The end of IBKR + VT: a cheaper, safer, less US-dependent alternative

[Brndete]

@leman

• synthetic ETFs beat physical Irish ETFs

A fair comparison would also match them against physical US ETFs.
From the first study, does Vanguard even offer swaps? I’d consider them biased against it, but share these conclusions:

• assuming that the swap provider passed the entire benefit of the tax advantage to the fund, which may not have been the case"
• All told, it is difficult to accurately assess the true risk and return profile of these funds
• providers are not always forthcoming about costs

The specific question hence isn’t answered by that summary. I do realize it’s a specific point and not part of the overall theme of your post, but maybe someone got something on it

IE-swap beating IE-physical for US stocks is understandable based on tax. In individual comparison they sometimes even beat US-physical, which is less intuitive or transparent to me.

And that’s for US stocks, only. For a global swap that “replicates the performance of the underlying index with one swap, not a basket of regional swaps stitched together” the result could be quite different.

For my own consideration, I wouldn’t want to swap the perceived risk or feeling of “US-politicians doing US-politicians things” that might affect my investments with “investment bankers doing investment bankers things” without understanding it better, incl. the expected results. The whole allure of passive ETF-investing is that it’s not only efficient, but easy and transparent.

[nabalzbhf]

When I checked recently I estimated the swap cost to around 10 bps.

It’s definitely annoying that it’s not transparent like the TER.

And unlike what @leman claims, they track the net return not gross. (And obviously then overperform their index since they can get full US dividend, but it doesn’t mean they outperform or even match the gross index).

[leman]

@nabalzbhf

unlike what @leman claims, they track the net return not gross. (And obviously then overperform their index since they can get full US dividend, but it doesn’t mean they outperform or even match the gross index).

That’s incorrect. It’s an assumption because we don’t know what’s on each swap contract. Invesco mentions a Custom Tax NTR index and it’s a bespoke benchmark whose assumed withholding rates are commercially negotiated between the index provider and the specific bank(s) behind the swap.

I personally choose to have a physical Developed ex-USA sleeve so it doesn’t matter to me, but I encourage you to quantify it and/or choose different assumptions for your own case.

@Brndete

A fair comparison would also match them against physical US ETFs.
The specific question hence isn’t answered by that summary. I do realize it’s a specific point and not part of the overall theme of your post, but maybe someone got something on it
In individual comparison they sometimes even beat US-physical, which is less intuitive or transparent to me.

Large physical funds such as VT rely on sampling. Meaning, they can underperform benchmarks and tend to carry meaningful tracking error. It’s not fresh news.

To sum up how I understand it:

- Physical ETFs tend to underperform benchmarks and carry more tracking error.

- There seem to be clear tracking superiority of synthetic ETFs for EM sleeve.

- Wider studies tend to show that replication method barely matters for developed equity ETFs.

There was a paper from a business school during Covid, not PhD level but it seemed to rely on good methodology, I’ll search for it later on. The bottom line was that synthetic funds track much more tightly, but also that provider execution matters as much as the replication method. I remember UBS and BlackRock showing larger physical versus synthetic gap compared to Amundi, Invesco and DWS.

@Brndete

The whole allure of passive ETF-investing is that it’s not only efficient, but easy and transparent.

I agree with you in principle, but to each their own. I find my own approach pretty simple (3 funds, US synthetic, strong ESG bias, no small caps) but many would disagree; for example I was very surprised to read @nabalzbhf mentioning margin requirements for UCITS at US brokers. Different strokes for different folks

@Brndete

For a global swap that “replicates the performance of the underlying index with one swap, not a basket of regional swaps stitched together” the result could be quite different.

While I am satisfied with the transparency of the synthetic S&P500 funds I use, I agree with you that it could be a different story with a broader index. If you want to dive into this one, you might want to use ICTax to get a feeling for the taxable return of something similar, such as UBS ACWI SF.

[nabalzbhf]

@leman

That’s incorrect. It’s an assumption because we don’t know what’s on each swap contract.

I mean the benchmark they follow is net return.

Obviously the swap for US stock is gross return (but that’s the implementation).

