Thursday, April 23, 2020

Asset Allocation: Should You Account for Taxes?

We can only buy food with after-tax money, so it might seem obvious that we should take into account taxes in any financial decision. However, Justin Bender, portfolio manager at PWL Capital, has some reasons why you might ignore income taxes when when calculating your portfolio’s asset allocation.

Justin has created a series of excellent articles going over a great many issues do-it-yourself (DIY) investors need to understand. The articles are organized as a series of portfolios with decreasing costs, but increasing complexity. The portfolio names are inspired by the comedy movie Spaceballs: Light, Ridiculous, Ludicrous, and Plaid. My focus here is on accounting for taxes in your asset allocation, but this is only a small part of the many useful ideas Justin explains in this series.

The main difference between the latter two portfolios in the series is that the Ludicrous portfolio ignores taxes when calculating asset allocation, and the Plaid portfolio takes taxes into account in what is called after-tax asset allocation (ATAA). The Ludicrous portfolio is fairly complex, and Plaid adds an additional layer of tax complexity. It’s certainly possible to create a simpler portfolio than Plaid while still retaining ATAA. The question we address here is whether you should want ATAA.

Let’s consider a simple example to illustrate ATAA. Suppose you have $100,000 worth of stocks in a TFSA, and $100,000 worth of bonds in an RRSP. Your asset allocation appears to be 50/50 between stocks and bonds. However, this is ignoring taxes. Suppose you expect to pay an average of 25% tax on RRSP withdrawals. Then your RRSP is only worth $75,000 to you after tax. You have $25,000 less saved than it appears, and your stock allocation is $100,000 out of a total of $175,000, or about 57%. So, your ATAA is about 57/43, and your portfolio is riskier than it appears.

Some might object that we can’t know our future tax rate on RRSP withdrawals, so they conclude that it’s best to ignore taxes. However, with before-tax asset allocation, you’re implicitly assuming that the tax rate will be zero, which is almost certainly very wrong. It’s far better to come up with a best guess, like the 25% in this example, than to use zero. There are other much more sensible arguments in favour of ignoring taxes in asset allocation than this one.

What difference does the asset allocation calculation method make?

One difference we’ve already seen is that it can mask your portfolio’s true risk level. Another is that it drives your asset location choices. The Ludicrous portfolio (that does not use ATAA) tends to fill RRSPs with bonds and leave stocks in taxable accounts, while the Plaid portfolio (that uses ATAA) tends to fill RRSPs with stocks and leave bonds in taxable accounts.

From the Ludicrous point of view, any gain in RRSPs will ultimately be heavily taxed, so we prefer to have stocks with their greater growth potential in taxable accounts. From the Plaid point of view, we know that the portion of RRSPs that really belong to us (the $75,000 in the example above) grows tax-free because the government portion ($25,000) grows to cover any taxes owning. So, we prefer to have stocks with their greater growth in RRSPs.

Which asset location method will give you the better retirement?

It depends. Justin says that the Plaid portfolio “is officially the most tax-efficient portfolio in the bunch.” If we control all portfolios to the same true risk level, “we find that the adjusted Ludicrous portfolio actually ends up with a lower after-tax portfolio value and return than any of our other portfolios.”

However, Ludicrous can make up for its shortcomings by taking on more risk. As we saw in the simple example above, not using ATAA leads to thinking the portfolio asset allocation is 50/50, when it’s really 57/43. By taking more stock risk, Ludicrous increases its expected returns. So, a riskier Ludicrous portfolio is expected to beat a lower-risk Plaid portfolio.

Of course, you could just change your Plaid asset allocation to 57/43 and get better returns than a Ludicrous portfolio at the same true risk level. So, the answer to the question of whether ATAA will give you a better retirement depends on your behaviour. Will you take more risk if you don’t use ATAA? If you use ATAA and put stocks in your RRSP, will you be more likely to lose your nerve and bail out of stocks when you see high volatility in your RRSP, even though part of these swings are really the government’s risk and not yours?

Reasons to ignore taxes when calculating asset allocation

1. Investor Behaviour

“The Ludicrous portfolio is the winner from a behavioural perspective, but a loser from a tax-efficiency perspective. The Plaid portfolio ... is the winner from a tax-efficiency perspective, but a loser from a behavioural perspective.”

Justin has extensive experience with his clients. They tend to view their RRSPs as entirely their own, and too much volatility in a stock-filled RRSP makes them nervous and prone to bailing out at a bad time.

The Ludicrous portfolio approach to asset location might even help some nervous investors who could benefit from more portfolio risk. Their portfolios appear less risky than they are, which helps calm them while the higher true risk level is better able to take them to their investment goals.

