In this tax-heavy episode, Justin discusses his part-time obsession – foreign withholding taxes. Throughout the show, he explains which specific ETFs are best held in each account type in order to reduce this largely hidden tax drag. He is joined by industry experts from BMO, BlackRock and Vanguard, who each discuss their company’s approach to mitigating this cost. ZEM and XEC also go head-to-head in the latest ETF Kombat. As a side-bar, Justin’s popular model ETF portfolios have been updated for 2020, and now include Light and Ridiculous levels of portfolio complexity (with Ludicrous and Plaid versions in the works). Justin finishes off the show by answering a listener question regarding the new iShares foreign equity ETFs that trade in U.S. dollars.
- Three ETF structures for accessing U.S., international and emerging stock markets [0:02:03.3]
- How Level I and Level II withholding taxes arise [0:02:28.3]
- How the type of account holding your ETF impacts foreign withholding taxes [0:03:44.3]
- The most tax-efficient U.S. equity ETFs for each account type [0:05:41.3]
- Why investors with smaller portfolios should generally avoid U.S.-listed ETFs in their RRSP (even when using the Norbert’s gambit strategy to cheaply convert their loonies to dollars) [0:06:43.3]
- When can purchasing U.S.-listed ETFs in TFSAs and taxable accounts still make sense? [0:09:03.3]
- Kevin Prins of BMO ETFs discusses some of their tax-efficient international and emerging markets equity ETFs [0:11:31.3]
- The worst fund structure for international equity ETFs [0:14:28.3]
- The most tax-efficient international equity ETFs for each account type [0:15:27.3]
- Steven Leong of BlackRock Canada discusses XEF’s transition from a U.S. wrap structure to one that holds the stocks directly, and why the U.S. wrap structure can still be a more cost-effective choice for some other asset classes [0:18.40.3]
- The most tax-efficient emerging markets equity ETFs for each account type [0:24:11.3]
- Why ZEM’s tax-efficient structure may still not beat XEC’s after all costs are considered [0:26:03.3]
- ETF Kombat: ZEM vs. XEC [0:27:51.3]
- The most tax-efficient global equity ETFs for each account type [0:31:23.3]
- Scott Johnston of Vanguard Canada discusses the recent tax-efficient changes to VXC’s fund structure [0:32:28.3]
- The overall cost of currency-hedged global fixed income ETFs (and why there’s still room for improvement) [0:35:42.3]
- Foreign withholding taxes on the Vanguard and iShares asset allocation ETFs [0:38:09.3]
- Ask Bender: Martin has a question about the foreign withholding tax implications of the new iShares USD ETFs (XUU.U, XEF.U, XEC.U and XAW.U) [0:39:28.3]
- Next episode, we’ll tackle asset location (i.e. where you should hold each of your ETFs for maximum tax efficiency). See you then! [0:42:06.3]
Hi Justin, super helpful information. I’m going to listen to this entire series.
For an all-in-one asset allocation ETF, such as XEQT, held in a non-registered taxable account, how do you go about dividing up the foreign withholding tax amount on your taxes? More specifically, given that this ETF holds assets from the US (via ITOT) and international markes (via XEF), how can you claim the foreign withholding tax credit for the US portion of the dividends when you also receive other international dividends? A T3 doesn’t exactly show you the breakdown of the US versus international dividends. Thanks in advance.
@Eric – I haven’t prepared my own taxes in awhile, but I seem to recall just selecting “Other” and writing “Various” for the country breakdown, and that seemed to work.
Hi Justin, I’m re-listening to your Foreign Withholding tax podcast for a second time since it was originally released. One because I like to be punished with numbers and second because I haven’t found any new content in Couch Potato Investing to listen to in a while.
In the podcast you explain the added cost of currency conversion when buying (For Example) ITOT yourself and that people may often forget the added cost versus holding a Canadian fund that holds ITOT for you. You also explain you can get some added savings by performing Norberts Gambit to complete the conversion. You used an example where the added cost of the conversion was about $600 versus holding the Canadian Fund so we need to keep that in mind.
My question is, isn’t that cost also there when you buy the Canadian version that holds ITOT? Your money still needs to be converted to buy ITOT regardless, except you don’t get to see the exchange rate or cost of conversion because it is hidden inside the fund? Possibly one of those costs outside the MER that you never get to see?
I don’t know if that’s actually the case, just asking for clarification?
@Martin – There would still be a cost, but it’s assumed that companies like Vanguard and BlackRock can access very low institutional FX rates.
