When you invest in a non-registered account, you need to be concerned about more than just your funds’ performance: you also need to know how much of your return will be eaten up by taxes. Unfortunately, while regulators are strict about the way ETFs and mutual funds report performance, fund companies in Canada have no obligation to estimate after-tax returns—something that’s been required in the US since 2001.
To help address this problem, Justin Bender spent the last several months creating a calculator for estimating the after-tax returns on Canadian ETFs. He was inspired by Morningstar’s US methodology, but he made many significant changes to adapt it for Canada. The new methodology is fully explained in our latest white paper, After-Tax Returns: How to estimate the impact of taxes on ETF performance. We have also made our spreadsheet available for free download so DIY investors can experiment on their own. (The spreadsheet is protected so the formulas cannot be altered. However, we have included detailed descriptions of these formulas in the appendix to the white paper.)
The methodology is quite complex, but here’s an overview in plain English:
- We begin by recording the ex-dividend dates for all the cash distributions an ETF made during the period being considered.
- We then look up the fund’s net asset value (NAV) on each of the ex-dividend dates, as well as the NAV for the start and end dates of each performance period, and enter these into the spreadsheet.
- Using the reports available at the CDS Innovations site (which readers will remember from our earlier white paper on calculating your adjusted cost base with ETFs), we record the amounts of all the cash distributions and their character (Canadian and foreign dividends, other income, return of capital or capital gains), as well as any foreign tax paid and any reinvested distributions.
- We next make an adjustment regarding foreign withholding taxes. When ETFs report their before-tax returns, they presume these withholding taxes are lost. However, the final level of foreign withholding taxes (which is designated as “foreign tax paid” in box 34 of your T3 slip) is generally recoverable when the ETF is held in a taxable account. So we felt it was necessary to add this back.
- The spreadsheet then assumes each distribution was immediately taxed at the highest combined federal/provincial rate that applied in that year: for example, interest from bond ETFs in 2013 are assumed to be taxed at 50%, while non-eligible Canadian dividends are taxed at 39.15%. These were the rates that applied that year in Nova Scotia and Manitoba, respectively, and provide a worst-case scenario for taxable investors in Canada.
- Finally, after the tax has been deducted and the adjustments made, we assume what’s left of the distribution is reinvested in the ETF at the price on the ex-dividend date.
After explaining the methodology and providing instructions for using the spreadsheet, the white paper presents our calculations for 13 ETFs in various asset classes. We present both the before-tax returns (adjusted to add back recoverable foreign tax paid) and the after-tax performance as estimated by our calculator. We also present the fund’s tax cost ratio, a calculation designed to capture the amount of a fund’s distributions that’s lost to taxes. (This ratio was also created by Morningstar and is discussed in more detail in the white paper.)
Here are some of the more interesting findings:
Bonds. Both Justin and I have written about the serious tax inefficiencies of traditional bond ETFs, and this point is driven home with a glance at their tax cost ratios. (The higher the ratio, the less tax-efficient the fund.) Real return bonds and high-yield bonds took the biggest hit, but even short-term and broad-based bond funds were terribly tax-inefficient. The lesson: hold traditional bond ETFs in an RRSP and look for other fixed options in a taxable account, such as GICs or ETFs that hold discount bonds.
REITs. Income investors love REITs, but a large share of their distributions are fully taxable. In 2013, for example, the iShares S&P/TSX Capped REIT (XRE) had a pre-tax return of about –6%, but we estimate the after-tax return at –8.56%, for a very high tax cost ratio of 2.72%. Over five years, both Canadian and global REITs had tax cost ratios over 1.6%, which would cause a huge drag on returns. The lesson here is that if you can’t hold REITs in a tax-sheltered account you may not want to hold them at all.
Dividends. While Canadian dividends are often touted as tax-friendly income, our analysis showed that higher yields also mean higher taxes. The iShares S&P/TSX Canadian Preferred Share (CPD) and the iShares Canadian Select Dividend (XDV) both showed much higher tax cost ratios than the iShares Core S&P/TSX Capped Composite (XIC). That means that over a period where these funds delivered the same before-tax returns, taxable investors in XIC would keep more for themselves. Granted, XDV outperformed quite dramatically over the last five years: investors will need to decide for themselves whether that should be expected in the future.
