Physical vs synthetic ETFs: how replication methods affect cost and risk

Physical and synthetic ETF replication are the two approaches an ETF issuer can use to track an index, and the choice between them affects cost, tracking precision, and the risk profile the investor takes on. Physical replication holds the actual underlying securities; synthetic replication uses derivative contracts—typically total return swaps—to replicate index performance without owning any of the underlying assets.

What ETF replication is

An ETF is designed to track an index or other benchmark, and replication method describes how it does so. For equity ETFs tracking well-known indexes, two primary approaches exist.

Physical replication means the ETF holds the actual securities in its index. Full physical replication holds every constituent in proportion to its index weight. For a large, liquid index such as the S&P 500, full replication is straightforward and cost-effective. For a broader index with hundreds or thousands of smaller, less liquid constituents—such as a global small-cap index—full replication becomes expensive and operationally complex. In these cases, ETF issuers use optimised or stratified sampling: holding a representative subset of the index that mimics the overall risk and return characteristics without including every security.

Synthetic replication does not hold any of the underlying index securities directly. Instead, the ETF enters into a total return swap with a counterparty—typically an investment bank. The counterparty agrees to pay the ETF the total return of the index (dividends and price changes), and the ETF pays the counterparty a fee plus the return on a collateral basket, which may consist of securities unrelated to the index. The ETF investor therefore owns shares in a fund whose return is determined entirely by the terms of a derivative contract.

How it works

In a full physical ETF, the fund buys and sells securities as the index rebalances, or as the creation and redemption mechanism requires. The fund directly owns the underlying assets, and its NAV reflects the market value of those holdings. Tracking error—the deviation between the ETF's return and the index return—arises primarily from transaction costs, dividend timing differences, withholding tax treatment, and securities lending income (which can partially offset costs).

In a synthetic ETF using a total return swap, the ETF's return is contractually specified. The counterparty delivers exactly the index return to the ETF on the swap settlement date. Synthetic ETFs can therefore achieve very low tracking error under normal conditions, since the return is locked in by contract rather than depending on the operational precision of portfolio management.

Synthetic ETFs using the funded swap structure hold only the collateral basket on their own balance sheet. In an unfunded swap structure, the ETF may hold no assets at all—the entire NAV is effectively a receivable from the counterparty. The collateral held against this exposure varies in quality and liquidity across providers.

What the evidence shows

Studies of European-listed ETFs tracking commodity indexes and emerging market equity indexes have found that synthetic replication generally produces lower tracking differences, partly because synthetic ETFs avoid the direct trading costs and securities lending friction that physical ETFs must absorb. This advantage narrows or disappears for liquid equity indexes where physical replication costs are already minimal.

The structural risk of synthetic ETFs became a concern during the 2008–2009 financial crisis, when counterparty defaults became a realistic scenario. In the European Union, UCITS regulation now limits counterparty exposure on unfunded swaps to 10% of NAV, and most synthetic ETF providers maintain collateral buffers and over-collateralisation to reduce this risk further. Nonetheless, the credit risk is structural and cannot be eliminated entirely in a synthetic structure.

Limitations and trade-offs

Physical ETFs introduce tracking error from transaction costs, rebalancing frequency, and dividend withholding tax treatment differences between the fund's domicile and the index's assumed treatment. Optimised-sampling physical ETFs accept a degree of tracking imprecision in exchange for lower operational costs. Securities lending—lending holdings to short sellers in return for a fee—can partially offset fund costs but introduces additional counterparty risk and complexity for the investor to consider.

Synthetic ETFs eliminate most sources of tracking error but introduce counterparty risk. If the swap counterparty defaults, the ETF investor's claim is limited to the value of the collateral basket held by the fund. The quality, liquidity, and correlation of that collateral with the index being tracked varies by fund and provider. For investors in jurisdictions with UCITS protections, the regulatory limits on counterparty exposure provide a partial safeguard, but the risk is not zero.

For straightforward exposures to liquid equity indexes, the practical difference in cost and tracking quality between physical and synthetic ETFs is small, and either approach is defensible. The choice matters most for hard-to-access indexes—commodities, frontier markets, or leveraged and inverse strategies—where synthetic replication is sometimes the only practical option.

ETF replication in pfolio

pfolio supports both physically and synthetically replicated ETFs within its investable universe. When selecting ETFs for a portfolio in pfolio, replication method is one of several characteristics worth reviewing alongside expense ratio, tracking difference, and liquidity. The fund's replication approach is typically disclosed in its key investor information document (KIID) or factsheet. ETFs available in pfolio are listed in the Assets section, and performance data is available via pfolio Insights.

Related articles

• ETFs explained: what exchange-traded funds are and how they span every asset class
• ETF structure explained: how exchange-traded funds are created and redeemed
• Total expense ratio: how ETF costs affect long-term portfolio returns
• Tracking error and tracking difference: how to measure ETF replication quality