Quantifying tax drag from US domiciled ETFs that hold non-US assets (eg VEU)

For investors outside the US. Personal investments, personal finance, investing news and theory.
Sister forums: Canada, Spain (en español)
---------------
Post Reply
Topic Author
Hockey Monkey
Posts: 93
Joined: Thu Aug 13, 2020 10:23 pm

Quantifying tax drag from US domiciled ETFs that hold non-US assets (eg VEU)

Post by Hockey Monkey »

I’ve attempted to quantify the tax drag on US domiciled ETFs that hold non-US assets.

Using VEU as an example, the resulting estimate is 0.32% which is higher than I expected. Hopefully someone more knowledgeable in this area can check my calculations and assumptions below.

When combined with the 0.08% MER for a total of 0.40%, VEU starts to look quite expensive. Couple it 50/50 with VTS with an MER of 0.03% and you have an overall MER of (0.03 + 0.40) / 2 = 0.215%

VGS is starting to look quite attractive at that point with an MER of 0.18% and none of these issues. Have I made a mistake somewhere?

--------------------------------------------------------------
Vanguard All-World ex-U.S. Shares Index ETF (VEU)

From the October 2019 annual report at https://www.vanguardinvestments.com.au/ ... /?overview

1,239,812,000 Net Dividends (A) - page 18
120,873,000 Foreign Withholding Tax (B) - page 19
1,360,685,000 Gross Dividends (A+B)
8.89% Foreign Withholding Tax Level 1 (B / (A + B))

VEU currently has a dividend yield of 3.06% and MER of 0.08%

Any tax drag applies when the investors home country (in this case Australia) has a double taxation agreement (DTA) with the country of the asset which would allow a tax offset if the assets were held in the same country as the investor instead of in the US.

Australia has a DTA with most countries of assets held in VEU. Cross referencing page 8-14 of the annual report with https://treasury.gov.au/tax-treaties/in ... x-treaties, 93.4% of assets by weight are covered by a DTA (All except Brazil, Colombia, Greece, Hong Kong, Kuwait, Portugal, Qatar, Saudi Arabia and UAE)

Different countries tax at different rates, however using 93.4% as an approximation we have a tax drag of 0.25%
3.06% (yield) * 8.89% (withholding tax) * 93.4% (DTA proportion) = 0.25%

There is also a 0.07% tax drag for the lost of franking credits for the 4.9% of Australian assets in VEU
4% (ASX yield) * 80% franking / 0.7 * 30% tax * 4.9% weight = 0.07%

Does this mean VEU has a total effective MER of 0.40%? (0.08% MER + 0.25% + 0.07% tax drag)?
Last edited by Hockey Monkey on Thu Aug 20, 2020 11:33 pm, edited 1 time in total.
User avatar
andrew99999
Posts: 1021
Joined: Fri Jul 13, 2018 8:14 pm

Re: Quantifying tax drag from US domiciled ETFs that hold non-US assets (eg VEU)

Post by andrew99999 »

Beyond my capabilities, but the one thing that may be missing is that the US likely has a bunch of tax treaties also, and I think the drag might would be the difference between the cost with Australian treaties less the cost with US treaties (don't forget that you can claim a tax deduction for the 15% US withholding tax on top)?
Topic Author
Hockey Monkey
Posts: 93
Joined: Thu Aug 13, 2020 10:23 pm

Re: Quantifying tax drag from US domiciled ETFs that hold non-US assets (eg VEU)

Post by Hockey Monkey »

Thanks Andrew, it sure is a complex area. This paper from Canada is interesting reading and appears to confirm your hypothesis.

Foreign Withholding Taxes
How to estimate the hidden tax drag on US and international equity ETFs
https://www.pwlcapital.com/wp-content/u ... linked.pdf

specifically scenarios

C. US-listed ETF that holds developed markets stocks directly.
Since the stocks are held by a US-listed ETF, Level I withholding taxes are payable at rates determined by tax treaties between these overseas countries and the US (not Canada).

and

E. Canadian-listed ETF that holds developed markets stocks directly.
With this type of fund, Level I withholding taxes are levied by the countries where the stocks are domiciled, and payable at Canadian rates. Since Canada has different tax treaties with the various countries, the foreign withholding tax rates may differ from those in Type D funds.