Fortunately, these behavioural concerns don’t apply to me. I know the government effectively owns a slice of my RRSP. I maintain a spreadsheet of my portfolio, where I make the after-tax view of all accounts prominent.

2. Regulation

Justin wrote that the Plaid portfolio has the disadvantage of “regulatory constraints due to higher before-tax equity risk.” So, it appears that regulators require Justin to measure portfolio risk on a before-tax basis. This is a major constraint for a professional portfolio manager, but as a DIY investor, it doesn’t concern me.

3. OAS claw-back

With stocks in your RRSP, you could end up with a large RRIF and be subject to OAS claw-back from large future RRIF withdrawals. This certainly has the potential to hurt portfolios with RRSPs full of stocks, but I think it’s likely to just shrink the advantage of keeping stocks in RRSPs rather than flip the balance to give the edge to keeping stocks in taxable accounts.

In simulations of my own portfolio that take into account OAS claw-back, the clear winner was to use ATAA and fill RRSPs with stocks. However, this only applies to my situation; we don’t know for certain about other circumstances.

One mitigation method I’m using against high tax rates on future RRIF withdrawals is to make small RRSP withdrawals each year now that I’m retired but nowhere near age 71 when RRIF withdrawals are forced. These lightly-taxed withdrawals have the disadvantage of reducing future tax-free growth in my RRSP, but have the greater advantage of reducing high taxes and OAS claw-back on future large RRIF withdrawals. Again, the size of these small withdrawals that gives lifetime tax advantages is highly specific to my situation.

This is a complex area, but I’m skeptical that it’s common for OAS claw-back to be likely to tip the scale in favour of filling RRSPs with bonds and taxable accounts with stocks. There might be cases where it helps to put Canadian stocks in taxable accounts to take advantage of the dividend tax credit, but it’s tough to beat tax-free stock growth in RRSPs.

4. Simplicity

There is a lot of advantage to keeping a portfolio simple. It’s amazing how even a simple-looking starting plan can become complex when you apply it to a real portfolio.

Justin’s Ludicrous portfolio is quite complex, and Plaid just layers on some more tax complexity. My own portfolio is simpler than Justin’s Plaid portfolio. It’s possible to embrace ATAA and its associated asset-location strategy and still simplify other aspects of the portfolio.

The main challenges ATAA brings are deciding on estimates of future taxes, discounting each account type with these tax rates, and taking any after-tax rebalancing amounts and calculating the before-tax amounts before placing trades. I certainly wouldn’t want to do all this by hand. That’s why I have all these calculations built into a spreadsheet. Building such a spreadsheet isn’t for everyone, but now that mine is done, my portfolio management is easy.

A very simple way to get the advantages of ATAA for the many investors who run their portfolios without worrying much about rebalancing or exact asset allocation percentages is to just use the asset location rules of the Plaid portfolio. This mostly means keeping bonds out of RRSPs. This only works as long as they remember that a slice of their RRSPs really belong to the government and they don’t load up on too many bonds.

Most DIY investors probably shouldn’t look past Justin’s Light portfolio, particularly if they have less than about $250,000 and little in taxable accounts.

5. Industry Bias

This wouldn’t apply to Justin, but the investment industry is much more concerned about its own revenues than whether their clients meet their investment goals. This is hardly surprising. After all, most people are much more concerned about their own salary than their employer’s goals. It’s not a crime to care more about yourself than others, as long as you still treat others, such as your clients or employer, fairly.

The investment industry makes the same money on RRSP assets as it makes on assets in TFSAs or taxable accounts. There may be some differences in administration costs, but they are small compared to the differences that income tax in different account types makes to investors. It’s hardly surprising that the bulk of the investment industry wouldn’t care much about investors’ taxes, particularly if almost all investors don’t understand these issues.

6. Momentum

Different ways of doing things always look a lot harder than just continuing to do what you’ve always done. The investment industry isn’t looking for make-work projects, and adopting ATAA methods is unlikely to make them more money. This might apply to Justin least given that he did so much work on the many details of his Plaid portfolio.


Justin’s conclusion is that “I personally use [the Ludicrous] asset location strategy in client accounts, mainly because the Plaid asset location strategy is not practical.” He cites investor behaviour, regulation, complexity, and unknown future tax rates as reasons for this impracticality.

The investor behaviour concern doesn’t apply to me. I’ve lived through stock market crashes in 2000, 2008, and 2020 without any panic selling. Justin’s regulatory constraints don’t apply to me either as a DIY investor.