IMO that makes it hard to fairly judge the performance against the benchmark. (And as I said I wish they’d publish the swap cost against the implementation (gross return), so we’d easily know how much the swap leaks)

[Cowi]

Thank you all who contributed to this very interesting thread. You gave me a lot of food for thought.

I tried to add the scenario DeGiro-XALL to the excellent calculations of assemblyrequired, Abs_max and leman, but am stuck with the opportunity costs of distributing vs. accumulating funds. Is this relevant or doesn’t it need to be considered?

Thanks leman that you raised the subject of Estate tax. This is something that I definitely underestimated. For younger people this might not be such an issue, but approaching retirement age this is a different situation. My wife would be overwhelmed if she had to deal with the IRS, so I will start to deinvest from Interactive Brokers.

[dbu]

@Cowi

start to deinvest from Interactive Brokers

The point is not the broker, but the (US) situs of the funds you use.

[leman]

@leman

There was a paper from a business school during Covid, not PhD level but it seemed to rely on good methodology, I’ll search for it later on. The bottom line was that synthetic funds track much more tightly, but also that provider execution matters as much as the replication method. I remember UBS and BlackRock showing larger physical versus synthetic gap compared to Amundi, Invesco and DWS.

Found it. Again, not PhD level but it seemed to rely on good methodology: 22 ETFs, 11 matched pairs, each pair from the same provider, same domicile, same currency, same exchange, same benchmark, isolating replication method as the only variable. Three of the pairs track US indices specifically:

Benchmark	Provider	Synthetic ETF (avg daily return vs. index)	Physical ETF (avg daily return vs. index)
S&P 500	iShares	I500: +0.136% (outperformed)	CSSPX: +0.058% (outperformed, less)
S&P 500	UBS	S5USAS: −0.188% (underperformed)	SP5USY: −0.736% (underperformed more)
MSCI USA	UBS	MUUSAS: −0.240%	USAUSW: −0.265%
MSCI USA	Xtrackers	XMUS: −0.184%	XDUS: −0.266%
Nasdaq-100	Invesco	EQQD: ~flat	EQQQ: large negative (but only 93 days of data, too short to trust)

Across the whole 22-fund sample (Nov 2018–Oct 2021, so it captures Covid), physical ETFs on average lagged their benchmark by roughly 0.69%, while synthetic ETFs on average slightly beat theirs, by about 0.08%. In the four reliable US pairs, the synthetic version tracked closer to (or beat) its benchmark than its physical sibling. Provider-level breakdown shows that UBS and BlackRock have a large synthetic/physical return gap, while Amundi, Invesco, and DWS have a much smaller one. So, provider execution matters as much as the replication method itself.

@Cowi

Thank you all who contributed to this very interesting thread. You gave me a lot of food for thought.

I tried to add the scenario DeGiro-XALL to the excellent calculations of assemblyrequired, Abs_max and leman, but am stuck with the opportunity costs of distributing vs. accumulating funds. Is this relevant or doesn’t it need to be considered?

Thanks leman that you raised the subject of Estate tax. This is something that I definitely underestimated. For younger people this might not be such an issue, but approaching retirement age this is a different situation. My wife would be overwhelmed if she had to deal with the IRS, so I will start to deinvest from Interactive Brokers.

My pleasure. I think that opportunity costs of distributing vs. accumulating should be negligible. Happy to add a Degiro+XALL column for you, following the same assumptions over 20 years:

Scenario	IBKR UK + VT	Saxo + XALL	Degiro + XALL	Degiro + ALLC
CHF 100’000 lump-sum	CHF 336’511 (−4.8%)	CHF 340’968 (−3.5%)	CHF 341’744 (−3.3%)	CHF 353’448
CHF 1’000/month	CHF 464’420 (−3.6%)	CHF 471’211 (−2.2%)	CHF 472’013 (−2.0%)	CHF 481’701
CHF 100’000 lump-sum + CHF 1’000/month	CHF 816’059 (−2.3%)	CHF 812’178 (−2.7%)	CHF 813’757 (−2.5%)	CHF 835’149
CHF 250’000 lump-sum + CHF 2’500/month	CHF 2’063’421 (−1.3%)	CHF 2’032’572 (−2.7%)	CHF 2’036’394 (−2.6%)	CHF 2’089’916
CHF 1’000’000 lump + CHF 10’000/quarter	CHF 5’094’130 (−1.1%)	CHF 4’991’911 (−3.1%)	CHF 5’002’959 (−2.9%)	CHF 5’152’773
CHF 5’000’000 lump + CHF 50’000/quarter	CHF 25’531’502 (−0.9%)	CHF 24’959’554 (−3.1%)	CHF 25’016’609 (−2.9%)	CHF 25’765’716