Another interesting point
As a group, emerging countries tend to have higher withholding tax rates on dividends paid within US funds.
User avatar
andrew99999
Posts: 1021
Joined: Fri Jul 13, 2018 8:14 pm

Re: Quantifying tax drag from US domiciled ETFs that hold non-US assets (eg VEU)

Post by andrew99999 »

Great link, and from PWL too!
You probably know this, but just in case - the 15% dividend tax that the US takes from VEU is claimable for Australians, so if your marginal tax rate is over 15%, FWT level 1 would be zero (or 15% if your MTR was below the tax free threshold).
Everything else you noted is interesting. It would be great to get numbers on this if it's possible.
Topic Author
Hockey Monkey
Posts: 93
Joined: Thu Aug 13, 2020 10:23 pm

Re: Quantifying tax drag from US domiciled ETFs that hold non-US assets (eg VEU)

Post by Hockey Monkey »

Yes, I understand that and the importance of the W8BEN to reduce this US tax from 30% to 15%

I'm mostly interested in understanding the hidden drag and ultimately the effective MER of VEU compared to VGS, IWLD etc.

Nothing has swayed me from the VAS/VTS/VEU 20/50/30 plan yet though for it's simplicity and diversification. A full replication IWLD without the current spread issues is probably the only thing that would, although I'm pretty sure the VGS MER will be halved if that occurs :)
String
Posts: 8
Joined: Sun Jul 08, 2018 11:19 pm

Re: Quantifying tax drag from US domiciled ETFs that hold non-US assets (eg VEU)

Post by String »

I think you are out by a bit with your franking credits loss. For a 4% yield with 80% franking this would be 0.04*0.8/0.7*0.3*0.049 which is about 0.07%.

Also using the us funds exposes you to the loss of level 1 + level 2 withholding if you ever end up with a 0% marginal tax rate.

I would think the australian domiciled funds which have the higher MER are more likely to get cheaper relative to the already cheap us funds.
Topic Author
Hockey Monkey
Posts: 93
Joined: Thu Aug 13, 2020 10:23 pm

Re: Quantifying tax drag from US domiciled ETFs that hold non-US assets (eg VEU)

Post by Hockey Monkey »

Thanks for the tip. I meant to go back and check that one. Total worst case is now 0.40%. VGS is getting tempting.
Topic Author
Hockey Monkey
Posts: 93
Joined: Thu Aug 13, 2020 10:23 pm

Re: Quantifying tax drag from US domiciled ETFs that hold non-US assets (eg VEU)

Post by Hockey Monkey »

A third drag I've seen mentioned elsewhere is the loss of the 50% long term capital gains discount on distributions. Can anyone confirm if this is real.

Eg assuming 1/4 of the 3.06% distribution was from long term capital gains.

0.25 * 3.06% distribution * 50% discount * 47% marginal tax rate = 0.18%

Total is now 0.08% MER + 0.25% withholding tax + 0.07% franking credit loss + 0.18% CGT discount loss = 0.58% total expenses.

Actually, rechecking the annual report https://api.vanguard.com/rs/gre/gls/1.3 ... ts/7608/au page 21, there have been no "Distributions from Realized Capital Gains" in the past 5 years. I suspect this potential CGT drag floating around is a red herring where people were comparing to VGS which has reported capital gains at times.

Think I'm ready to pull the trigger on a VTS/VEU portfolio as the best of the current bunch (VGS, IWLD etc).
Last edited by Hockey Monkey on Sat Aug 22, 2020 4:16 am, edited 2 times in total.
String
Posts: 8
Joined: Sun Jul 08, 2018 11:19 pm

Re: Quantifying tax drag from US domiciled ETFs that hold non-US assets (eg VEU)

Post by String »

I think the effect of capital gains distributions is likely to be the opposite. I can't remember where you find this but i don't think veu or vts have ever distributed capital gains (outside maybe their year of inception). The australian funds on the other hand have regularly had small capital gains distributions.
Topic Author
Hockey Monkey
Posts: 93
Joined: Thu Aug 13, 2020 10:23 pm