The complexity of taking full account of taxes was a concern for me before I automated most of my portfolio decisions and actions in a spreadsheet. I rarely have to look at it, a script emails me when I need to do something, and it calculates trade amounts for me. I also combat complexity by making other aspects of my portfolio simpler than Justin’s Plaid portfolio.

As for unknown future tax rates, ignoring taxes in calculating asset allocation is the same as assuming taxes are zero. This is further off the mark than any guess at future tax rates would be. It’s better to make your best guess than forget taxes entirely.

Overall, Justin’s reasons for using his Ludicrous portfolio are sensible on balance given his position. His clients’ lack of understanding of these issues as well as regulatory constraints make his decision clear. However, most of his reasons don’t apply to me, so I use after-tax asset allocation in running my own portfolio.


  1. Great explanation!

    I don't quite see why professionals would put clients in the "ludicrous" type portfolio (bonds in RRSPs) unless they're really comfortable with the effective asset allocation. IMHO if you're not going to do the ATAA, why not just stick with the approach of having the same asset allocation in each account, where things are clear?

    1. Potato: Thanks!

      I think the answer comes from a number of traits that are typical of a professional's clients but don't apply to me. Suppose you can sort out a client's emotional risk tolerance, and that it's based on before-tax asset allocation because the client thinks his RRSP is all his. Suppose further that the client really would be better off with more risk to increase the odds of a good outcome for retirement. Then the Ludicrous portfolio really does maximize returns under these peculiar circumstances.

      My own approach with friends and family is to try to explain important concepts like the government owning a slice of the RRSP so that I can better optimize after-tax returns. But, since I'm not trying to make my living at this, I'm perfectly content to be "fired" by friends and family, not that this has happened yet.

    2. Yeah, that point of "tricking" clients is an important one (and a point I've made myself before)...

    3. Potato: Another possible explanation of the status quo is that regulations drive everything. Risk is measures pre-tax by mandate, and if you want to survive as a pro you have to then maximize expected returns subject to this constraint you can't change. With this theory, all other explanations (behavioural concerns, etc.) are simply "stated reasons" as opposed to the real regulatory reason.

  2. That's a very interesting piece too, that I hadn't considered before, and a great addition to the conversation!

  3. Thank you for your blog. After 20 Years, alas you and Justin have answered my questions. I have asked numerous people, financial planners, advisors, bank portfolio managers, none of them would use ATAA. It's so obvious what's in our RRSP does not all belong to us. Thus, our net worth are obviously less than what's shown in my bank website front page. I only this week encountered Justin's portfolio articles and I am reading your blog after you had a link back to your blog. Then I noticed that these articles and blog posts are very recent, in the past week.
    Is this a recent discussion by planners? I wonder if this will be a popular topic, especially amongst DIYers. I have lots of reading to do and catch up as I am out of date, lots of learning to do.

    1. Hi Mike,

      I've certainly been thinking about ATAA for quite a while; I can't speak for anyone else.

      I agree it's obvious that our RRSPs don't belong entirely to us. I'm not holding my breath for advisors to take this into account in how they run portfolios.

  4. Hi Michael,

    Great discussion of Justin's Portfolio's and the reason he does not use the Plaid Portfolio with his clients. The Plaid portfolio has really opened my eyes to the need to rethink the traditional portfolio asset location strategies. For the past 10 years or so I've been running a Ludicrous portfolio based on the ATAA thinking this was the way to go. I, as you, have a nice spreadsheet set up to calculate AT for all investments in all accounts including Capital gains taxes owed in taxable accounts.

    For me to change to the Plaid portfolio is relatively simple for me logically, but with 8 accounts between my wife and I (2 taxable, 2 TFSA, 2 rrsp, 1 spousal rrsp and 1 LIRA) technically it will take quite a few trades at $10.

    I'm just into retirement will only be withdrawing from now on, excluding TFSA deposits from the withdrawals. The Plaid portfolio actually makes my situation simpler in that right now I'd only have Bonds in Taxable account, and only Cdn equities in the TFSA. I will then have a little bit of everything in the RRSP. For now, rebalancing will be very easy.

    As I start to withdraw funds I see that the rebalancing could get more complicated as my initial withdrawals are coming from taxable accounts thus reducing overall bond totals...leading to have to shift more bond % to the RRSP.

    Can you comment on any of this from your experience? Personally I don't think this will be any harder than what I had been doing before retirement on an ATAA approach which was using deposits instead of withdrawals.

    Thanks for the article. It was very helpful for me.