Download the white paper and the calculator and let us know what you think. Some investors will quibble with the methodology, but we hope it will at least start a discussion on this important and neglected part of performance reporting.
Thanks for the post. It would be interesting to see the tax cost ratio at lower personal marginal tax rates. It is important to see the effects of taxation, but for a lower income investor the impacts won’t be as extreme and it would likely change some investment decisions.
Great analysis, thank you. For those of us who are easily challenged by numbers, and are daunted by the prospects of actually finding the data to populate the spreadsheet, even if we understand it, would it be possible to provide a much simplified analysis of some of the most popular ETF’s please?
The analysis could assume say $10,000 invested in ETF a, b, c, d etc, at a common date each year, at a marginal tax rate x, and sold 12 mos later of the next year.
A bar chart would quickly compare fund gross and net returns within fund families, and by fund categories, by time period.
Or even better create an online tool that would permit investors to produce their own comparisons with funds they are interested in, over the time lines they prefer.
Really good! First of all, thanks for taking the time to write these. Really enjoy reading these whitepapers. Second, agree totally that this is an often-neglected topic.
I think this will serve to highlight the difference between “efficiency of distributions” versus “total tax cost”. For example, everyone generally knows that bonds funds, like XSB, are “inefficient” in taxable accounts, losing nearly or sometimes more than 50% of their return to taxes. Likewise, most generally believe that Canadian Eligible Dividends are more efficient, so holding a fund like XDV in a taxable account might make sense because a lower portion of the dividends are lost. Your paper does a great job of showing that, in the case of the two funds above, it’s pretty much a wash – the much higher yield of XDV wipes out the efficiency advantage, such that the total tax loss is almost identical.
I know you focused on pre-liqudation to keep it simple, but what I think needs to be baked-in as well is “expected capital appreciation”. If you bake that in, then you have to believe that, for example, XDV will produce higher capital gains than XSB in the long run (no one can see the future, but if you cannot subscribe to the idea that stocks should outperform bonds over time, then really the whole risk/return model and modern portfolio theory needs to be thrown in the trash). Therefore one should absolutely prefer to hold XSB in an unregistered account rather than XDV – once capital gains are baked in, XDV should be the obvious loser.
I once tried to construct a similar Excel test that would do what you have done and also include estimated capital gains as well as allowing you to vary the marginal tax rate. It was more broad, but at least 2 decimal places less accurate than what you have done here. :) I assume that all equity markets returned roughly equally over time, and that return was somewhat higher than bond returns in the long run. (Again, nothing certain, but if you are reading this site than I think you have to at least buy into that) The surprising, if coarse, conclusion was that, depending on those future capital gains and marginal tax rates, holding fixed income in an unregistered account is may be not as bad an idea as conventional wisdom suggests. Looking at some of the other FI options as you suggest makes it even better.
@Willy: Glad you liked the paper. I would challenge you on one of your points, however: “One should absolutely prefer to hold XSB in an unregistered account rather than XDV – once capital gains are baked in, XDV should be the obvious loser.” If all you’re looking at is the dollar amount you’d pay in income taxes, then I suppose it’s true that XDV “loses” to XSB in a taxable account. But that’s ignoring the bigger picture. When you consider the total after-tax return on a diversified portfolio spread across RRSPs and taxable accounts, an ETF of premium short-term bonds is a terrible choice in a non-registered account and Canadian equities are an excellent choice. The whole point of the tax cost ratio is to consider the impact of taxes on a relative basis, not on an absolute one.
@CCP: Thanks for the feedback. Not sure I follow, though? I guess what I am questioning is, why is it you suggest that it’s the relative taxes (ie “tax efficiency”) that is more important than the absolute taxes (“total tax cost”)? The former is certainly informative, but if I’m an investor, my goal is to keep the most money possible, isn’t it? And so it’s the latter on which I should base my decisions? I’d rather keep only 50% of an investment with a long-term, after (all) tax rate of return of 3%, than keep 75% of an investment with a long-term, after (all) tax rate of return of 8%. The second is “more efficient”, yes, but the first actually leaves more money in my pocket.