[Cowi]

Thanks dbu, of course you are right.

I thought one step ahead without explaining it. I use Interactive Brokers because of the access to US Funds. For UCITS Funds I’m quite happy with DeGiro and Flatex.

[nabalzbhf]

@leman

Across the whole 22-fund sample (Nov 2018–Oct 2021, so it captures Covid), physical ETFs on average lagged their benchmark by roughly 0.69%, while synthetic ETFs on average slightly beat theirs, by about 0.08%.

I assume benchmark is net return, right? (So hard to compare with US domiciled + DA-1, which is what want to compare to)

(Edit: yes net return is what they benchmark to)

[nabalzbhf]

@leman

Again, not PhD level but it seemed to rely on good methodology:

It’s weird they don’t even mention the tax reason for overperforming…

[leman]

They do mention it casually, as a given. I imagine it’s not the point of the paper because tax mechanics are well-known, and predictable. The actual finding to me is that provider execution matters as much as the replication method itself.

We know that physical replication tends to carry more tracking error but at the end of the day, the replication method doesn’t make much difference.

[Abs_max]

I understand.
So the difference in performance for VT vs WEBG is 0.09% as per your calculations. I have similar number in mind.

I don’t fully understand the synthetic stuff. It feels like derivative and not the actual stocks. So I am mainly using physical replication for now.

Maybe someday I will warm up to synthetic ETFs too.

[Brndete]

@nabalzbhf

compare with US domiciled + DA-1

Wouldn’t comparing actual returns, work? Thus skipping the benchmark altogether, or adding one for reference? I only started with manual samples, which is cumbersome, but would be easy if data is available.

Maybe morningstar etc. will aready do the job? trackingdifferences.com gets you overviews with ETFs by index, already marked with physical and swap, but alas it’s UCITS, only.

[nabalzbhf]

@Brndete

Wouldn’t comparing actual returns, work?

Yep that’s ideal (but lots of UCITS funds are kinda new and you need to have funds tracking the same index, and have both funds be distributing).

[Cowi]

@leman

Scenario	IBKR UK + VT	Saxo + XALL	Degiro + XALL	Degiro + ALLC
CHF 100’000 lump-sum	CHF 336’511 (−4.8%)	CHF 340’968 (−3.5%)	CHF 341’744 (−3.3%)	CHF 353’448
CHF 1’000/month	CHF 464’420 (−3.6%)	CHF 471’211 (−2.2%)	CHF 472’013 (−2.0%)	CHF 481’701
CHF 100’000 lump-sum + CHF 1’000/month	CHF 816’059 (−2.3%)	CHF 812’178 (−2.7%)	CHF 813’757 (−2.5%)	CHF 835’149
CHF 250’000 lump-sum + CHF 2’500/month	CHF 2’063’421 (−1.3%)	CHF 2’032’572 (−2.7%)	CHF 2’036’394 (−2.6%)	CHF 2’089’916
CHF 1’000’000 lump + CHF 10’000/quarter	CHF 5’094’130 (−1.1%)	CHF 4’991’911 (−3.1%)	CHF 5’002’959 (−2.9%)	CHF 5’152’773
CHF 5’000’000 lump + CHF 50’000/quarter	CHF 25’531’502 (−0.9%)	CHF 24’959’554 (−3.1%)	CHF 25’016’609 (−2.9%)	CHF 25’765’716

I tried to buy ALLC at DeGiro but I got an error message in the morning and now the funds can’t be found any longer (only in USD on Xetra, which is uninteresting to me).