Re: Quantifying tax drag from US domiciled ETFs that hold non-US assets (eg VEU)

Post by Hockey Monkey »

Yes I think you are right. I just updated my post above. VEU has not reported capital gains in the past 5 years
Topic Author
Hockey Monkey
Posts: 93
Joined: Thu Aug 13, 2020 10:23 pm

Re: Quantifying tax drag from US domiciled ETFs that hold non-US assets (eg VEU)

Post by Hockey Monkey »

A followup after further research on tax drag from US ETF's that hold non-US assets

From https://www.passiveinvestingaustralia.c ... on-the-asx

There are currently three main options to invest in the world from Australia

1) VGS/VGE 90/10, gross MER 0.21% (excluding small caps)
2) IWLD/VGE 90/10, gross MER 0.13%
3) VTS/VEU 55/45, gross MER 0.05%

The following ETFs currently suffer from tax drag due to holding non-US assets in a US based ETF.
VGE - Vanguard FTSE Emerging Markets Shares ETF (Underlying fund VWO on NYSE)
IWLD - iShares Core MSCI World All Cap ETF (includes the US ETF iShares Core MSCI EAFE ETF)
VEU - Vanguard All-World ex-U.S. Shares Index ETF (US domiciled)

Although the VTS/VEU combo appears really low cost (0.05% overall MER), all options end up within 4 basis points of each other once tax drag is factored in.

After factoring in tax drag
1) VGS/VGE, effective MER 0.23%
2) IWLD/VGE, effective MER 0.24%
3) VTS/VEU, effective MER 0.20%

It is almost like the funds price with tax drag in mind to remain competitive with each other.

IWLD is rumoured to be planning to switch to direct replication once it reaches a certain size, however the MER is also likely to increase when that happens based on a similar change in Canada https://www.canadianportfoliomanagerblo ... revisited/

I think I'm convinced to stop worrying about tax drag and prioritise ETF selection based on other factors like dividend reinvesting (DRP) availability and US estate issues.

Tax drag factors calculated from recent annual reports of yield and withholding tax
VGE withholding tax drag 0.20%
IWLD withholding tax drag 0.08%, franking credits tax drag 0.02%
VEU withholding tax drag 0.25%, franking credits tax drag 0.07%
User avatar
andrew99999
Posts: 1021
Joined: Fri Jul 13, 2018 8:14 pm

Re: Quantifying tax drag from US domiciled ETFs that hold non-US assets (eg VEU)

Post by andrew99999 »

Great effort there, thank you.
Hockey Monkey wrote: Tue Sep 08, 2020 5:52 am I think I'm convinced to stop worrying about tax drag and prioritise ETF selection based on other factors like dividend reinvesting (DRP) availability and US estate issues.
+1
Topic Author
Hockey Monkey
Posts: 93
Joined: Thu Aug 13, 2020 10:23 pm

Re: Quantifying tax drag from US domiciled ETFs that hold non-US assets (eg VEU)

Post by Hockey Monkey »

Hockey Monkey wrote: Tue Sep 08, 2020 5:52 am IWLD is rumoured to be planning to switch to direct replication once it reaches a certain size, however the MER is also likely to increase when that happens based on a similar change in Canada https://www.canadianportfoliomanagerblo ... revisited/
I made an incorrect assumption reading the above article that the Canadian XEF fund increased their MER on switching to direct replication. In fact it looks like XEF originally launched in 2013 with a higher MER of 0.30% based on this article https://canadiancouchpotato.com/2013/04 ... the-world/.

Then at some point the MER was reduced to 0.22%

Let's assume that when switching IWLD to direct replication, BlackRock continue to hold IVV (S&P 500) and IJR (S&P 600 small caps) at 0.04% and 0.07% respectively to keep costs low and only switch the Ex-US and Canadian assets to avoid tax drag. This would mean they need to be able to directly hold MCSI EAFE + Canada for around 0.18% to maintain the overall MER of 0.09%.

