Blessed by the Potato

So you liked the idea of all-in-one funds when they came out. You like-liked them. Finally, you love the idea of all-in-one ETFs and are going all-in. You’ve rounded off your asset allocation, and stuck it all in a single ETF. All new money that you save, you also stick in this one ETF.

Then about 20 years later*, your risk tolerance has shifted enough that a different all-in-one ETF is appropriate for you (e.g., VGRO now becomes VBAL).

How do you switch? What are the costs and trade-offs? This is one of the remaining sticking points for a lot of potential all-in-one investors (well, that and taming the need to tinker).

RRSP/TFSA: there are no tax issues to worry about, just go ahead and do it in one go. Costs a commission to sell (and depending on your brokerage, one to buy the new fund) — not material. Thanks for visiting, enjoy your all-in-one fund without any worries, you can skip the rest of the post. (And if you’re not sure how to buy an all-in-one ETF and invest and all that, check out the course to teach you just that).

Non-registered: it gets more complicated, as you might expect. When you sell a fund you’ll have to realize the gains (and thus pay tax). Turning over your entire portfolio at once might have some fairly large tax consequences. You can take an alternative approach, like buying a bond fund at some point (perhaps selling a bit of your all-in-one fund) to make a two-fund portfolio later.

That looks like you’ll trade simplicity today for a bit of a headache later, in either complexity at some point or a tax bill. That’s likely still a good move, though you’ll want to go in with eyes open on what the cost of maintaining simplicity is. Fine, let’s math it out to get an estimate.

TL;DR: doing a big switch-over in your non-registered account (like you might in a TFSA/RRSP) may end up adding something like double-digit basis points to your equivalent MER costs in realizing the taxable gains, which may push you towards more complexity later — which to be fair is a problem for future-you to deal with, and is still better than complexity all the way along, though it means at some point you’ll be pushed to learn how to rebalance).

Disclaimer: I did the spreadsheet and wrote the post on the train back from Montreal, so there may be errors. The cost ended up about double what I had guessed it would be, which may just mean the math is a caution to my intuition, and that people who want simplicity in a non-registered account might have to suck it up or look to the e-series funds, or it may be a sign there is indeed an error in here.

How big is your non-registered? Bear in mind that the benefit of simplicity will accrue to all your accounts, though the tax cost may only hit your non-registered. So if you’re planning on heading into retirement with a million-dollar portfolio, odds are that you may have most of that in registered accounts, so even a “big” tax hit to your non-registered may not be big in the context of the work you’re saving by using all-in-one funds for several decades in all your accounts.

Unfortunately, the switching cost is not a simple percentage that you can directly compare to an MER — it’s going to depend on future capital gains, your tax rates, etc. And looking at just the cost of switching an all-in-one fund will over-state things a bit, as you would still have to pay gains to rebalance a 3- or 4-fund portfolio when the time comes (or to sell it and use the money to live off of).

So how much is this going to cost? Let’s use a specific example to try to put some numbers to it. Let’s say you throw $5,000/yr into a non-registered account (remember, this is on top of what you’re saving in your TFSA/RRSP) and buy VGRO. Let’s say the market is well-behaved for the sake of our spreadsheets, and that investment grows at 6% nominal (your real return [which matters for most planning purposes] will be lower due to inflation, but for tax purposes nominal returns matter, and in this case this doesn’t include dividend/interest distributions — those we’ll say are part of our $5,000 annual [nominal] contribution).

20 years later, your risk tolerance has shifted enough that you feel VBAL is a more appropriate choice, and you have to switch over. Your VGRO at that point is worth ~$189k, with a cost basis of $100k. Switching in one big go means realizing a gain of ~$89k, so ~$45k gets added to your taxable income. That feels ouchy, though you’ve reset your cost basis so you’ll pay less tax when you eventually sell the funds for living expenses. Also remember that this is 20 years in the future with nominal dollars, so it’s not quite as bad.

If you have an income of $60k in today’s dollars (~$90k in 2039 dollars), the tax bill of the big bang approach would be (very roughly) $13k (in future dollars). That’s not all an incremental cost of using all-in-one ETFs though, as those gains have to be paid eventually if you don’t have a big switch.

There are some other options to spread the cost out and reduce it a bit, with just a bit of added complexity:

Option 1: spread it over multiple years. One of the issues with the all-in-one big-bang switchover is that it could add a lot of taxable income in a single swoop, which would then run up the marginal tax rates to add insult to injury (in this example, up to the ~38% bracket for someone normally in the ~30% one). You can pay less tax by just spreading the hit over a few years. That’s a bit of extra complexity, but not much. You could do it over a few years, which requires fiddling with your portfolio a few times, but timed right you can spend just a few days focusing on the trade to spread the tax hit over two years (by selling half in the last few trading days of December, then the other half in the first few trading days of the new year).

For a two-year hit, you would save about $600 in taxes as all the capital gain fits into the bracket you’re already in (so there’s also no added benefit to further spreading it out — other situations, like having more gains, may change that).

