If you have compared two ETFs tracking the same index and found one described as “physical” and the other “synthetic”, you have hit one of the few genuinely technical choices in ETF selection. Most fund decisions come down to cost and domicile. Replication method is different: it changes what the fund holds, how it earns its return, and which risks you carry. After eleven years of buying UCITS ETFs I have owned both, and the honest answer still surprises people.
Here it is up front. For most core holdings — a global equity fund, a broad bond fund — physical replication is the sensible default. It is transparent, you effectively own the underlying stocks through the fund, and it is what the largest, cheapest funds use anyway. But synthetic replication is not the villain it was made out to be after 2008. On specific indices, particularly US equity, a well-built swap ETF can track the benchmark better than a physical one for a structural, legitimate reason. The skill is knowing which method you own and why.
Below I explain both plainly, show where each one wins, and cover the risk that physical-fund fans tend to forget: securities lending.
What physical replication actually means
A physical ETF does the obvious thing. To track the MSCI World, it goes out and buys the shares in the MSCI World — Apple, Microsoft, Nestlé, and the rest — in roughly the index weights. When you buy a share of the ETF, you own a slice of a fund that owns those real companies. This is the model behind almost every household-name UCITS ETF you can think of.
There are two flavours. Full replication holds every constituent, which suits large, liquid indices like the S&P 500 or FTSE All-World. Optimised or sampled replication holds a representative subset, used when an index has thousands of small or illiquid names (broad emerging markets, global aggregate bonds) where buying every line would cost more than it is worth. Sampling introduces a little tracking noise but keeps costs down.
The appeal of physical is straightforward: you can see what you own, there is no third party promising to pay you the return, and if the fund provider went under, the assets are ring-fenced from the manager. For a long-term core position, that transparency is worth a lot of peace of mind.
What synthetic replication actually means
A synthetic ETF gets to the same index return by a different route. Instead of buying the index stocks, it holds a basket of collateral — often a diversified pool of large-cap shares or high-quality bonds — and signs a total return swap with a counterparty, usually an investment bank. The deal is simple in principle: the swap partner pays the fund the exact return of the target index, and in exchange the fund pays over the return of its collateral basket. The tracking is delivered by contract rather than by ownership.
That contract is the source of the historic worry: counterparty risk. If the swap partner failed to pay, the fund would be exposed. This is exactly why the structure earned a bad name during the 2008 crisis. But UCITS rules were built for this. Under UCITS, the mark-to-market exposure to any single swap counterparty is capped at 10% of the fund’s net asset value, and in practice providers manage it far tighter, resetting the swap frequently so the gap rarely gets close to the limit.
Funded, unfunded, and multiple counterparties
Modern synthetic funds go further. Many use multiple swap counterparties to spread the risk, and post collateral that often exceeds 100% of the exposure, held at a third-party custodian. In an unfunded swap the fund owns the collateral basket outright; in a funded swap the collateral sits in a segregated account as security. Either way, the aim is that if a counterparty defaulted, the fund could seize and sell the collateral to make investors whole. It is not zero risk, but it is a long way from a naked promise. The residual risk is real but small, and it is documented in the KID and the fund methodology, which is where you should look before buying.
Where synthetic replication quietly wins
Here is the part that gets left out of the “physical good, synthetic bad” story. On some indices, synthetic replication delivers a measurably better return, and it is not a trick — it is tax.
When a physical fund holds US shares, the dividends it receives are subject to US withholding tax. An Irish-domiciled physical S&P 500 ETF benefits from the US–Ireland treaty and pays 15%, but that is still a drag on the gross index return. A synthetic fund never holds the US shares, so at the fund level there is no dividend to be taxed in the same way. A well-structured swap can therefore capture close to the gross total return of a US index, giving it a small but persistent edge over an equivalent physical fund. On the S&P 500 that difference has historically been worth a few basis points a year — not dramatic, but real, and it compounds.
The same logic helps on hard-to-access markets. Some emerging or frontier exposures are expensive or operationally awkward to hold directly; a swap lets the provider deliver the index return without wrestling with local custody and settlement. This is why synthetic funds still thrive in specific corners even as physical dominates the broad-market core. To see how these small structural gaps show up in the numbers, it helps to understand the difference between tracking difference and tracking error before comparing two funds on past performance alone.
