Physical replication buys all or a sample of index securities. Synthetic replication obtains index performance through a derivative, often a swap, alongside a collateral or substitute basket. Neither label is enough on its own: documents, counterparties, collateral, costs and tracking quality matter.
Two ways of obtaining exposure
A physical ETF buys securities, either all index constituents or a sample. A synthetic ETF uses a derivative contract to obtain the specified performance. This describes a mechanism, not an automatic ranking between a good product and a bad one.
What the fund actually holds
Full replication aims to hold the index constituents; sampling selects representative holdings and may introduce deviations. A synthetic fund’s basket can differ from its benchmark. Read both the promised exposure and the assets actually held rather than assuming they are identical.
Further reading: how stock indices work.
A fair comparison
Two funds can track the same index using different methods and publish different TERs. Compare realised replication, spread, liquidity and derivative terms too. A lower advertised expense ratio does not establish that the complete structure is better for every use.
Further reading: how the stock market works.
Counterparties and protection
A swap creates dependence on a counterparty performing its obligations. Collateral, limits and resets can mitigate but not eliminate that exposure. Structures vary, so the prospectus and reports need to explain how the protections work in the specific product.
Further reading: the risks and rights attached to stocks.
Physical replication also needs scrutiny
Securities lending can earn revenue but introduces risks and revenue-sharing arrangements. Custody, sampling and difficult-to-trade markets also matter. Owning the securities does not eliminate market losses, liquidity constraints or the consequences of operational decisions.
A difficult index has no universal solution
Large indices or expensive market access can complicate full replication. Sampling and derivatives address different practical problems. Do not infer a required method from the market name alone. Compare the actual mandate with what the manager discloses.
Separate market loss from structural failure
If an index loses 15% and the fund follows, the loss does not prove a replication malfunction. Counterparty, custody or execution failures are different issues. Distinguish the desired economic exposure from the mechanism delivering it.
Questions for the disclosures
Look for counterparties, collateral composition, eligibility criteria and exposure management. For securities lending, examine revenue allocation and safeguards. Answers should come from the specific fund’s documents rather than assumptions attached to the words physical or synthetic.
Further considerations
- Full physical replication is intuitive but may be inefficient for very broad indices or difficult markets.
- Sampling holds a representative subset and adds model and deviation risk.
- A swap adds counterparty risk, mitigated under UCITS through limits, collateral and resets but not eliminated.
- A synthetic ETF’s substitute basket can differ from the index and should be checked in disclosures.
- Securities lending may also be used by physical ETFs and creates its own risks and revenue-sharing questions.
This is general educational material, not personalised financial advice. Goals, taxes, time horizon and capacity for loss differ from one investor to another.
Related guides
- ETFs explained: how they work, costs and risks
- Accumulating and distributing ETFs
- ETF TER and costs it misses
- Physical and synthetic replication
- Tracking difference and tracking error
- Portfolio diversification
- Dollar-cost averaging