I guess this is within the realms of possibility given they can do MCSI EAFE in Canada for 0.22%, although XEF has CAD $3.6B Net Assets, so economies of scale apply. I'm not sure an IWLD at AUD $200M Net Assets or AUD $70M of the MCSI EAFE + Canada portion could be achieved for the same cost. Australian iShares MSCI EAFE (IVE) ETF has an MER of 0.31% today and that’s with the US economies of scale and suffers the same tax drag we are trying to avoid.

We will just have to wait and see. Until this occurs, IWLD is about the same cost as VGS due to tax drag and suffers from higher spreads, turnover and tracking errors due to the lack of mid caps. Until IWLD does switch to full replication, I plan to continue to hold VGS
mtatt
Posts: 3
Joined: Thu Oct 22, 2020 5:14 pm

Re: Quantifying tax drag from US domiciled ETFs that hold non-US assets (eg VEU)

Post by mtatt »

Hockey Monkey wrote: Thu Aug 20, 2020 7:32 pmAny tax drag applies when the investors home country (in this case Australia) has a double taxation agreement (DTA) with the country of the asset which would allow a tax offset if the assets were held in the same country as the investor instead of in the US.
I may be misunderstanding Level 1 FWT.

Level 1 relates to withholding taxes levied by countries where the stocks are domiciled. An ETF can reclaim foreign withheld taxes depending on where it;s domiciled and the tax treaties of that country. Assuming that the tax treaties of Australia and US are the same, would a (hypothetical) VEU domiciled in Australia only be able to claim and recover the same amount of FWT as the VEU domiciled in the US? Hence, virtually no difference between the two? (except Level 2 FWT which is recoverable by the individual investor in Australia).
alex_686
Posts: 13286
Joined: Mon Feb 09, 2015 1:39 pm

Re: Quantifying tax drag from US domiciled ETFs that hold non-US assets (eg VEU)

Post by alex_686 »

I will throw in another wrinkle.

“ 120,873,000 Foreign Withholding Tax (B) - page 19” represents the dividends withheld under the treaty rate, not the foreign taxes paid.

Many countries exempt mutual funds from taxes so the funds can claw back those withheld taxes. India and the EU come to mind - or at least when I was last doing this professionally. It often requires one to hire a local tax firm and fill forms with copious data. Turn around time is often 2 years. Often the expense is just not worth it.
Former brokerage operations & mutual fund accountant. I hate risk, which is why I study and embrace it.
mtatt
Posts: 3
Joined: Thu Oct 22, 2020 5:14 pm

Re: Quantifying tax drag from US domiciled ETFs that hold non-US assets (eg VEU)

Post by mtatt »

alex_686 wrote: Mon Dec 28, 2020 9:58 pm “ 120,873,000 Foreign Withholding Tax (B) - page 19” represents the dividends withheld under the treaty rate, not the foreign taxes paid.
By looking at this:
The fund has filed tax reclaims for previously withheld taxes on dividends earned in certain European Union countries. These filings are subject to various administrative and judicial proceedings within these countries. Such tax reclaims received during the year, if any, are included in dividend income. No other amounts for additional tax reclaims are reflected in the financial statements due to the uncertainty as to the ultimate resolution of proceedings, the likelihood of receipt of these reclaims, and the potential timing of payment.
(from the VEU annual report)

I would assume then that:

Gross Dividend = dividends received before any taxes
Foreign Withholding Tax = foreign taxes withheld plus less any reclaimed taxes received during this FY
Net Dividend = not actual cash received, but rather Gross dividend minus the above

I guess?
Last edited by mtatt on Tue Dec 29, 2020 2:02 am, edited 1 time in total.
alex_686
Posts: 13286
Joined: Mon Feb 09, 2015 1:39 pm

Re: Quantifying tax drag from US domiciled ETFs that hold non-US assets (eg VEU)

Post by alex_686 »

mtatt wrote: Mon Dec 28, 2020 10:11 pm Foreign Withholding Tax = foreign taxes withheld plus any reclaimed taxes received during this FY
It would be “less”, not plus. And who knows how long the process takes. I would think most of the clawed back income would be over a year and land somewhere else on the income sheet- but it has been a few years since I last dealt with this. And maybe Vanguard is highly efficient.

