TER/Tax optimized portfolio

Thanks for taking the time to study my portfolio and also for your response:

You’re making a good point and I agree that a TER difference of around 0.05 % is probably not “measurable”. However, my reason for choosing the above portfolio is not only to TER- but also to tax-optimize. This brings me to your statement in the previous thread:

I read on the Mustachian Wiki and also on other websites that the withholding tax can make a difference of up to 1.00 % per year.

Assuming an investment of CHF 100’000.- with an expected return of 8% p.a. and the compounding effect of 20 years this can lead to quite some savings:

Delta in Costs Total Savings
0.05 % CHF 4’335.-
0.10 % CHF 8’708.-
0.20 % CHF 17’570.-
0.30 % CHF 26’589.-
0.40 % CHF 35’768.-
0.50 % CHF 45’109.-
0.60 % CHF 54’615.-
0.70 % CHF 64’289.-
0.80 % CHF 74’133.-
0.90 % CHF 84’151.-
1.00 % CHF 94’345.-

In order to “unlock” these savings (even just the CHF 8’708.- with a cost difference of 0.1%) all I have to do is to find a combination of tax/ter optimized funds. From there investing in 6 instead of 1 funds is not too much of a hustle and the fee’s with IB are negligible. Isn’t that worth it?

To quote the post behind that link:

“a Swiss investor holding shares of VT will effectively receive only (1-0.3)(1-0.35)=45.5% :scream: of the dividends UBS distributes to VT. This is significant, as it roughly equates up to 1% cost p.a. (assuming the dividend yield is 2%)”

That statement is plainly wrong.

Average dividend yields in developed markets are between 2 and 3%. Residual withholding tax isn’t anywhere near 33 or 50%. For U.S. securities through VT, it is in fact zero. Also…

The U.K. will (generally) have no withholding tax on dividends. So no further optimisation here.

Half of last dividend paid by UBS (which is half of UBS’ 2020 dividend paid in 2 times) was tax free. So no swiss whitholding tax.
EDIT : UBS dividend for 2017, 2018 and 2019 were 100% tax-free.

I think I found the answers to questions b) & c):

b) Are “German ETF’s” domiciled in Ireland tax optimized
It looks like Ireland has 9 double taxation treaties with EU countries and Germany is one of them. Luxembourg apparently has 8 treaties but in practices it looks like there is hardly a big ETF performance difference between the two. That means I’m going to stick to an EMU fund based in Ireland.

c) TER/Tax wise, what’s the better choice for emerging markets/asia pacific ETF’s.
I found a document with a list of countries with a double taxation treaty with the USA under the following link:


As China and Japan are the biggest chunks in the EM ETF resp. Asia Pacific ETF, my selected Vanguards ETF’s based in the US appear to be TER/TAX optimized and I’m sticking to them.

@San_Francisco & @REandSTOCK thanks for your answers.

So you would argue that because of all the US double taxation treaties, the zero withholding tax in the UK and the situation like with UBS, where there is no withholding tax on Swiss dividends, there is basically no significant TER/Tax advantage between choosing the VT and an portfolio that is split up like mine?

How representative is the UBS example for the overall Swiss market? Just wondering because a lot of sources (NZZ, Cash, VZ) are arguing that Swiss investors should be careful to pick SMI/SPI ETF’s that are located in Switzerland and not abroad.

CHSPI for example, has a yearly special dividend (the last one) that regroups all the tax-free dividends it received during the year. That means there is no whitholding tax outside the fund as well for these companies’ dividends. If you have VT, you will have your 15% whitholding tax when you get dividend from the fund. You will be usually able to recover those 15%, but sometimes not, and you need to do the work.

The sole fact that the U.S. have a DTA with Japan and China does not necessarily (in and by itself) mean that a U.S. ETF would be most efficient tax-wise. It depends on the applicable tax rate.