Thanks again Justin. Every time you post a Podcast i learn a little more…
I have 70 K in USD that i currently have in a US$ high intertest saving account. Looking to increase Fixed income return at the same time keep it in US$. I have both equities ITOT and IEMG investments with a T1135 above 100K under 250K. So i like the .U method of ETFs if we can keep the cost down. My choices are USIG ( and go above the 250K- Foreign Holding and a lot more reporting) or go to something like ZIC.U as an equivalence. Your Thoughts? Do you have any other suggestion ?ZUP.U?. On one of your pod cast you mentioned buy Bond ETF focused on discount vs Premium. ( a could not find the comment.).
Hi Justin- I just came across your blog, this is fantastic resources, thanks for sharing.
I returned to Canada in ’18 and was holding VXC in a foreign brokerage. Unfortunately didn’t move it to a Canadian brokerage yet. For 2019, I have the tax document of dividends and foreign taxes and obviously no T5, can I still claim dividend tax credits as if these were invested in a Canadian brokerage ? Any other nuances from a reporting standpoint of domestic vs foreign source income in this case?
Thanks,
Raghu.
@Raghu: I would recommend speaking with a tax accountant for this question.
Amazing work Justin! First time posting on your blog, but the content quality on here has been amazing and truly informative. I have learned a lot about the technicalities of ETF investing (WHT, ABC, etc) and the labour in your podcast is evident. Look forward to continued learning.
Question regarding the “.U” ETFs noted in this podcast – why would someone with US dollars invest in these items versus the US-listed ones? As noted in the podcast, the economies of scale, tracking error, etc. are much more beneficial using the US products instead of the CDN products.
@EZ: I’m so glad you’ve been enjoying the blog/podcast!
If an investor did not want to convert their U.S. dollars to Canadian dollars, the Canadian-based “.U” ETFs could avoid T1135 reporting and possible U.S. estate taxes in very limited circumstances.
Hi Justin. Great episode it really answered many of the question I have re:FWT. This last year I have switched to holding VT in my RRSP. Just to keep it simple and max tax efficiency. The question I now have is: as I get older, I will be purchasing more bonds in the RRSP. I was thinking about using BNDW. On the one hand, it would be easier to balance with VT as both in USD and also globally diversified. It will be easier to keep the 80:20 or 60:40 ratio. But I know it’s not currency hedged to CAD. The other option is buying a canadian ETF like ZAG but it makes balancing to the ratio harder as two different currencies and as a result more trading with Norbit’s in the event of a market correction. What are your thoughts on your clients using BNDW in their RRSP?
@Francis: I wouldn’t recommend trading one portfolio management issue (rebalancing between CAD and USD securities) with another (holding unhedged foreign bonds).
You could consider keeping a small amount invested in a Canadian-listed foreign equity ETF, like VXC or XAW, and use this to rebalance between your Canadian stock and bond ETFs when markets go up or down.
When will this be uploaded on your youtube channel?
Wait it looks like it is there. Don’t know why I didn’t see it the first time
Hi Justin,
This was just a small point you mentioned but I was interested in how you calculated the cost of executing Norbert’s Gambit. I think you said that assuming the costs are two $10 trade fees (so $20 total) and the bid-ask spread, then the cost for converting $20,000 is 0.27%, and for $2,000 is 1.31%.
The cost for the trading fees seem to be 0.1% (for $20K) and 1% (for $2K) if I am calculating properly, so I back into you calculating the cost of the bid/ask spread being 0.17% (for $20K) and 0.31% (for $2K). I was trying to think through how you calculated it. I would have assumed the % cost of the bid/ask spread would be the same, as it is the same per share.
And if you don’t have time to get to this I totally understand!
@Jeremy: The cost of Norbert’s gambit changes depending on the following:
– Current spot rate
– Amount of conversion
– Commissions charged
– Bid-ask spread % cost of DLR/DLR.U
Based on current rates, you would do the following:
1. Lookup the current spot rate to convert CAD to USD (currently around 0.7614). So if you’re converting $20,000 CAD at zero cost, your benchmark is $15,228 USD ($20,000 CAD x 0.7614)
2. Now take your CAD amount, subtract the trading commission, divide this result by the ask price of DLR, multiply by the bid price of DLR.U, and subtract the second trading commission. In this case, ($20,000 CAD – $10 CAD) / $13.28 CAD x $10.10 USD -$10 USD = $15,193 USD
3. Subtract your benchmark USD proceeds from your Norbert’s gambit USD proceeds. So $15,193 USD – $15,228 USD = -$35 USD
4. Convert the -$35 USD to Canadian dollars, using the spot rate from Step 1. So -$35 / 0.7614 = ~$46 CAD
5. Divide this result by your original $20,000 CAD conversion amount. So $46 CAD / $20,000 CAD = 0.23%
Therefore, in this example, your expected cost of Norbert’s gambit is around 0.23%.
