Finance and the high-dividend portion with one of the ETFs in the IRA.

  2. Brilliant idea. I can’t believe the industry has not done this already. It seems so obvious, but great ideas often seem obvious in hindsight.

      1. Brilliant indeed, Karsten! There is, however, the rub of Ibbotson study, that even on a total return basis, the dividend yielding cohort returns more over multiple decades than no dividend cohorts. This study was done from 1929 till 2012 as I recall and is covered in one of my Investing Series posts. Assuming that there is no difference in total return between dividend yielding and non-dividend stocks as a group, then your idea makes eminent sense for tax-minimization.

        1. That doesn’t invalidate anything I wrote. You keep the higher-yield (and higher return) stocks in the tax-deferred account and you should roughly replicate the S&P500 return in aggregate but with better tax efficiency. That’s all claimed here.
          This is not about style investing. The benchmark is the S&P500. That’s because I can’t guarantee that the dividend tilt will continue to outperform in the future.

          1. I didn’t say it invalidates anything Karsten. I only wanted to point out that any style based (dividend or no dividend) distortion can inadvertently skew the final result – which I assumed here to be maximizing tax-adjusted total return. I also can’t guarantee a dividend tilt will lead to superior total returns in the future, just because they have done so in the past as Ibbotson study and others have shown.

              1. Now you are asking a specific question for which answer depends on individual tax situation. If I have significant accumulated tax credits, I wouldn’t mind loading up on carefully selected dividend ETF (not the purely “high dividend” type) in both taxable and tax deferred accounts. If my allocation called for a certain percentage, then I would load up first with tax deferred and then move to taxable account for obvious reasons. The US laws are structured such where its possible to earn $80K in dividends from taxable account and still pay zero federal and state taxes for a MFJ living in a tax friendly state as well. Even if I take a high quality dividend ETF like SCHD yielding just 2.6%,it would take substantial assets (over $3 million) in taxable accounts to generate enough dividends to cross the tax threshold for MFJ filers. God, I wish I had that problem!

                1. Thanks for sharing! Sorry for putting you on the spot! 🙂
                  But just for the record, I think we’re on the same page! I will move some equity holdings to high or higher-dividend stocks while in retirement!

            1. But again, just for the record: Your point about different dividend-styles having different total returns (certainly different returns in the past, potentially different expected returns going forward) is exactly the reason I like to stay away from style biases. I propose to construct two ETFs that, when combined, exactly yield the S&P500 without any style bias.

  3. Excellent idea! Certainly cheaper and more efficient that tax loss harvesting strategies at the robos.

  4. One possible issue is that the total return and growth of the high dividend investment in the tax-deferred account will eventually have to be withdrawn, with RMDs, and depending on the tax situation, could be subjected to much higher effective tax rates than the dividend rate youre trying to avoid. Otherwise, I have been doing something similar for clients for years!

  5. This is a very practical ETF idea with tangible benefits. I’m trying to see if its possible to implement this fully and cheaply for a typical user, but it looks like M1Finance doesn’t let you go under 1% allocation for a specific stock and Robinhood still doesn’t support fractional shares. I think doing a partial representation in M1Finance like you’re saying would work well enough, but unfortunately it’ll require continuous monitoring to make sure all the % allocations stay in line with the index. Would be nice if M1 added pie sharing (similar to Motif) so only one person in the community would need to maintain it.

    1. I’d go with a 50 share pie in M1 and set the initial % to roughly match the market caps. It won’t be exact because the steps are 1%. But should be close enough!
      And completely agree: “pie sharing” would be nice!

    2. “but it looks like M1Finance doesn’t let you go under 1% allocation for a specific stock”

      You can with some work. You can nest pies within one another.

      If you have a stock that’s allocated to 1% in a pie, and then put that pie into another parent pie that’s allocated as 50% of your portfolio, that 1% stock is 0.5% of your portfolio.

  6. First up, I want to say that you think about way more serious things than I do while enjoying early retirement 🙂 My biggest thoughts of the day are “what am I going to eat for lunch”, “what should I watch next on Netflix”, and “will a second cup of coffee REALLY prevent me from sleeping tonight?”.

    On to the meat of your article – Clever idea! I was never in a high tax bracket and always paid 0% on qualified dividends (which SP500 is 100% qualified IIRC), so I didn’t care much. But even in my low tax status, I still paid (and continue to pay) an effective 21% to 35% marginal tax rate on dividends (6% state tax + 15% IBR student loan repayment implicit tax + ~14% loss of ACA premium tax credit subsidy). So this strategy makes some sense for even the “poor” on paper like me (in the $40-42k AGI range typically).