    1. Hi Ed,

      I'm glad my explanation helped you. I never ran my portfolio using Justin's Ludicrous approach, so I'm not sure how much work it would be to switch to Plaid. It would depend on a number of factors such as the number of different ETFs you own and whether you have significant capital gains built up in non-registered accounts. You might choose to spread out a full switch over time if you'd end up recognizing capital gains.

      One small way to save on trading costs is to combine a withdrawal with a switch. Suppose you want to change one ETF for another in your RRSP, and you want to sell something in your RRSP to raise cash for a planned withdrawal. You might be able to combine these steps to make fewer total trades than if you do these two steps separately.

      The only part of making withdrawals that I find challenging is dealing with accumulated capital gains in a non-registered account. I find this fairly easy to deal with in my Plaid-like plan, but I imagine it would be harder with Ludicrous.

    2. Hi Michael,

      Thanks for the quick reply. I've been running the Ludicrous ATAA portfolio for the past 10 years so the only difference going to the Plaid would be the asset location changes, I'm totally familiar with running my allocation on an after-tax basis. Yes, I understand the capital gains problems but I'm lucky??? to have some larger capital losses to use against them. Long story.

      I'm only questioning the move to Plaid over a move to a Ridculous as I still have large capital losses that I won't be able to use if I go the Plaid route.

      Based on my account sizes, etc. in the Plaid version I'd have only bonds in my taxable account leaving no way to use up the capital loss that I have.

      As there is only a minor difference ~ 23 bps to going the Plaid route then maybe it's worth using up my capital loss.

      I think i have a rebalancing method that works and minimizes the number of trades as I never do a full rebalance. I monitor the entire AT asset class values and when they fall out of the range I have set, I rebalance the account that will provide the largest benefit towards the rebalancing and do that one. If that triggers another rebalance on another asset class (usually cash) then I do that rebalance similarily. So sometimes I rebalance back to the desired allocation but most times I just move closer the desirred allocation.

      What you think? Plaid or Ridiculous?

    3. Hi Ed,

      As it happens, I also have some old capital losses that will help with future capital gains. However, my stocks don't all fit in our RRSPs and TFSAs, so I own some stocks in my non-registered account. That will use up my capital losses eventually.

      If the 23 bps difference you mentioned is before taking into account your capital loss, have you attempted to quantify whether the capital loss will make up this difference? It's possible the best path is some hybrid of Plaid and Ridiculous.

      I'm afraid I just can't tell what approach is best for you without a lot more information, and even if I had it, I likely couldn't do the "perfect" analysis.

      Something to keep in mind, though, is that it's better to settle on an approach that's close to best than it is to fret over small differences between good portfolios.

  5. The 23 bps / yr difference is what Ben Felix has mentioned as the only real difference in optimizing the asset location over a Ridiculous Portfolio. Unless I misunderstood.

    "Something to keep in mind, though, is that it's better to settle on an approach that's close to best than it is to fret over small differences between good portfolios."

    Excellent reminder.

    1. Hi Ed,

      Keep in mind that Ben Felix's entire discussion was based on before-tax asset allocation. With ATAA, things would be different.

      Rather than hold slavishly to the Plaid rules, I run my portfolio with a bias to holding stocks in my RRSP and TFSA, with U.S.-listed ETFs in my RRSP. If I'm rebalancing and it happens to be easier to put some money into a bond ETF in my RRSP, I do so. Then with future rebalancing, I might sell it again.

      As long as I have the overall calculation of my ATAA across all my accounts automated, I don't fret about minor deviations from my asset location rules.

  6. Hi Michael,

    I really appreciate your article. Thank you. I've read through Ben's series of 'Spaceballs' portfolios. Unless I missed it, Ben examples do not seem to account for the tax liability from capital gains in taxable accounts. Does your model account for that as well? Extrapolating from the logic, it should based on a combo of accrued gains plus expected long-term capital growth rate, right?


    1. Hi RR,

      If I'm not mistaken, Justin Bender treats all taxes as being paid immediately, whether RRSP income or capital gains. This is pessimistic in a long-term scenario.

      Yes, I do account for capital gains taxes. I do this by first estimating how much my capital will grow (taking into account the fact that I will selling it off over time). Suppose I expect the total capital gains over the decades to be roughly equal to my current capital level. Then half my capital will end up as capital gains, and with the 50% inclusion rate, one-quarter of my capital will end up as regular income. So, then I'd use one-quarter of the tax rate I pay on regular income as the discount rate for this capital. (Note that this is my average tax rate, not my marginal tax rate.) Of course, I'd have to add in taxes on dividends as well.

      I'm not too concerned with getting this figure exactly right. What matters is getting closer than the 0% estimate implicit in ignoring taxes altogether.

    2. Thanks for clarifying.