Thanks. I’ve read those sources some years ago, but still don’t get it.
The PWC source mentions 10% for companies, but the summary shows 15% for private investors. Can you pin point to the part where the 10% should apply? Or even better in the DBAs?
The colorful “Japan” is a hyperlink. You can click on it. The other countries and the withholding taxes by Japan are listed within.
Don’t be too smug about it 
Can you pin point the part where the 10% should apply, or not? With a reliable source?
I’m aware the DTA states 10%, yet here it mentions the 15.3% that I get withheld.
Same here. I quote the English translation, my Japanese isn’t good enough for reading tax code:
Aren’t there many countries where the withholding tax is fixed and may be higher than the treaty rate? I thought the US is rather the exception than the rule (and brokers need to be IRS qualified intermediaries for W-8BEN).
In those cases, you’re eligible to get a refund from that country. The time and effort required for this depends on the country.
I think e.g. Swiss-domiciled funds (and possibly also individual companies) have to pay out net dividends. I.e., the withholding taxes are paid directly by the fund, not by the broker. I’m not completely sure about the exact process but with such a system, it’s impossible for a broker to apply a reduced treaty rate based on the tax residence of the investor.
Sometimes things are rather straight forward, but some people give me the runaround, by just not looking at information. Putting up even better sources doesn’t work. Reflecting the demand to do their thinking for them works.
You want to say you have actually seen the 10% in both sources and chose to go back and screenshot the unrelated resident and non-resident section? I don’t believe you and some humility would do you good here.
Regardless, since you now did dig into some sources: Why should the DTA not apply? Art. 10 Abs. 2 (German) states clearly that:
I didn’t see any caveats in the vicinity. Also, the information you posted refers to the non-treaty status.
1 Like
Probably. I tested that Canada works the same (IBKR files an NR4). But for European and Australian securites you can only get a withholding tax voucher.
1 Like
In german/french only
for PDF of all withholding rates according to Swiss Treaties. ooen the pdf named “Vertragliche Begrenzungen der ausländischen Steuern”
DBA Japan
https://www.fedlex.admin.ch/eli/cc/1971/1720_1725_1725/de
1 Like
Are you being rude? Why?
I simply can’t make the connection between the 10% in the DTA and the 15% from the other document titled “Übersicht über die Auswirkungen des Abkommens”. IB withholds 15%. Therefore, I took the net 5% as granted so far. So for me, it’s not straight forward yet at all.
What am I missing? Is IB making a mistake? Can these last 5% be reclaimed seperately? Since you seem to know so much more, why not give a hint?
It states explicitly that the withholding tax get’s reduced to 10% for everyone that is not a pension fund or major (>10%) shareholder.
That Japan applies different withholding taxes for public and private companies has no bearing on what the treaty demands. It demands a maximum of 10%.
This is further supported by the fact that you can only claim 10% tax credit in Switzerland. Because that is what Japan can legally withhold.
That IBKR doesn’t care and also doesn’t help you with reclaiming the excess withholding tax is a you-problem for everybody else (IBKR, Japan, Switzerland).
Switzerland links the form for having the correct withholding tax applied (below page in German, but can be switched to EN):
Looking at it, it can work similar the US and Canada. You send your form to the Japanese tax office and the Japanese dividend paying company. Then they will only withhold the treaty rate.
The problem probably is that IBKR holds the securities as a broker in an omnibus account. So neither do you get the US-Japan treaty rate of 10%, nor can you claim, that you are the actual owner and get the 10% of the Switzerland-Japan treaty. You get the 15% non-treaty rate.
According to the document of the National Tax Agency JAPAN you posted you can also reclaim it later:
https://www.nta.go.jp/english/taxes/withholing/Information/13002.htm
But it requires also filing the first form, and also suffers from the same problems. You could of course try with a printout of the dividend statements from IBKR, but well…
1 Like
Thanks for the clarification. Your second link is for non-dividend refunds, but I got a form for dividends.
I’m not convinced whether it’ll be succesfull, but I guess I will give this a try in the coming days to see if and how long back I can get these 5% net L2TW down to 0%.
Doing taxes is supposed to be fun, filling this kind of forms is not.
What you can do to estimate cost of specific etf is going through their yearly report and look how much dividends they had and how much was withheld.
Thanks for the advice. I checked the VT annual report but couldn’t figure out which dividend values are relevant to be honest.
Even though it’s difficult to get to final number easily, I think the real difference between VWRL and VT will amount to 0.1 to 0.15% per annum after taxes in terms of their tax efficiency and tracking differences. I don’t worry too much about TERs because they can be misleading sometimes.
I tried to recreate some of calculations from the original google sheet and the Bogleheads Wiki post using the assumptions of % for L1WT and assuming 0% L2WT for both ETFs. I came to similar values. Total costs of 0.35% for VWRL and 0.13% for VT. However, I did not account any TD. I read people mentioning index providers use maximum withholding tax rate and comparing TD of IE and US domiciled ETFs is difficult. TrackingDifferences.com shows the average TD over several years but not for US domiciled ETFs. I found those values only on Trackinsight.com but only for the previous year.
I think total cost of VWRL (0.35%) includes 0.22% TER. However the tracking difference of VWRL is much less than 0.22%. In other words even though VWRL should be 0.22% behind its index, it’s actually not. So 0.22% is not real cost of holding this ETF.