Option 2: break the simplicity. You can still get ~20 years of simplicity with an all-in-one fund, then sell just enough to buy enough of a bond fund to get you back to your desired asset allocation. You’ll be in for more complexity for the rest of your investing days, but by that point it may be worth it.

Math of that: Sell 25%** of your VGRO, buy a bond fund, and get your 60% overall stock exposure. Then you’d be paying about $3.3k on the smaller realized gain portion (again, nominal dollars).

Our counterfactual is investing in a 4-fund portfolio. In that case, you may be able to keep rebalancing by just buying bond funds with your new money each contribution. (We’ll ignore the potential differences in return from having progressively more/less in equities at different points than the all-in-one discrete route — though to be fair, that should also be open as part of option 2, where you just buy bond funds along the way and break the simplicity of the all-in-one fund earlier without having to realize any gains).

Then you start selling down, perhaps at a lower base income. For simplicity of modelling, let’s say that you want to sell enough to generate $10,000/yr in after-tax income (in 2039 dollars) for 20 years and it’s all in the lowest tax bracket. But because the cost bases will be different, we’ll have to sell different amounts of the portfolio. That is, the now-VBAL portfolio will need us to sell a bit less to get the income we need because with a freshly reset cost basis, it will start off almost all tax-free (but because you paid more in taxes in 2039, you’ll also have less to start with). Let’s also throw 5% nominal growth on that portfolio just to make my job modelling in Excel harder.

Big bang VGRO: take off $13.9k in taxes to start, then model growth, sell-off, and finally final value. In year 1 you have to sell $10,096 to get $10,000 after-tax (just one year of 5% growth to pay taxes on), and by the end you’ve collected $10,000 x 20 after-tax, and have $138.2k left (of which $89.3k is unrealized gains from the new growth).

Two-year VGRO: take off slightly less in taxes ($13.3k), so it comes out pretty similar: an ending value of $139.9k, of which $90.3k is unrealized gains.

Hybrid: Take off $3.3k in taxes for selling just a quarter of your VGRO, which leaves you with a higher unrealized gain as you start to withdraw your funds. To get $10k after-tax, you have to pull out $11.0k in the first year. By the end, the portfolio is worth $145.4k, of which $115.8k is unrealized gains.

4-fund alternative: here we’re assuming that you’ve never had to realize any gains, yet somehow made the same return as VGRO/VBAL. So when you start pulling your $10k/yr income out, you have to sell $11.1k in the first year pre-tax to get that. After 20 years of that, you’ll have $151.7k left, of which $123.3k is unrealized gains.

Then what happens with those unrealized gains can matter a bit. If you just discount them by 10% (half the lowest tax rate), then we find the after-tax value of the portfolio is $10.1k lower for the all-in-one user — the tax hit and lack of continued compounding does hurt, though some of the tax hit at the big-bang rebalance will be made up in later years as the unrealized gains did eventually have to be realized to generate income out of the portfolio. And if this point represented death where all gains are realized at once, then the bigger unrealized gains may push your estate up into a higher bracket, closing the gap further.

Again, it’s hard to put this into MER terms. An ending value that’s something like 7% lower after 40 years would be like 18 bp more in costs, which sounds higher than I was expecting when I started doing the math, honestly. Of course, I may be being overly fair to the 4-fund alternative’s ability to match VGRO’s growth without any selling to rebalance along the way — there likely would be some gains realized along the way. If you accept a 2-fund approach in the future (only selling a portion to buy bonds), the hit would be more equivalent to 10 bp. And you also have to consider how much of your portfolio is non-registered — here I’m considering just the non-registered portion, but if most of your money is in registered accounts (as most Canadians can expect to be the case) and you want to consider spreading that tax hit over a larger denominator, the MER-equivalent type hit might be less than a third of this.

As always, balance any costs against the value: the extra simplicity of an all-in-one fund is practically a no-brainer when the MER is only about 10-12 bp higher than a 4-fund portfolio. If it’s more like 30 bp when you add in the tax hit of a big-bang rebalance, then you may have to think about it a bit more (esp. when you can start getting into TD e-series funds at about that level and automate your purchases — though the all-in-one funds would still work in your registered accounts).

Edit May 27: to clarify, the tax hit is coming from a few sources. First is just realizing the gain and missing out on the ability to continue to defer taxes and compound the growth. Next is the fact that the gain is at a higher tax rate (I’ve assumed here that the hit comes while working, at a minimum of 30% tax rate, and higher for the “big bang” switch, while in retirement gains are realized at a 20% tax rate). So there’s a few factors mixed together in this scenario.

* – Why 20 years in all the examples? If we take the basic rule-of-thumb of basing your asset allocation on your age, that implies that you shift your allocation by about a percentage point per year, and the different all-in-one funds shift by 20 percentage points… Your own case will be your own case, though.

** – Why not 20%, which may be the intuitive first-guess answer? Because you’re selling part of your bonds, too. To get to an overall 60/40 split, you’ll need 75% in VGRO and 25% in straight bonds.