I wait for an answer from DeGiro if the funds will be available again on SIX and else back to square one.

Unfortunately the interesting synthetic US-stocks ETFs seem not to be available in CHF on SIX (at least not on DeGiro).

[leman]

@Abs_max

I understand.
So the difference in performance for VT vs WEBG is 0.09% as per your calculations. I have similar number in mind.
I don’t fully understand the synthetic stuff. It feels like derivative and not the actual stocks. So I am mainly using physical replication for now.
Maybe someday I will warm up to synthetic ETFs too.

I totally agree with you. This discussion goes back and forth from discussing life and death, to discussing piffling amounts of money.

Choosing Amundi Prime means a tax drag smaller than one day’s worth of volatility over 20 years, while sparing you from this long stretch of smallcaps underperformance. It really is a simple solution.

@Brndete

Wouldn’t comparing actual returns, work? Thus skipping the benchmark altogether, or adding one for reference? I only started with manual samples, which is cumbersome, but would be easy if data is available.
Maybe morningstar etc. will aready do the job? trackingdifferences.com gets you overviews with ETFs by index, already marked with physical and swap, but alas it’s UCITS, only.

I did it in this very thread a few weeks ago, for near-identical Blackrock S&P500 funds. Feel free to have a look.

Many insisted on doing it for global funds (which I don’t understand - the perceptible tax effect is on the US part) but you will struggle finding near-identical funds for that comparison. These are tiny amounts anyway, that can swing one way or the other.

The bottom line is that it doesn’t matter, especially on the day where your loved ones learn about your inevitable death.

@Cowi

Unfortunately the interesting synthetic US-stocks ETFs seem not to be available in CHF on SIX (at least not on DeGiro).

Degiro can add new funds on request (and for free, unlike Saxo and the others).
Customer support competence can vary wildly, but you get there at the end. You can save a lot of time by including the KIID in your national language. I would use this template:

• ISIN: LU3086265710
• DICI: https://api.fundinfo.com/document/c310d8fb5cf548c76be5cb9cf302229d_141165/PRP_CH_fr_LU3086265710_YES_2026-06-30.pdf
• Nom: BNP Paribas Easy MSCI ACWI UCITS ETF (Acc)
• Devise: CHF
• Bourse: SWX
• Ticker: ALLC

[Ardius]

I received 500 chf with the DA-1 for 2025. I will certainly switch to ucits etf. My mortgage will increase and the interest will again reduce the da1 reimbursement.

It does not worth the money and I will increase peace in my mind…

[Abs_max]

Someone reading your comment might say „whattttt 500 CHF is not worth it :)“ when we go after every last cent in TER%

But I agree that it all depends on the relative value. 500 CHF for investment value of 500K CHF might mean it’s meaningless in big scheme of things

[Tony1337]

In 10 years that may be 5000 CHF though. Depending on where you are in your investment journey.

[xerox5003]

When I look back at how I started investing, it was all about rock-bottom costs, TER, squeezing every last basis point. That phase was fun as hell - every cent counted.

The wealthier I get (and I’m still nowhere near my goals), the more other things start mattering. I’m past the stage of hunting for the absolute cheapest broker or the absolute cheapest product. Reliability and security have become way more important to me than they used to be.

I’ve reached the point where I just don’t want the US situs tax headaches anymore, or to leave my future heirs with that mess. Especially now that European TERs have gotten so low that they’re basically on par with the US-domiciled ones. Rationally I don’t think IBKR or a US-domiciled ETF will actually cause major problems. But European alternatives have improved a lot, and I’m glad they have.

I still use IBKR for some things and I don’t expect it to become a problem in the future. Still, a Swiss broker just feels a lot better to me - and I’m happy to pay the stamp taxes and fees for that peace of mind. I no longer feel the need to nickel-and-dime every last cent.

Personally I think I’ll eventually drop IBKR completely and just run a Swiss broker, maybe with one European broker on the side at most.

Right now I’ve fully replaced my VT position with VWRA and will do another switch once the UCITS version of VT becomes available.