Hi Justin,
Thank you for this informative podcast. My question relates to TD E-Series.
As Dan was mentioning earlier
https://canadiancouchpotato.com/2019/08/30/td-e-series-funds-the-next-generation/
TD has switched the benchmark to Solactive.
I’m wondering how much is FWT for
–TDB902 US E-Series
–TDB911 International E-Series
and how many layers of FTW are there (1, 2)?
Thanks.
@Ale: The U.S. and international TD e-Series funds would have a similar FWT drag as a Canadian-listed U.S. equity ETF (like ZSP) or a Canadian-listed international equity ETF that holds the stocks directly (like ZEA). So between 0.25-0.30% per year in a TFSA/RRSP, or fully recoverable in a taxable account (there is only 1 layer in both cases).
Hi JB,
Thanks again producing more amazing content. I can tell that it would take a lot of time to produce the podcasts and blog and I really appreciate what you do. I really enjoyed this episode and I’m eagerly awaiting the next episode.
I know that there are more moving parts with passive investing in corporate accounts. I have wondered if there are ideal choices for which ETF if a person is comfortable with Norbert’s Gambit and is building a portfolio similar to your model portofolios (ie. VCN plus ITOT and/or XUU plus XEF and/or IEFA plus XEC and/or IEMG). Could you comment on what you recommend and if it’s possible to guess at the magnitude of discrepancy in withholding taxes and/or total returns so that a person could judge if the cost and hassle of Norbert’s Gambit is worthwhile. The part I’m having trouble guessing is the amount of recoverability of the withholding taxes: is there simple calculation like US funds holding International stocks can recover 80% of withholding taxes and Canadian funds holding International funds can recover 100% of these taxes, but the fund of fund structure can recover 0% of level 1 withholding taxes and 80% of level 2 withholding taxes? I very much appreciated the other blog posts you did that showed the inputs on the various lines of the corporate and personal tax returns, and I was just wondering if there is an estimate or a pattern that jumps out without crunching the individual numbers and controlling for all other variables. BTW: the tax gain harvesting article was awesome too.
If you assume that an investor can pay enough dividends to recapture recoverable taxes, and if an entire year passed with no currency fluctuations, and assuming TER discrepancies are negligble and tracking errors are similar, then I would guess the following:
US: ITOT would have recoverable or partially recoverable foreign taxes (withholding taxes) but these taxes are not recoverable at all with XUU. XUU would have about 0.26% that is unrecoverable and XUU has 0.04% higher MER too. Thus, perhaps if you held both XUU and ITOT corporately, you might see ITOT outperform by about 0.3% annually. So it’s best to hold the US fund that holds US stocks directly (ie. ITOT), rather than the Canadian Fund of a US fund that holds the stocks (XUU).
International Developed: My guess is that XEF’s structure would make it be taxed in a similar way as IEFA and taxes would be recoverable in a similar way in corporate accounts.Thus, for this pair of ETFs, maybe the discrepancy between these choices would be small and mostly consist of XEF’s higher MER (0.14%), slightly higher TER and slightly higher tracking error. And my conclusion would be that if the difference was less than 0.2% annually, maybe it’s not worth doing Norbert’s Gambit and doing more accounting work because of the different currencies.
Emerging Markets: IEMG would avoid one layer of withholding taxes that would be unrecoverable for XEC and thus IEMG would outperform by maybe 0.31% (unrecoverable level 1 withholding taxes), some portion of level 2 withholding taxes (?), the MER discrepancy (0.12%), a tiny amount for TER discrepancy, and small tracking error discrepancy.
That reader question was really great: I had wondered why something like XUU.U would be created if it doesn’t have the same CUSIP as XUU. I just looked up the bid-ask spread of a ETFs I mentioned in this comment and noted that all the US-listed ones (ITOT, IEFA, IEMG) had a one cent bid-ask spread, while XUU, XEF, XEC all had 2 cent spreads and XUU.U had a 3 cent spread. The volumes for XUU.U are very low and I would guess that this would make XUU.U unadvisable to buy.