    A couple thoughts on ways to implement the strategy.

    One would be a Vanguard Tax-Managed Large Cap fund that would be an index-ish fund that does sort of what you suggest without the complementary Tax-Sucky-Managed Fund that you would stick in your IRA. This fund sort of exists (the “Vanguard Tax-Managed Capital Appreciation Fund Admiral Shares (VTCLX)”) but the dividend yield seems to be half way between the Vanguard SP500 and the Mid-Cap index, so not a real significant tax savings vs just owning the two index funds that make up the Tax Managed Cap Appreciation Fund. Maybe it performs more tax-efficiently at other times?? And now that I write all of that out, it’s basically saying “just invest in even more tax-efficient focused investments than a regular index fund” though it still won’t get you that complementary fund to stick in your IRA (though maybe a dividend-focused index fund might get you close?? Like “Vanguard High Dividend Yield ETF (VYM)”).

    Another way to implement a big “basket of stocks” portfolio would be with the Motif Investing platform. I can’t remember the details on commissions but it’s pretty low (maybe $5-10 to trade a basket of 30 stocks??). Then you could do 1-2 baskets of low div stocks in taxable accounts and similar for high div stocks in tax-deferred. If you needed to add or subtract funds from either of these 2 accounts then it’s relatively straight forward in terms of execution. I say that as someone who’s read about Motif but not as someone who’s actually used Motif’s platform 🙂

    1. Haha, thanks for the compliment! 🙂
      I saw the Vanguard tax-managed fund. I’m shocked and disappointed that it still has a pretty high dividend yield (>1.5%). Not much of a tax saving and this fund also has a higher expense ratio.
      Motif is an interesting idea. One would probably need around 2 motifs for the low-dividend portion and then simply use a high-dividend ETF for IRA. I checked the motif site and all dividend-themed motifs are about high dividends. If anyone comes up with a low-dividend theme: you’d be the first and could corner the market there!

      1. I was surprised at how little difference there is between VG’s tax managed funds vs the regular index fund too. I decided to invest in the small cap tax managed fund many years ago in a taxable account and now it has doubled or tripled so I’m pinned down with the cap gains curse, even though the delta in div yield vs the small cap index fund is only 15 basis points (yet I’m paying an extra 5-6 bps in ER!!). Not the smartest move in hindsight.

        1. Ah, OK, interesting! Small caps tend to have a lower yield. If you go with small-cap growth you can probably push that to below 1%. But there’s the style bias. Better not do this with 50% of the portfolio! 🙂

  7. Turnover rate is another tax event that is out of our hands with funds or ETFs. Although capital gains are more favorable, depending on your income, they still count in the calculation of the Obamacare subsidy.

    This is really an interesting post. I’ve recently started rereading an oldie from Jeremy Siegel, “The Future for Investors”, his follow-up to “Stocks for the Long Run”. He had tons of free labor, called students, and they went back to see what happened if you bought and held all the duds of the SP500 of the past. The findings showed that you would have done better not to purchase the newer companies being added, but stuck with the old, boring businesses. Even if they closed up shop, you would get shares of other companies who bought them out, etc. Bottom line was that you made these better gains because of reinvesting dividends and buying at better prices because the companies are not as popular. So, as to your proposed ETFs, my guess if you waited 20 years, is that the high dividend value ETF would beat the growth.

    You are so right that it is amazing that with thousands of options, there really aren’t many good choices for the taxable account. PhysicianOnFIRE wrote about selecting Berkshire Hathaway for that reason. Now that we are keeping income low enough for Obamacare subsidies, those “pesky” dividends are sometimes a “problem”!

    1. Thanks! Good point! Maybe another reason to not use the ETF because that would churn index constituents both as they leave/enter the index and as then go from low to high yield bucket or vice versa. Just buy and hold will do well enough!

    1. Nice! This post of him, though sound more like you’ll want to stay away from dividends. Period. Not just in the taxable account but also tax-deferred account! Ouch! Thanks for sharing, very thought-provoking!!!

  8. Regarding caveat #1, if one’s plan includes tax gain harvesting in early retirement, one could move at no tax cost to the split system. There may be opportunity cost in foregone Roth conversions though.

    I am finding that in early retirement I will Roth convert maybe two years out of three (to the top of the 22% bracket, so no tax free LTCG those years), then gain harvest to the zero cost limit the third year. The gain harvesting is to raise funds for living expenses and conversion taxes, but this year I moved some sales proceeds right back into equities, thinking to have something available for sale with a high basis.

    1. Very interesting! Thanks for sharing. True, I assumed zero capital gains in retirement. The tax advantage is diminished (but not eliminated) when you assume that you pay capital gains in retirement.