So in my view that should be used to adjust the total costs of holding VWRL.
I think best way would be to ignore the TER difference of VWRL and VT. And just focus on the tax loss portion to make a decision. The tax loss I think is between 0.1-0.15%
Your comment about index assuming highest WHT is correct . But it should be same approach for FTSE all world (benchmark for VWRL) and FTSE all cap (benchmark for VT)
Sorry, I should have clarified. Yes, I meant including the TER. TWR alone I calculated 0.06% for VT and 0.13% for VWRL. Although based on what seems to be a bit outdated L1WT numbers from the wiki post.
Thanks for the clarification that sounds reasonable. Interestingly, the TD for VT was reported at -0.91% for last year. Anyways, by the looks of it, it seems I am splitting hairs. I was just generally curious about the differences and wanted to understand withholding taxes of dividends.
Well, it’s more complicated than this. I will refer specifically to MSCI indices, FTSE is probably similar, and give 1 example. Most ETF’s performance is reported on a “total return”, with dividends reinvested, basis, even if they are distributing. They track an index, usually a net total return version. To calculate it, it is assumed that dividend distributed by each stock is reinvested after deducting a tax of 30%.
Now, an Irish ETF on US stocks pays 15% withholding tax on dividends distributed to it, and there are no taxes for payouts from the fund. Assuming as an example 2% dividend yield, 0% TER and ignoring compounding in 1 year time frame, this ETF’s performance will be 0.3% better than that of it’s index, i.e. TD of -0.3%, although the fund didn’t do anything but receiving and distributing or accumulating dividends.
ETFs also earn money by lending stocks. iShare is rather transparent about it. A fund can earn more than 0.1% p.a. with this. These earnings also decrease TD.
1 Like
Thanks for more clarity.
Just a question.
Let’s say there are two ETFs -: ETFA and ETFB. Both track MSCI USA. ETFA is domiciled in US and ETFB is domiciled in IE.
Would both of these ETFs track exactly the same Total Return Indexes as benchmark? Or they track two different Indexes because the underlying WHT assumptions need to be different ?
It’s up to the management company to choose which index their ETF track. Most US ETFs on US stocks would track a US-based index, which would count no withholding tax, I guess. PBUS is vanilla MSCI USA.
1 Like
By the way, I checked the fact sheet of VT and VWRL, it seems their benchmarks are assuming pre-tax dividend reinvestment for VT.
But I could not understand clearly for VWRL because I can find different comparisons. I saw one where only index price is compared so I think it excludes dividends anyways. Then I saw one where all gross income is reinvested but it’s not clear what happened to WHT.
I think VWRL performance vs benchmark might not have this WHT effect (at least based on link below) and can be purely attributed to its lending or simply its effective operations. But I am not 100% sure.
I think you might be looking into this comparison because you might want to buy VWRL but worried about money you might lose by not buying VT instead.
If that’s the case , then please consider following
- These two ETFs are not exactly the same. VWRL doesn’t invest in small caps while VT does. So it’s not just choice of ETF but also about exposure to small caps.
- VT is more tax efficient for CH residents, it is not insignificant, but it is also not so huge difference.
I think in long run, you will be fine with either one.
#1 VT , #2 VWRL for global investing with one ETF
2 Likes
I moved all my equity allocation from VT to VUG+BRK.B shares last year.
Why? Main reason was marginal tax drag on dividends is huge in Suisse Romande.
Being single, a high earner, having military tax, and living in Geneva my marginal tax rate is 50%. Since VT pays ~2% in dividends per year my marginal tax rate gives a 1% tax drag on holding VT.
Compare that to VUG with ~0.5% dividends and BRK.B with 0%. Holding them in equal weight gives ~0.25% dividends or 0.125% tax drag. Expense ratios are also slightly lower but this is less relevant.
Essentially, as long as VUG+BRK.B does not underperform VT by >0.9% over the long run then I am still better off.
Summary of differences between VT and 50% VUG + 50% BRK.B:
- Tax drag: VT has 1% while VUG/BRK.B has 0.13%
- Market correlation: VT has 98% correlation and VUG/BRK.B has 94%
- Growth/Value: Holding a growth fund like VUG and value focused holding company like Berkshire gives rough diversification here
- Geographies: VT is a global fund but as all market cap weighted funds are, is dominated by the US with 65% of its holdings there. VUG+BRK.B is 100% US.
I appreciate that this approach has more US exposure, more tech exposure, and is quite concentrated in specific stocks like AAPL. However, I don’t necessarily see those as bad things going forwards and so for the sake of almost 1% lower tax drag I am content to hold this for the long run.
Performance since 2015. This is starting with $100k and investing $5k inflation adjusted per month since. It is not adjusted for the tax savings (so for VT subtract 1% per year and VUG+BRK.B subtract 0.13% per year).
Note that only 2 years since 2015 (2015 and 2019) did VT outperform. During periods of international stock outperformance VT should do better. Or if there is a big US tech crash and Berkshire doesn’t offset substantially. Otherwise VUG+BRK.B has had pretty incredible returns over the last decade+.
3 Likes
I’m not sure growth is the way to go here. It had some good years recently, true. But isn’t it an anti-factor (of the value factor)?
Since you state that you earn a lot (and therefore have or will have a lot of money), did you consider warpping your investments into a company to shield it from tax? (And leaving Geneva, because 50% wtf)
1 Like