An unrelated question:
I noted that ZSP and ZSP.U don’t have the same CUSIP either and thus can’t be used for Norbert’s Gambit. Is there anything like DLR and DLR.U that has currency risk and broad market exposure risk (it could be a regular holding and then you could sell on either the US side or the Canadian side depending what currency you wanted it to settle in)? Do you think any ETF providers would make a product in the future that would be the same CUSIP? It would be nice if there was a product that gave the market exposure that so many people want and also allowed Norbert’s Gambit. It seemed like HXS and HXS.U were almost great products but these seem like not great choices for many reasons now.
@Diego Revere: Thanks for listening and reading! The podcast and blogs do take an enormous amount of spare time, but it’s worth it when I receive such positive feedback :)
I’m going to attempt to tackle the asset location in corporate accounts discussion in a future blog post, so stay tuned for that (I’m sure it will be a beast).
But from a foreign withholding tax perspective in a corporate account:
– XUU and ITOT would be similar (they both have 15% withholding on U.S. dividends, but these are generally recoverable)
– XEF would be more tax-efficient than IEFA, as it only has one layer of withholding tax, which is generally recoverable
– IEMG would be similar to XEC (but it has lower fees, so will likely perform better)
With that said, foreign dividend income does not flow as well out of the corporation (once the refundable taxes come into play) – this is an outdated blog, but tries to illustrate the issue:
https://canadianportfoliomanagerblog.com/taxation-of-foreign-income-in-a-corporate-account/
As for better alternatives to DLR/DLR.U – I haven’t found any that I’m comfortable with. If these ETFs had a higher price but a similar bid-ask spread, they would be incredible (but the market makers probably wouldn’t like this too much). Maybe some product providers will step up to the challenge in the near future – we’ll have to wait and see.
Hi Justin,
In your new Vanguard Ridiculous model portfolio, what is the reasoning under the “RRSP, RRIF, LIRA and LIF Accounts” section behind using VIU (MER 0.23%) and VWO (MER 0.12%) over VXUS (MER 0.09%)?
They seem to have the same holdings, but if you are going to be buying VWO in USD anyway, why not just get the cheaper VXUS and cut out VIU entirely?
@David Nixon: Having the exact same region weights across the Vanguard Light and Ridiculous portfolios will make my future asset location discussions easier for investors to digest (and also to illustrate, as I will need to break-up the international and emerging markets regions once I go Plaid ;)
I’ve also posted a blog/calculator that shows investors how they can determine the weights to use for other similar ETF combinations (such as VCN/VTI/VXUS, VCN/VTI/VEA/VWO, and VCN/VT):
https://canadianportfoliomanagerblog.com/more-alternatives-to-vanguards-asset-allocation-etfs/
Thanks Justin for another great podcast! Most of it was familiar to me from reading your whitepaper (which definitely helped make my portfolio more tax efficient – thanks again!), but I still managed to learn a few new things. The statements from the ETF providers were really interesting.
I’m curious about the US bonds foreign tax – I had previously read that Canadians were exempt from US bond interest – where is the ~0.1% you mentioned coming from?
Also, this isn’t directly related to the current podcast, but I was looking at your latest “Ridiculous” model portfolios and wondering why is the equity distribution different between the iShares and Vanguard portfolios (eg. for 100% equity, 25/45/25/5 vs 30/39.78/22.5/7.71) – I understand the last two values will be different due to South Korea and a couple other countries, although shouldn’t the iShares version allocate more to emerging than Vanguard does?
@Graeme: I’m glad you enjoyed the episode!
Good question on the small amount for foreign withholding tax on U.S. bonds – I’ll reach out to Steven Leong and see if he can shed some light on the specific types of U.S. bonds that are still subject to this tax.
In the “Ridiculous” model portfolios, they use the exact same equity weights as the “Light” asset allocation ETFs (these weights have been chosen by the product providers, iShares and Vanguard). I made them the same so that investors can directly compare the product fees and foreign withholding taxes of the Light portfolios vs. the Ridiculous portfolios.
iShares uses custom target weights for their U.S., international and emerging markets exposure. If they used a market-cap weighting (like Vanguard does), they would have slightly different weights in their 100% equity ETF (XEQT): 25% Canadian, 43.79% U.S., 22.31% international and 8.9% emerging markets
Thanks for the explanation! Have you seen any reasoning from iShares for under-weighting emerging markets so heavily (nearly half of what would be expected with cap-weighting)?