Anyways, I would be skeptical of trying to get to 2 decimal points. A better way is to measure the tracking error between the fund, the tax adjusted index, and the unadjusted index.
Former brokerage operations & mutual fund accountant. I hate risk, which is why I study and embrace it.
Topic Author
Hockey Monkey
Posts: 93
Joined: Thu Aug 13, 2020 10:23 pm

Re: Quantifying tax drag from US domiciled ETFs that hold non-US assets (eg VEU)

Post by Hockey Monkey »

mtatt wrote: Mon Dec 28, 2020 9:45 pm
Hockey Monkey wrote: Thu Aug 20, 2020 7:32 pmAny tax drag applies when the investors home country (in this case Australia) has a double taxation agreement (DTA) with the country of the asset which would allow a tax offset if the assets were held in the same country as the investor instead of in the US.
I may be misunderstanding Level 1 FWT.

Level 1 relates to withholding taxes levied by countries where the stocks are domiciled. An ETF can reclaim foreign withheld taxes depending on where it;s domiciled and the tax treaties of that country. Assuming that the tax treaties of Australia and US are the same, would a (hypothetical) VEU domiciled in Australia only be able to claim and recover the same amount of FWT as the VEU domiciled in the US? Hence, virtually no difference between the two? (except Level 2 FWT which is recoverable by the individual investor in Australia).
A VEU domiciled in Australia but where the underlying assets are held in the US would still suffer from lost level 1 FWT.

DTA’s allow for FWT to be offset against domestic tax when the foreign assets are held directly in the country that has the DTA. For example in Australia, VGS, VISM and VAE all hold ex-US foreign assets directly in Australia.

I found this great video from Ben Felix explaining the issue after creating this thread
https://www.youtube.com/watch?v=_gC2y4j5s0M
Topic Author
Hockey Monkey
Posts: 93
Joined: Thu Aug 13, 2020 10:23 pm

Re: Quantifying tax drag from US domiciled ETFs that hold non-US assets (eg VEU)

Post by Hockey Monkey »

alex_686 wrote: Tue Dec 29, 2020 1:27 am
mtatt wrote: Mon Dec 28, 2020 10:11 pm Foreign Withholding Tax = foreign taxes withheld plus any reclaimed taxes received during this FY
It would be “less”, not plus. And who knows how long the process takes. I would think most of the clawed back income would be over a year and land somewhere else on the income sheet- but it has been a few years since I last dealt with this. And maybe Vanguard is highly efficient.

Anyways, I would be skeptical of trying to get to 2 decimal points. A better way is to measure the tracking error between the fund, the tax adjusted index, and the unadjusted index.
Great to have someone onboard that has dealt with foreign withholding taxes professionally. I’m very much an amateur.

Wouldn’t any reclaimed (recovered?) taxes boost the dividend paid to unit holders rather than be reported as foreign withholding taxes?

Have you reviewed this report from PWL capital? That’s ultimately how I came up with the formulas used above.

https://www.pwlcapital.com/wp-content/u ... linked.pdf
alex_686
Posts: 13286
Joined: Mon Feb 09, 2015 1:39 pm

Re: Quantifying tax drag from US domiciled ETFs that hold non-US assets (eg VEU)

Post by alex_686 »

Hockey Monkey wrote: Tue Dec 29, 2020 5:21 am Great to have someone onboard that has dealt with foreign withholding taxes professionally. I’m very much an amateur.

Wouldn’t any reclaimed (recovered?) taxes boost the dividend paid to unit holders rather than be reported as foreign withholding taxes?

Have you reviewed this report from PWL capital? That’s ultimately how I came up with the formulas used above.
We may be getting into semantics here.

Foreign Withholding Tax is a debit. i.e., a negative number. Adding in reclaimed taxes would reduce that debit. i.e. -3 + 1 = -2.