Also, the PDF that you linked isn’t relevant with regard to the treatment and withholding tax on Asian dividends paid to a U.S. fund.

Where to?
Did you mean UBS dividend payments to foreign ETFs were free of withholding tax?

I think it’s not a good sign if an ETF outperforms other ones based on the same index. An ETF is meant to hug the index as much as possible.

Btw I once tried to compare VT with VWRD, but the different close price time points, difference absolute unit price, different dividend dates and values made it challenging. I wanted to see how much is truly lost when holding VWRD.

Yes. (Tax-free in general.)

1 Like

Eventually every one is going to lag the index. The question is how much.
No ETF will perfectly replicate the index (especially not the physically-replicating).

I happen to have previously posted about this is as well, taking and compiling actual price/NAV and distribution numbers.

Result: roughly 1% over a period of 6 years. Not a figure I’m going to lose sleep about.

Both follow different indices, but that shouldn’t make much of a difference either.

CHSPI paid 16.4% of its 2019 dividends and 17.8% of its 2018 dividends free of tax.

In your calculation: did you reinvest the dividend? Was the withholding tax applied correctly in both cases? Did you make sure you compare start and end price for the same moment in time? LSE and NYSE have different trading hours. Imagine LSE closes the day on price X, then at NYSE it goes up a further 1%.

But even if it’s only 1% over 6 years, that’s an amount you can get practically for free.

All this work to save a little bit on taxes? Is it really worth the trouble?

1 Like

It all doesn’t really matter or change things:

  • Reinvested dividends: No calculation on reinvested dividends, but the difference (additional compound interest on an additional 0.1 or 0.2%) would be negligible over the period.
  • Withholding tax: There’s none on the Irish ETF. With W8-BEN there would be 15% on the U.S. one, which you could claim credit for with DA-1 if you’re paying (enough) taxes in Switzerland.
  • Start and end price and trading hours: Again, they don’t matter, since we know that these two funds will move virtually identically, with close to 100% correlation anyway (even though the indices and holdings differ slightly - but again, differences are probably negligible, in addition to being unpredictable).

Edit: more bothersome might be VWRD‘s one-time distribution from capital gains though.

ok, agree

My question was rather: in the dividend per piece stated by you: has the withholding tax already been deducted from it or not? If so, did they deduct 15% or 30%? If there was a deduction, you should include it back in, as it will impact the calculation. 15% of 2% is 0.3%, per year!

Disagree. Like I said, imagine that the closing price in Europe was $70.00. But the US exchange remained open. The closing price there for that day was $72. But for you these prices are equivalent. And the same will apply to the end of the period. Europe closes with $110, but US drops to $108. In the end you will have 110/70 for Europe and 108/72 for US. This can distort your calculation by a few percent.

At what tax rate is VWRD cheaper than VT without any deductible interest from a mortage?

Oh, sorry. Yes, gross amounts, as communicated by the funds themselves and/or the Swiss tax administration.

In theory, they should perform the same over the long-term. Given same indices, holdings and distribution in full. However…

  • The indices and holdings are largely - but not fully - the same. But I’m not going to account for that, since it’s intended to just compare holding the two funds.
  • VWRD has curiously had this one-time (?) distribution from capital gains
    So that assumption might not hold water.

But how would you measure it?

I don’t know how, maybe I would take a whole sequence of start and end dates and draw a chart to see if there is a big difference between different reference points.

And I don’t know anything about the capital gains payout of VWRD. It’s interesting thought.

I just realised that the last fiscal reform changed something. If the swiss company pays a dividend with reserve from capital contributions, the company needs to pays a taxable dividend of the same ammount from other reserves. If it doesn’t do so, half of the “tax-free” dividend won’t be tax free. That’s probably the reason UBS did that in 2020. I checked their last annual report and they can still pay 4.68 USD tax free to share-holders, that means 13 years of 50% of tax-free dividends. We might see the percentage of tax-free dividend from the SPI drop a bit in 2020.