Also, if it weren’t for the assert allocation ETFs, I’m curious what sort of breakdown between Canada/US/International you’d prefer (with the caveat I realize no one can predict the future)? It’s interesting that both of the funds have a fair bit more US and less Canada than you’d get with the 1/3-approach Dan’s talked about.
And I’ve got to ask – what are you planning for the “ludicrous” and “plaid” portfolios :)?
@Graeme: Not sure for the reasoning for iShares under-weighting emerging markets. Like most decisions based on back-testing, it probably just looked better ;)
With our PWL Toronto clients, we still invest most of their portfolios in a 1/3 Canadian, 1/3 U.S. and 1/3 international/emerging markets split. When I first started at PWL in 2006, this foreign equity allocation was very close to market-cap weights. Since then, the U.S. has increased to become a larger component of the global market, but we haven’t made adjustments to our portfolios.
The Vanguard portfolios only have around 3% less in Canadian stocks than global, relative to a 1/3 split (30% vs. 33%). The iShares portfolios do have noticeably less Canadian home bias though.
The “Ludicrous” portfolios will use the same ETFs as in the “Ridiculous” portfolios, but will employ a traditional asset location process (i.e. equities in TFSA first, taxable accounts next, RRSPs last). The “Plaid” portfolios will also use the same ETFs as in the “Ridiculous” portfolios, but they were employ an after-tax asset allocation/location process :)
@Graeme: Here’s the response I received from Steven Leong of BlackRock Canada regarding the small amount of withholding tax on U.S. bonds:
On the US bond interest and WHT question, the issue is the following. While it is true that direct interest payments are generally exempt from US WHT, the payments made by ETFs (or specifically, registered investment companies) are subject to US WHT (even if the underlying income to the ETF is bond interest).
More recently (late 2015), the law in the US has changed such that an investment company can declare some or all of its distributions as deriving from “qualified interest income”, which is exempt from US WHT. I’ve included a link below that shows the $ and % amounts of QII for our US-listed bond ETFs.
https://www.ishares.com/us/literature/tax-information/qualified-interest-income-qii-percentages-2019.pdf
The limitation though, is that interest received by the US ETF is only “qualified” if it is received from a US-domiciled bond issuer.
Looking at the two US bond ETFs held in XBAL, XGRO etc GOVT (which holds only US treasuries) and USIG (which holds only corporates) for 2019:
GOVT: 99%+ is “qualified”
USIG: ~68% was qualified
There are a large number of foreign corporate issuers into the US domestic bond market, which explains the differences above.
One other comment on the .U series and Norbert’s Gambit for your benefit:
The reason XUU.U trades under a different CUSIP than XUU is that if they traded under the same CUSIP, the distributions on both symbols would have to be in the same currency (either USD or CAD). There would be no way of paying USD to investors who purchased the US listing and CAD to investors who purchased the CAD listing under a single CUSIP. We ultimately concluded that a large majority of investors who wanted to invest in USD would also prefer to receive distributions in USD (rather than receive CAD in a USD account and potentially face forced conversion costs).
Hope points are both helpful. Please let me know if you need anything else.
Thanks for a great episode. I admire Dan and your leadership in this topic. I enjoyed learning how one of the major firms didn’t even realize their fund had a competitive advantage in tax efficiency until reading your analysis. I am sure it has had an impact on bringing down costs for all of us.
This might be in the chutzpah category since it involves a free resource you have created for the community. I was wondering if you might have an updated version of the foreign tax withholding calculator available. I used the existing one to “follow along at home” while listening to this podcast and it missed a few of the ones discussed and some recent changes of course. And could you add a column for the fund structure? (US ETF, Canadian fund holding US etf, Canadian fund holding the underlying assets).
I also had an idea to float for others feedback. I have been thinking about the foreign tax efficiency impact of using equity or bond futures rather than ETFs. I trade these anyways in an interactive broker account and am a regular DIYer so you don’t need to be an expert. My understanding is they don’t have any foreign tax withholdings and are taxed as capital gains (unless you are a day trader). Their use is limited since I believe you can’t trade them in a TFSA or RRSP account and taxable accounts are already most tax efficient for withholding taxes purposes.
However, I thought of this when you were discussing foreign bonds funds and their high cost due to foreign withholding tax inefficiency and MER. They might be interesting in this case as a combined cost over 0.70% for bonds is tough long-term.
Thanks again!
@Jeremy: I’m glad you enjoyed the show! I’ll be posting blogs that closely follow along with the narration, so this should help as you listen to the podcast.
I’m not familiar with the types of global bond futures available (and the cost of these), so perhaps you could provide readers with your own experience, and how it works.