I did not read the report in depth. The one you are pointing to came out after I left that space. It seemed to be pitched towards Canadian investors, but I have read similar ones by PWC aimed at US fund accountants.
Former brokerage operations & mutual fund accountant. I hate risk, which is why I study and embrace it.
Topic Author
Hockey Monkey
Posts: 93
Joined: Thu Aug 13, 2020 10:23 pm

Re: Quantifying tax drag from US domiciled ETFs that hold non-US assets (eg VE

Post by Hockey Monkey »

alex_686 wrote: Tue Dec 29, 2020 6:57 am
Hockey Monkey wrote: Tue Dec 29, 2020 5:21 am Great to have someone onboard that has dealt with foreign withholding taxes professionally. I’m very much an amateur.

Wouldn’t any reclaimed (recovered?) taxes boost the dividend paid to unit holders rather than be reported as foreign withholding taxes?

Have you reviewed this report from PWL capital? That’s ultimately how I came up with the formulas used above.
We may be getting into semantics here.

Foreign Withholding Tax is a debit. i.e., a negative number. Adding in reclaimed taxes would reduce that debit. i.e. -3 + 1 = -2.

I did not read the report in depth. The one you are pointing to came out after I left that space. It seemed to be pitched towards Canadian investors, but I have read similar ones by PWC aimed at US fund accountants.
As I understand it we have

Gross dividends = Net Dividends (paid to unit holders) + Foreign Withholding Tax (withheld by level 1 country)

What I’m saying is that recovered taxes would boost Net Dividends which are passed on to the unit holders.

When the assets are held in the same country as the unit holder and that country has a DTA with the level 1 country, the FWT can be offset against domestic taxes. When this is not the case, the unit holder has to pay full tax on the net dividend, resulting in double taxation
alex_686
Posts: 13286
Joined: Mon Feb 09, 2015 1:39 pm

Re: Quantifying tax drag from US domiciled ETFs that hold non-US assets (eg VE

Post by alex_686 »

Hockey Monkey wrote: Tue Dec 29, 2020 7:33 am As I understand it we have

Gross dividends = Net Dividends (paid to unit holders) + Foreign Withholding Tax (withheld by level 1 country)

What I’m saying is that recovered taxes would boost Net Dividends which are passed on to the unit holders.

When the assets are held in the same country as the unit holder and that country has a DTA with the level 1 country, the FWT can be offset against domestic taxes. When this is not the case, the unit holder has to pay full tax on the net dividend, resulting in double taxation
Like I said, we are getting down to semantics. We can reorder your equation to:

Gross - Withholding = Net

This is closer to the operational accounting order. Gross dividends are fixed. Whatever is not withheld flows into net. When taxes are reclaimed it goes against the Withholding account, thereby reducing the withholding amount, thus increasing Net.

Unless the dividends were paid in 2018 and it has taken this long to reclaim, which is a very real possibility. Gross, Withholding, and Net are all current year values - and for many funds 2019 taxes where just filed. So we are going to have to tuck that reclaimed taxes in some other income account.

And I am probably making a mountain out of a mole hill.
Former brokerage operations & mutual fund accountant. I hate risk, which is why I study and embrace it.
Topic Author
Hockey Monkey
Posts: 93
Joined: Thu Aug 13, 2020 10:23 pm

Re: Quantifying tax drag from US domiciled ETFs that hold non-US assets (eg VE

Post by Hockey Monkey »

alex_686 wrote: Tue Dec 29, 2020 8:47 am
Like I said, we are getting down to semantics. We can reorder your equation to:

Gross - Withholding = Net

This is closer to the operational accounting order. Gross dividends are fixed. Whatever is not withheld flows into net. When taxes are reclaimed it goes against the Withholding account, thereby reducing the withholding amount, thus increasing Net.

Unless the dividends were paid in 2018 and it has taken this long to reclaim, which is a very real possibility. Gross, Withholding, and Net are all current year values - and for many funds 2019 taxes where just filed. So we are going to have to tuck that reclaimed taxes in some other income account.

And I am probably making a mountain out of a mole hill.
Is there anything we can do with this additional knowledge to improve the accuracy of the percentage tax drag estimate?
alex_686
Posts: 13286
Joined: Mon Feb 09, 2015 1:39 pm

Re: Quantifying tax drag from US domiciled ETFs that hold non-US assets (eg VE

Post by alex_686 »

Hockey Monkey wrote: Tue Dec 29, 2020 3:13 pm Is there anything we can do with this additional knowledge to improve the accuracy of the percentage tax drag estimate?
Not much. Don't go to 2 digits of precision. It will give you a false sense of certainty.