The physical risk nobody mentions: securities lending
People choose physical for safety, then overlook the way many physical funds earn extra income: securities lending. The fund lends out its shares to short-sellers and other borrowers in exchange for a fee, which can offset costs and improve returns. It also reintroduces a form of counterparty risk — the borrower might fail to return the securities — which is the very thing physical investors thought they had escaped. It is mitigated by collateral and lending limits, but it is not nothing, and the revenue is not always shared fairly with the fund. I covered this in whether your broker is lending out your ETF shares: “physical” does not automatically mean “no counterparty exposure”. The broader point, that both methods carry hidden mechanics, is in the companion piece on the hidden risks in some EU ETFs.
Learning ETFs the slow way?
The free 12-page Starter Kit gets you to a first, sensible portfolio faster: the 7-filter fund shortcut, the 3-question broker matrix, and the 7 costliest beginner mistakes.
| Feature | Physical | Synthetic (swap-based) |
|---|---|---|
| What the fund holds | The actual index stocks (full or sampled) | A collateral basket plus a swap contract |
| Main risk | Securities-lending counterparty risk (if used) | Swap counterparty risk (UCITS 10% cap, collateralised) |
| Transparency | High — holdings published daily | Lower — you see collateral, not the index directly |
| Tracking on US equity | Good, but dividend withholding tax is a drag | Often slightly better (no fund-level US dividend tax) |
| Best use | Core global equity and bond holdings | Specific exposures with a tracking edge |
| Availability | Dominant across major indices | Niche but strong in US and hard-to-access markets |
My verdict: what I actually do
For my core — the funds I expect to hold for decades — I use physical replication. Not because synthetic is dangerous, but because transparency matters more to me than a few basis points on a position I will never trade away, and the biggest, cheapest global funds are physical anyway. For a beginner, physical is the answer that needs the least explaining.
Where I am happy to own synthetic is on a specific US-equity sleeve where the withholding-tax edge is documented and the provider’s swap structure is well collateralised across multiple counterparties. That is a considered choice, not a default. If you are weighing two named funds against each other — say a physical against a swap-based world tracker — read the methodology in each KID, compare their real tracking difference rather than headline TER, and use a screener like our UCITS ETF comparison tool to line them up. The comparison I keep coming back to as a worked example is Invesco’s swap-based MSCI World against the iShares Core physical fund, which shows exactly how small the practical gap has become. Whichever you choose, the one rule that matters is the one people skip: know which method you own, and check that the KID’s risk section still says what you think it does.
Frequently asked questions
Q: Is a synthetic ETF safe for a long-term European investor?
A: Broadly yes, within UCITS rules. Counterparty exposure is capped at 10% of net assets, swaps are usually collateralised above 100% and reset frequently, and many funds use several counterparties. The residual risk is small but not zero, so read the fund’s KID and methodology and make sure you are comfortable before buying.
Q: Why would a synthetic ETF track a US index better than a physical one?
A: Because a physical fund holding US shares pays withholding tax on the dividends it receives, which drags on the return. A synthetic fund never holds those shares, so at fund level there is no equivalent dividend tax, letting a well-built swap capture closer to the gross index return. On the S&P 500 that edge has historically been worth a few basis points a year.
Q: How do I tell whether an ETF is physical or synthetic?
A: Check the fund’s factsheet or KID, which state the replication method directly, often as “physical (full or optimised)” or “synthetic (unfunded/funded swap)”. If a fund holds a collateral basket and names a swap counterparty, it is synthetic.
Q: Does physical replication mean there is no counterparty risk?
A: Not necessarily. Many physical funds lend out their shares for extra income, which reintroduces a borrower default risk, mitigated by collateral but still present. So the difference is really about which type of counterparty risk you carry and how it is managed, not whether it exists at all.
Q: Is synthetic replication still legal and common in Europe after 2008?
A: Yes. Synthetic UCITS ETFs are fully regulated and widely available; the post-2008 reforms tightened collateral and disclosure rather than banning the structure. Physical dominates the broad-market core, but synthetic remains strong on US equity and hard-to-access markets where its tracking edge is genuine.
Q: Which should a beginner choose for a first global ETF?
A: Physical, in almost every case. It is transparent, it is what the largest and cheapest global funds use, and it needs the least explaining. Once you understand replication well enough to judge a specific swap fund’s structure and tracking edge, you can add synthetic where it earns its place.
Rules and structures evolve, so treat any figures here as approximate 2026 illustrations and verify current details on the fund’s own factsheet and the regulator’s guidance. The European Securities and Markets Authority (ESMA) publishes the UCITS framework that governs how both methods work.
This guide is part of our hub on sector and thematic UCITS ETFs — browse the full cluster for related situations.
European UCITS/ETF investor with 11+ years building a global dividend and index portfolio, and founder of Financial Expert Class. I write every guide here from first-hand experience investing across EU brokers and tax regimes — not theory. More about me →