  9. Fantastic idea. As my taxable balance grows, I become more and more aware of those pesky dividends. I’d certainly consider buying these funds you suggest over the full index.

    If I were to set it up myself, I’d prefer to just split the “no dividend” stocks from the others so you can have an investment that avoids all of them if you want. You can then just weight your portfolio with the appropriate ratio of each.

  10. Great idea! One of those things where it seems so obvious you can’t believe that it’s not been done already. In terms of a UK perspective, taxes on dividends are less of an issue (gains taxes are more painful). However, we’re still quite limited in ETF options. I’m still waiting for a true global all-cap that isn’t 15 basis points more than in the US (unfortunately we’re starting to get some of the ‘garbage’ ETFs over here instead).

    Thinking about other options, I came up with two. But being honest, I don’t think either works.

    First, is a pair of Factor ETFs. One Low Volatility; one Momentum. Generally speaking low volatility shares tend to have higher yields, whilst momentum shares by design will tend to have lower yields. Trouble is, it’s an active strategy so you can’t recreate the index using it.

    The second one I thought up was to use Sector ETFs. The benefit of this is that you can recreate the index. It won’t be perfect as even low yield industries will have some high dividend payers. In fact, after a quick eye-ball on Vanguard, I don’t think the yield spread is wide enough to make it viable (IT: 1.07% vs Utilities: 3.37%). Besides, re-balancing would be utterly painful!

    Anyways, thank again for the interesting post!

    1. Nice, thanks for sharing the UK experience! Every country has its own tax hacks.
      Agree: I think it’s best to stay with a passive to avoid turnover and fees.
      I thought about sector ETFs too. Fidelity has very low-cost ETFs. But the bottom 50% of cap weight (IT + Consumer Disc. + FInancials or Health) would have an average weighted yield of about 1.35%. Lower than the S&P but much higher than sorting by individual security yield. 🙂

  11. Vanguard Tax Managed Capital Appreciation takes the dividend down a bit to 1.66% (compared with 1.94% on the VOO) with a very similar return profile. Perhaps the algorithm that creates this lower yield can be tuned up a bit.

    The FTSE Social Index Fund carries a yield of 1.62%, in part due to the overweighting of some of the lower yield “growth” names. But it tracks somewhat further away away from VOO.

    Beyond these available potential solutions, I would worry that if you start splitting into multiple funds, per your suggestion, it will be easy for things to get out of whack and complicated, and tracking error will rear its ugly head.

    I am a big fan of ERN (!) but am not loving this idea.

    1. As I said earlier: 1.66% is still way too high to be considered a valuable tax hack. Same for 1.62. Much of that minimal tax gain will be eaten by a higher expense ratio. So, I’m not loving that approach. 🙂

      I’m less worried about the style drift (getting out of whack). Recall, that if you start with cap weights in the beginning, the cap weights stay intact within the two buckets. Only across the two buckets you might have to reshuffle a bit. But you can do that by, for example, redirecting some of your existing S&P500 index ETF dividends in the taxable account into the new low-div bucket to make up for the lower yield there.

  12. Very clever idea! I agree that there would be some details to work through (like minimizing the churn caused by companies changing their dividend rate), but it doesn’t seem insurmountable (e.g. institute a smoothing function that ensures that a company’s allocation won’t change by more than X percentage points in a given year).

    Such a strategy would of course have to compete with the standard dogma of using tax-advantaged space for bonds — while paying tax on qualified dividends is painful, paying tax on unqualified dividends seems even worse! This too seems like a small quibble, though, as I suspect there are relatively few people that fill their 401(k)s with *only* bonds (and of course, there are plenty of investors that for various reasons just don’t need/want a very high bond allocation).

    1. Thanks!
      Exactly! Most of us in the FIRE community have the problem of too high a percentage in tax-deferred accounts, so you shouldn’t hit that constraint. Especially in the accumulation phase, you shouldn’t have much in bonds anyway.

  13. Hey Ern,

    I’m my taxable account I have both vanguard develop (vea) and emerging (Vwo) market fund. I keep it at 85% develop 15% emerging as my foreign. For tax purposes is it best to have just total int market?

  14. Hi ERN! Long time listener, first time caller.

    It’s not exactly what you’re asking for here, but in Canada we have our own tax-hacked version fo index fund ETFs. They’re swap-based and the idea is they track an index+dividends, while keeping everything as capital gains. Ideal for taxable accounts. There is some (small) counterparty risk.

    See and

  15. There are Accumulating ETFs like the SWDA that do not distribute dividends. They’re generally domiciled in Ireland and traded in several non-US stock exchanges. Not sure how access and tax would work for US investors.