But how can you be sure which asset will return 3% and which 8%? I can’t, obviously. But as I noted above, do you not at least have to accept that, over time, equities ought to produce higher returns than bonds? Else a good amount of what’s on this site becomes upside down? ;) My assertion is that if you count on this – SOME general, long-term outperformance of equities – and you include the cap gains that you will inevitably have to pay on those outperforming equities – it is not at all hard, given very reasonable assumptions, I think, to suggest that you could actually end up after 30-40yrs with more after-tax money to your name had you held fixed-income in the taxable account.
(Realizing I’m tangenting here as your paper clearly and intentionally ignored post-liquidation taxes and focused just on “annual” tax cost due to distributions, so not really in scope. My point though is just that if you did or had included those things, I think the story would be different. It would show that the guy who held his XSB position in his unregistered account would have done better than the guy who held his XIC in unregistered instead – or pretty darn close)
@Willy: This is a huge topic with a lot of unknowns (future asset class returns, future tax rates, withdrawal strategies, etc.) and no hard-and-fast answers. But I’ll refer you to our previous white paper on asset location, which address many of the questions:
@CCP: No argument there. Most definitely. I’ve read that other whitepaper and agree with the arguments within. And for the record, I personally hold my XIC position in my unregistered account today, and my XBB position in my RRSP. So definitely playing Devil’s Advocate here! My point is only that while the pre-liquidation “tax efficiency” can be defined and “predicted” to pretty high accuracy (as you have), the post-liquidation, all-in tax rate (which depends heavily on future returns) is much harder to define, but is actually the ultimate determinant of how much money is left in your hands vs. the governments. So while the former clearly indicates one direction as best, I’m not sure the reality of it is that cut & dry. Do apologize though for straying a bit outside the scope of your paper… cheers!
How do tax-friendly ETFs such as HXT and HXS compare to their counterparts (XIU and VFV)? (I think that HXT has no swap fee and that HXS has a 0.30 swap fee in addition to the MER).
And are the potential tax savings from these tax-friendly ETFs worth the increased counterparty risk and the decreased diversification when compared to XIC and VUN?
How would Mawer Tax Effective Balanced Fund 105 compare with Berkshire Hathaway ‘B’ stock (no dividend – only capital gains when sell) in a non-registered account?
Also, how would you report Mawer vs. Berkshire on your CRA? Both are long term holds (buy and hold).
@Essy: If you hold a mutual fund or a stock in a taxable account you’ll get a T-slip at the end of the year with the details about the distributions (interest, dividends, etc.) and you report these on your tax return. If you sell shares and incur a capital gain or loss, you’ll get a gain/loss report from your brokerage at the end of the year also.
@Jeff: I’ve written a lot about these swap-based ETFs:
Dan, I have read your blog faithfully for years and want to let you know that the blog’s revised format is much more user friendly. I recently wanted more specific info on preferred shares ETFs and your search device along with “related articles” got me there faster than the old website design. Thanks for all you do for us
Just wanted to say that this article is a real tour-de-force. Exceptional work guys — it’s rare to see information this well prepared.
@cmj: Many thanks for the feedback, and glad you like the new design!
@Chris: We appreciate the kind words. Justin worked extremely hard on this one and we’re hoping the industry will take some notice.
PWL white papers are so good I forward them to others.
I am glad you have such an academic approach and get into the level of detail required because too much financial planning seems to be reactive and incremental, sometimes motivated by sales considerations, rather than based on sound research such as this.
Investors and advisors often devote much time and energy on the prospects for markets to strategically attain “alpha” and do not focus enough on the things we actually have some control over – savings rate, compounding time, fees, asset allocation and in this case minimizing taxes.
@Andrew: Many thanks for the compliment. Probably the most important thing I have learned from Justin is the importance of focusing on things within our control, and the importance of analyzing the issues thoroughly so you can quantify them. It’s a challenge to present these ideas to the investing public, because it’s not always a compelling story that people want to hear, so it’s great to hear you appreciate it.
Dan, I want to invest in S&P 500 as a Canadian ETF, not hedged. ZSP.U is interesting since it is in $US funds. I have quite a bit of US cash and didn’t want to use it on US exchange because of watching the amounts of foreign investments/estate and Form T1135.
