The short answer: a physical ETF owns all or a sample of the index securities. A synthetic ETF receives the index return through a swap while holding collateral or a substitute basket. Physical is usually easier to understand; synthetic can track some markets more efficiently. Neither label proves that a fund is safer or cheaper.
The useful comparison is the exact fund structure: what assets it holds, which counterparties it uses, how collateral and securities lending are managed, and how closely it has tracked the correct benchmark. “Physical means no counterparty risk” and “synthetic always tracks better” are both false shortcuts.
Replication compared
| Feature | Physical replication | Synthetic replication |
|---|---|---|
| Primary return mechanism | Owns all or a sample of index securities | Receives index return through a swap or similar derivative |
| Main implementation gap | Sampling, trading, taxes, cash and rebalancing | Swap terms, collateral, reset process and counterparty performance |
| Counterparty exposure | Can arise through securities lending and service providers | Explicitly arises through the derivative counterparty |
| Holdings resemble the index? | Usually, but an optimised sample may omit names | The collateral or substitute basket need not match the index |
| Potential use case | Liquid, accessible underlying markets | Markets or strategies that are costly or difficult to replicate directly |
Full replication and sampling are both physical
A full-replication fund aims to hold every index security at its index weight. An optimised or sampled fund holds a representative subset. Sampling can reduce dealing and custody costs in a broad or less liquid index, but it adds model and selection differences between the fund and benchmark.
Physical funds may also lend securities. That can offset part of the fund’s costs, while introducing borrower, collateral and operational risks. Review the lending policy, collateral, revenue split and historical contribution instead of treating the word “physical” as evidence that no counterparty exposure exists.
How synthetic replication works
In a common unfunded-swap structure, the fund owns a basket of securities and enters a swap that exchanges the basket’s performance for the benchmark return. In a funded structure, the mechanics and collateral arrangement differ. The prospectus and annual report should explain the model, eligible counterparties, collateral and reset policy.
A swap can make it easier to track markets with access restrictions, difficult tax treatment or expensive rebalancing. The trade-off is reliance on the swap contract and its counterparty. Collateral and UCITS limits reduce exposure but do not make operational or counterparty risk disappear.
What the UCITS counterparty limits mean
For an uncleared derivative transaction, UCITS Article 52 limits exposure to one counterparty to 10% of fund assets when the counterparty is a qualifying credit institution and 5% in other cases. That is an exposure limit, not a statement that 10% of the fund is expected to be lost. Collateral, valuation and swap resets can keep actual net exposure lower, but the current prospectus and reports are needed to see how a particular fund manages it.
Tracking needs evidence
Synthetic replication can produce a tight benchmark result, but that is not guaranteed. Physical replication can also track closely, especially for liquid indexes. Compare the fund’s published tracking difference over consistent periods, and check whether the benchmark is gross, net or price return. A comparison against different benchmark conventions can create a false winner.
The total expense ratio is only one contributor. Withholding tax, securities-lending revenue, swap pricing, transaction costs, sampling and cash all affect the realised gap.
Questions to ask before choosing
- Does the fund use full physical, optimised physical or synthetic replication?
- If synthetic, who are the counterparties and how often is exposure reset?
- What collateral is accepted, how is it valued, and where is it held?
- If physical, does the fund lend securities and how is the revenue shared?
- What has the tracking difference been against the exact stated benchmark?
- Are the domicile, tax treatment, spread, assets and listing suitable for the intended account?
Replication is one dimension of fund selection. A transparent synthetic fund can be a more faithful implementation of a hard-to-access index than a heavily sampled physical fund. A straightforward physical fund can be easier to understand. The prospectus-level facts decide the comparison.
Frequently asked questions
Can a physical ETF still have counterparty risk?
Yes. Securities lending can expose a physical fund to a borrower, and service providers add operational dependencies. The policy, collateral and revenue split matter.
Does a synthetic ETF own nothing?
No. A synthetic UCITS ETF can own a substitute basket or have collateral held under a funded structure. Those assets may not resemble the tracked index, which is why the prospectus and collateral disclosure matter.
Does synthetic replication always track better?
No. A swap can reduce some trading and tax frictions, but pricing, collateral, resets and counterparty performance still affect results. Compare matched-period tracking difference against the exact benchmark.
Sources and limitations
ESMA’s guidelines on ETFs and other UCITS issues define index tracking through synthetic or physical replication and set disclosure and risk-management expectations. The UCITS concentration rule sets the 10%/5% counterparty-exposure limits for relevant derivative transactions.
This guide compares structures at a high level. Exact legal form, collateral, securities lending and counterparty exposure vary by fund and can change. It is not a solvency assessment or a recommendation of either method.