To estimate tax drag you have to estimate dividends. Dividends vary, so you are going to need to throw in some error bars. You are then going to need to estimate withholding, which will also vary thanks to the claw back provision. Now, the error bars on your dividends estimates are going to swamp the error bars on your estimated withholding.
Former brokerage operations & mutual fund accountant. I hate risk, which is why I study and embrace it.
Topic Author
Hockey Monkey
Posts: 93
Joined: Thu Aug 13, 2020 10:23 pm

Re: Quantifying tax drag from US domiciled ETFs that hold non-US assets (eg VEU)

Post by Hockey Monkey »

Vanguard released their AMMA tax statements today (Attribution managed investment trust member annual statement) which prompted me to have another look a tax efficiency of VTS/VEU vs VGS

Looking at the following portfolios in Vanguard Portfolio Builder
Option 1 - 12 month yield 2.04%
76.5% VGS - MSCI developed LC + MC
13.5% VISM - MSCI developed SC
10% VGE - Emerging Markets

Option 2 - 12 month yield 1.38%
57% VTS - US Total Market Index
43% VEU - All-World Ex-US Total Market Index

The difference in yield appears to be mostly due to distributed capital gains in option 1, whereas option 2 VTS/VEU appears to have never distributed capital gains, only distributing dividends.

Even though these distributed capital gains are mostly long term, they are still taxable, so for someone in the top 47% tax bracket in Australia, with a 23.5% tax (50% CGT discount) x 0.66% yield difference = 16 basis points additional tax.

This tips the balance back in favour of Option 2, even when accounting for tax drag from lost level 1 withholding tax and franking credits if investors are comfortable with W8BEN forms and US estate tax risks.

Is there something unique about ETF implementation in the US that avoids distributed capital gains? eg better liquidity? Or is the nature of VTS/VEU being total market funds avoiding mid cap sized stocks migrating from VISM small cap to VGS mid cap and vice versa triggering capital gains?
daffyd
Posts: 199
Joined: Sun Sep 29, 2013 11:51 pm
Location: Australia

Re: Quantifying tax drag from US domiciled ETFs that hold non-US assets (eg VEU)

Post by daffyd »

Hockey Monkey wrote: Mon Aug 02, 2021 2:01 am
Is there something unique about ETF implementation in the US that avoids distributed capital gains?
Yes, the US doesn't treat creation/redemption of ETF units by authorised participants (i.e. market makers who keep the etfs price in line with the underlying shares) as taxable events. So a well designed, highly liquid US-domiciled ETF will send over individual shares with the highest capital gains in redemption baskets and the taxable event only happens when the individual sells. This can even be used for strategies with slightly higher turnover to avoid incurring capital gains when trading.

Aus domiciled ETFs can still be slightly more efficient than a standard fund but not this efficient.
Topic Author
Hockey Monkey
Posts: 93
Joined: Thu Aug 13, 2020 10:23 pm

Re: Quantifying tax drag from US domiciled ETFs that hold non-US assets (eg VEU)

Post by Hockey Monkey »

daffyd wrote: Mon Aug 02, 2021 2:36 am
Hockey Monkey wrote: Mon Aug 02, 2021 2:01 am
Is there something unique about ETF implementation in the US that avoids distributed capital gains?
Yes, the US doesn't treat creation/redemption of ETF units by authorised participants (i.e. market makers who keep the etfs price in line with the underlying shares) as taxable events. So a well designed, highly liquid US-domiciled ETF will send over individual shares with the highest capital gains in redemption baskets and the taxable event only happens when the individual sells. This can even be used for strategies with slightly higher turnover to avoid incurring capital gains when trading.

Aus domiciled ETFs can still be slightly more efficient than a standard fund but not this efficient.
Really appreciate your in depth response here. All starting to make sense now. When it comes to tax efficiency, US ETF > AUS ETF > AUS Managed Fund.
Post Reply