What I don’t understand is on Friday when the DOW closed up 208 points and the Canadian dollar was down .69 to 88.94 why would ZSP.U be down .28%? ZSP not hedged is up 2.17% which makes sense since DOW up and US currency up.
>> “… on Friday when the DOW closed up 208 points …”
ZSP.U doesn’t track the Dow. The S&P 500 was up 1.1%.
>> “the Canadian dollar was down .69 to 88.94”
ZSP.U trades in USD so changes in the price of CAD have no effect on it. The effect of currency changes is at the portfolio level in valuing your holdings of ZSP.U in CAD.
>> “why would ZSP.U be down .28%?”
According to BMO the market price of ZSP.U was indeed down 0.27% – but the NAV/ unit of the ETF was up 1.07% on the day … almost exactly the same as the index at 1.1%.
The difference between market and NAV would be due to a change in the discount/ premium at which the ETF trades relative to NAV.
@cmj: Steve’s answers are correct, though it’s worth following up on the last one and asking why the market price and the NAV of this fund would be so wildly different. Some difference is normal, but it’s certainly unusual to see the market price go down when the NAV went up over 1%. It’s really hard to say without digging into the weeds. I suspect it has to do with the fact that ZSP is set to pay a distribution when the market opens on Monday, October 6. Although one would usually expect that price drop to occur on the ex-dividend date, which would have been earlier this week.
In any case, this is something of a pricing anomaly that occurs with ETFs occasionally, and they are almost always short-lived. Over any meaningful period the price of an ETF should be very closely tied to its NAV.
Thanks Steve and Dan for your input on ZSP.U. and the added links on Couch Potato to understand its performance.
I appreciate posts on this topic and all white papers, much respect for all the work Justin must have done on this one. Having read the white paper it seems like following all the steps outlined would be very time consuming and achieve a limited reward.
About 18 months ago following a Financial Wisdom Forum discussion on this topic, newguy wrote a tool to make after tax return comparisons a lot easier. It can be customized to each investor’s chosen parameters(tax rates, dividend yield, capital gains rate, MER, in/out RRSP rates, type of account, etc). It also includes disposition taxes based on the investor’s chosen holding period. I helped newguy work out some of the initial kinks and suggested some of the default parameters(they can be changed). Here’s the link: http://newguys.freeoda.com/taxCompare.html
Hopefully other couch potatoes find this tool as useful as I did, it can also help in tax location decisions. If you find a bug please report it in the thread linked in the notes. If Justin can find the time to compare its results to his own software/spreadsheets/calcs all the better.
There is a newer version than what gsp posted at- http://newguys.freeoda.com/Returns.html
You can ask me questions here-
I am currently helping a relative that came into some settlement money and needs a little help with allocating into ETFs/Equities. We have the opportunity to invest $500,000 in the following ETFs/Equities. 1. What would be the best portfolio to maximize tax efficiency?
2.Will creating a combination of XRE and Dream Global REIT (TSE:DRG.UN) in the TFSA be the best option? We are thinking of taking XEM out of the mix.
Current 2015 income is $10,000.
Numbers (current holdings):
***RRSP – $0 Invested
***Non-registered accounts – $0 Invested, $500,000 CASH CAD available.
***TFSA (all TSE iShares)-
$10K XDV iShares Dow Jns Cnd Slct Dvdnd Indx Fnd
$10K CPD Claymore S&P/TSX Canadian Preferred Share ETF
$10K XHB iSHARES CDN HYBRID CORP BOND IDX ETF
$5K XEM iSHARES MSCI EMERGING MARKETS IDX ETF
$6K XRE iShares S&P TSX Capped REIT Index Fund
$6389 is CASH ready to be invested.
What we are looking at adding:
iSHARES US HIGH YIELD BOND INDEX ETF (TSE:XHY) – Non-registered account- withholding tax refund
iSHARES CORE SP 500 INDEX ETF (TSE:XUS) OR VANGUARD S&P 500 Index ETF (TSE:VFV) – Non-registered account – withholding tax refund
Dream Global REIT (TSE:DRG.UN)
TransAlta Renewables Inc (TSE:RNW)