  16. Regarding your baseline presumptions to …”Hold the tax-inefficient asset classes (bonds, REITs, etc.) in tax-deferred accounts and equities in taxable accounts”.

    I discredit that common rule of thumb for Asset Location in the article posted at :
    The Bogglehead methodology for Asset Location you reference in a reply above also wrongly uses this ‘tax efficiency’ metric.

    The industry’s starts from a wrong understanding of where the benefits of tax shelters come from. If you don’t know the benefits, you cannot optimize them. I encourage you to take up my ’cause’ to re-educate the industry, or find errors in my math and logic. This issue has few knck-on impacts in the US, but has resulted in more than a dozen industry promoted – but wrong – pieces of advice in other countries.

    If you want to play with different variable inputs there are a number of tabs for Asset Location at

    1. I’ve thought about this topic as well. It’s not as clear-cut as the proponents of the “stocks in taxable, bonds in retirement accounts” always want to make it. Totally agree!
      But I also identified some logical erros in the calculations of the critics, example:
      I’ll have to go through your sheet first to see what’s going on and whether you make the same error as the WCI article. I can already tell you that your Excel calculation of the taxable account in tab “RRSP benefits” is incorrect. You have to split the dividend yield (tax drag every year at the marginal rate) vs. the capital gains (taxed only once in the end). So, the taxable account makes more than 6.8% after tax. You treat the taxable account as though it’s taxing the capital gains every single year, which is clearly unrealistic.

      But rest assured, I am thinking about this issue too and I’m planning a post on this topic when I find some more time later this year! 🙂

      1. I don’t accept your criticism of the opening tab at

        My math requires some assumed tax rate on profits in a Taxed account. This will always be the EFFECTIVE rate that incorporates the impact of capital gains deferral, and preferential rates, and possible tax credits, etc. You can start with the tab “Effective Capital Gains Rate” to find the equivalent yearly tax rate when gains are delayed many years. Then use the tab “Weighted Average” to find the tax rate for all the profit types generated by that asset.

        The tax rates used by the model for contributions and withdrawals are also EFFECTIVE rates that include the iimpact of income-tested benefits from other programs, and averages of multi-tax-brackets, etc.

        The math and logic hold true always. You just did not know how to convert a periodic tax payment into an equivalent yearly rate. I can answer any other ‘issues’ you believe you find.

        However it was the the Advisor Perspective article I was most interested in bringing to your attention.

        1. You can assume an “EFFECTIVE” tax rate all you want. I look up ACTUAL tax rates in the IRS code and work with them.
          And maybe because you’re from Canada you’re not aware of the subtleties of the US tax code. “Profits in taxabale account” are not taxed at one single fixed rate. Dividends and long-term capital gains are taxed at a certain rate. Bond Interest are taxed at another rate. Long-term capital gains are not taxed until realized, thus yet another separate calculation: you’d have to track your cost basis numbers along the way and figure out the tax. So, setting the tax rate a priori independent of the horizon is incorrect for stocks.

          In the advisor perspectives article, you make some of the same logical mistakes as WCI in his post. So, yeah, I will write a blog post on this topic soon. I got my notes and calculations ready, I “just” have to write it all up.

          1. No, the superficial differences in tax systems between Cda and US won’t make any difference.Neither would the British tax system make any difference. I tried in my response above to walk you though exactly how you get from the IRS tax rates for different types of income to the Effective tax rate that takes into account all the issues you raise.

            Use one tab of my spreadsheet to find the equivalent annualized rate that results from capital gains deferred for whatever period you like. Use that rate for use in the other tab of the spreadsheet where you add the other interest and dividend portions of profits (with their own rates) to find the weighted average tax rate for ALL types of profits from the asset.

            The end result is that the tax rate you should use to calculate the ‘tax efficiency’ metric that you want to use for Asset Location (Tax rate * Rate of return), is the same one you would use for the first tab of my spreadsheet that you object to. It is the equivalent annualized rate of tax taking into account all types of income and any deferral of capital gains.

            1. OK, gotcha, regarding the cap gains taxes. I still prefer to do the tax calculations the way I do it, not as the blended average, but it probably won’t make much of a difference. Understand now!
              But how did you get 7.7% after-tax return in the taxable account in the advisor perspectives article?

