ETFs & Funds · Intermediate · 4 min

Replication: physical, sampling, synthetic

You can explain the three ways an ETF copies its index and name the risk each method adds, especially counterparty risk in swap-based ETFs.

Same index, different inside

Two ETFs track the same index. Their returns are almost identical. Yet one owns every share in the index, and the other may own none of them.

How an ETF copies its index is called replication. It is written in every factsheet, usually in one short line that most investors skip. But it decides what the fund actually holds and which risks come with it.

Full, sample, swap

Physical replication means the fund buys the securities in the index.

Full replication: the fund holds every stock, in the same weights as the index. This is simple and transparent, and works well for indices with a manageable number of liquid stocks.
Sampling: for indices with thousands of holdings, many of them small and illiquid, the fund buys a representative subset that behaves like the whole. This saves trading costs, but the fund can drift a little further from the index, so the tracking error is usually higher.

Synthetic replication works differently. The fund holds a basket of securities, which can differ from the index, and agrees a swap with a bank: the fund pays the return of its basket and receives the return of the index. The basket serves as collateral.

This adds counterparty risk: if the bank fails, the swap stops paying, and the fund falls back on its collateral. EU UCITS rules limit this risk. For derivatives that are not cleared through a central counterparty, exposure to a single credit institution may not exceed 10% of the fund's assets, and 5% for other counterparties. In practice, the swap is reset regularly so the exposure stays within those limits.

Synthetic ETFs can track some indices more precisely or cheaply, for example markets that are hard for foreigners to access. Physical ETFs often lend out securities to earn extra income, which carries a similar, smaller counterparty risk.

Collateral covers the gap

A fictional synthetic ETF holds a collateral basket worth 95 million. Its swap with a bank promises the index return. The index rises, and the fund's total value reaches 102 million: 95 million in collateral plus 7 million owed by the bank under the swap.

Then the bank fails. The fund keeps the 95 million basket, but the 7 million owed under the swap is at risk until the bank's insolvency is settled. That 7 million is the counterparty exposure, and under UCITS rules it must stay within 10% of the fund.

Common mistakes

Assuming every ETF holds its index. Check the replication method.
Treating synthetic as unsafe by default. It is a different risk, limited by rules, not automatically a bad one.
Ignoring securities lending. Many physical ETFs lend shares and carry counterparty risk too.
Judging by the TER only. Replication affects tracking difference as well.

The takeaway

ETFs copy an index by holding all its shares, a sample, or a swap. Each method has trade-offs in cost, accuracy and counterparty risk.

Check yourself

Quick quiz

4 questions. Instant feedback, nothing is saved.

1Tap the method that depends on a counterparty paying the index return.

Tap the chart.

A synthetic ETF receives the index return from a swap counterparty, usually a bank. If it fails, the fund relies on its collateral.
2Under UCITS rules, how large may the exposure to a single bank as OTC derivative counterparty be at most?

Article 52 of the UCITS directive caps it at 10% for credit institutions and 5% for other counterparties, for derivatives not cleared centrally.
3Why would an ETF on a global index with thousands of small stocks use sampling?

Sampling saves trading costs but can increase the tracking error.
4What does a synthetic ETF actually hold?

It holds a collateral basket and swaps its return for the index return. The basket can look quite different from the index.

of 4 correct.