              1. The article states (in point #3) in brackets that I presumed cap gains never taxed, and a 2% dividend taxed at 15% for stocks with an 8% total return.
                For a $100 asset the yearly $tax would be 100 * 2% * 15% = $0.30
                The total return would be $100 * 8% = $8.00
                The after-tax return would be 8.00 – 0.30 = $7.70 on the $100 asset = 7.7%

                1. Ah, OK, gotcha. It’s a decent assumption for some investors (stay in brackets 1/2 and no state income tax). But for most people, I’d assume some tax rate >0 when withdrawing. But in your case this still works, because you give the equity in taxable account the best possible assumption and still show that sometimes it doesn’t work. With taxes on cap gains you’d find it even easier to show your results.
                  Anyways, interesting work! Great job!

                2. Hold On. You have missed the point completely. The argument of if / how Asset Location choices should be made, is determined by the math, by the process, and NOT by the particular variables chosen.

                  I come to absolutely no conclusions regarding the AL of generic stocks or bonds or anything. My input variables for the generic asset type will be different from yours, and different from the next guy’s. The reality that the outcome of the math will change with different variable inputs NEVER permits you to dis/ like the math itself just because you dis/ like the variables chosen. The reality of different outcomes means that no one should be creating generic AL rules in the first place. That was a huge point of my paper.

                  I am NOT trying to prove that stocks are better in tax shelters. In fact in the paper you see how that conclusion would be wrong for shorter time frames, and if there is rebalancing, and probably if there was a penalty from a higher w/d tax rate,

                  I AM proving in the article that none of the AL rules used by the industry are generically true. They are not wrong because of their chosen variables. They are wrong because their math, their process, is wrong.

                3. My conclusion is the same: There is no slam dunk on the AL decision. I don’t know why you keep carrying this on and on and on. I am the host here and will now shut down this discussion. I just don’t like to hear that I “completely missed the point” from someone who completely missed the point on the issue of parameter-dependence of the AL decision (tax rates, expected return parameters, horizon length parameter). And who completely missed the point of this particular post on the tax efficiency of ETFs. Thanks!

          2. Maybe if you gave me a specific situation I could walk you though the math. You need
            * the %return from each type of income the asset produces Eg 2% div, 6% cap gains.
            * the deferral period #yrs for triggering cap gains.
            * the IRS tax rate for each income-type

              1. But you failed to specify your capGain delay period …. and that is exactly the issue you use to refuse to accept an annualized tax rate. The WhiteCoat example used a 30 yr delay, so I will work with that. Below are the screen shots from the spreadsheet tabs you can use to calculate the effective tax rate.


                First find the cap gains rate = 7.8% See A
                Than add that to the 2% annual dividend income to get the 9.6% blended annualized tax rate. See B
                Then input the RofR and Tx% into the Cumulative Benefits tab to see that at your 30yr period, assuming no rebalancing, stocks are better inside shelters compared to bonds. See C
                The chart over time shows the cross over at year 25. See D

                And yes, the chart over different time spans is still valid for you even though your tax assumptions used a specific 30 yr capgain delay. It shows how the your tax-sheltering benefits would grows over time given your assumptions, not the benefits for all stock owners with all delay assumptions.

                Be clear that your objections on this thread so far have all been to do with the calculation of net benefits from tax shelters (specifically the input variables A and B) and nothing to do with the Asset Location issue (C and D).

    2. Aside from that: The bond vs. stock discussion had really nothing to do with this particluar post. This post is actually from the perspective of someone (like yours truly!!!) with a 100% stock portfolio who wonders how to shuffle the high-yield and low-yield stocks between the different account types.
      Even for the folks who do hold some bonds somewhere (whether taxable or IRA is irrelevant) you’ll still likely have enough stocks in your portfolio that you hold stocks in both taxable and tax-deferred accounts, and here again there’s a tax arbitrage if you could split up your S&P 500 into two separate ETFs with different dividend yields.

      So, again, I see the problem with the “rule of thumb” you mention, but whether or not you believe it, it has nothing to do with this particular blog post. Whether or not it’s true it says nothing about the validity of the point I’m making in my post here.

      1. I agree. It was only the comment I quoted that brought to my attention your acceptance of the ‘tax efficiency’ metric for Asset Location. I just used the opportunity to bring the AL issue to your attention as a possible future article ….. (obviously hoping you will promote my debunking of all the received wisdom of course).

  17. Hi Karsten — I don’t think I understand ETF pairs in this context. Wouldn’t the ETF provider just create four ETF’s, one with the lowest quartile dividends (0-25), one with the second quartile 25-50), third quartile (50-75), fourth (75-100)? Then people could choose their split by buying the same dollar value of each but putting them into each account? I think you end up with fewer ETF’s to cover more combinations this way.
    Going to quintiles or deciles would mean more ETF’s, and probably more expenses (stocks will move between segments from time to time).

Leave a Reply

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