Physical vs. synthetic ETFs: Which replication method is better?

Many investors first check whether an ETF uses "physical" or "synthetic" replication. This can quickly lead to a simple conclusion: physical replication sounds tangible and safe, while synthetic replication sounds complex and therefore riskier. However, this distinction is too simplistic.

ETF stands for exchange-traded fund. An ETF is an investment fund that is traded on a stock exchange. Many ETFs aim to track the performance of an index as closely as possible. How successfully they do so depends not only on the label attached to the replication method, but also on costs, risks, market structure and the way the method is implemented.

• What does replication mean for an ETF?
• How do full physical replication and optimised sampling work?
• How does a synthetic ETF track an index using a swap?
• Is a physical ETF safer than a synthetic ETF?
• How can investors assess the quality of ETF replication?

Replication means tracking or reproducing an index. An index-tracking ETF aims to mirror the performance of its underlying index as closely as possible. To do so, the fund must translate the index's calculated performance into the return generated by the fund's own assets.

A simplified example helps illustrate the difference. If EUR 100 is invested in a physically replicating ETF, the fund generally invests the money in the securities included in the index. If one company has a 5% weight in the index, this would correspond to EUR 5 in the simplified model.

A synthetically replicating ETF pursues the same objective in a different way. It enters into a financial contract with a bank, known as a swap, through which the fund receives the index return. This service creates costs within the swap structure.

Both methods aim to do the same thing: track the index. The difference is whether the fund holds the index constituents directly or receives the index performance through a contract.

With full physical replication, the ETF generally buys all the securities in the index in line with their respective index weights. When the composition of the index changes, the portfolio must be adjusted accordingly. The method is straightforward to understand but may require a large number of individual transactions.

However, not every physically replicating ETF holds every index constituent. A broad international index may contain numerous securities across different countries, exchanges and currencies. Limited tradability, currency restrictions, low liquidity or other operational requirements can make full replication difficult or expensive.

Optimised sampling is an alternative. The ETF invests in only a selection of the index constituents. This selection is designed to reflect the index's main characteristics closely enough for the fund to perform in a similar way.

A physical ETF tracking a broad index such as the MSCI World does not therefore have to hold every single stock in the index. "Physical" means that the fund buys securities directly. It does not necessarily mean that the ETF replicates the index in full, position by position.

Sampling can reduce operational effort. At the same time, the selected portfolio must be managed so that deviations from the index remain limited. The actual quality of replication is therefore more important than the label alone.

A synthetic ETF usually uses a swap. A swap is a financial contract under which cash flows are exchanged between the fund and a counterparty, typically a bank. Economically, the arrangement is designed to deliver the performance of the target index to the ETF.

Three concepts are central to this structure: the substitute basket, the swap and the counterparty.

The substitute basket consists of the assets that the fund actually holds. These assets do not have to match the constituents of the index being tracked.

The swap specifies which index return the counterparty will deliver to the fund. Depending on the structure, the capital invested may remain in the fund and be invested in other assets. Under another arrangement, capital may be transferred to the counterparty as part of the swap structure.

The substitute basket of a synthetic ETF may therefore look very different from what the name of the tracked index suggests. What matters is that the substitute basket and the swap together generate the economic index exposure.

This structure can offer advantages in hard-to-access markets, certain bond segments, very broad indices or other operationally demanding areas. Direct purchases of securities may be more complex, more expensive or subject to restrictions. This does not make synthetic ETFs universally superior, because swaps also involve costs and specific risks.

The statement "physical is safer" sounds plausible but overlooks important details. The two replication methods have different risk drivers. Investors should therefore consider not only which securities an ETF holds, but also any additional transactions and contractual relationships.

Where a fund is structured so that its assets are legally segregated, the fund's assets must be kept separate from the management company's own assets. This segregation does not protect investors against market losses on the assets held. Nor does it automatically remove every risk arising from additional transactions or claims.

In a physically replicating ETF, the securities purchased directly form part of the fund assets. However, the fund may temporarily lend securities to other market participants. Securities lending creates the risk that a borrower does not return the securities on time or in full.

Securities lending can also be collateralised. Even so, it should not be ignored when assessing the risks of a physical ETF. A physical ETF is therefore not automatically free of counterparty risk.

In a synthetic ETF, the main counterparty risk arises from the swap. If the index in the simplified EUR 100 example rises by 5%, the bank may owe the fund EUR 5. If the bank defaults, there is a risk that this claim will not be met in full.

Collateral can reduce this risk. Suitable assets are provided to help secure the claim in the event of a default. However, collateral does not automatically eliminate the risk completely, because its quality, valuation, liquidity and legal enforceability also matter.

There is therefore no simple ranking. Physical ETFs may carry risks from securities lending, while synthetic ETFs carry counterparty risk through the swap. The actual level of risk depends on the specific fund structure, the collateral and how the arrangement is implemented in practice.

Many investors begin comparing ETFs by looking at the total expense ratio, or TER. It is important, but it does not capture every factor that ultimately affects the ETF's performance. Transaction costs, swap costs, securities lending revenue, tax effects and structural factors may also influence returns.

Tracking difference is therefore an important measure. It shows how far an ETF's performance has deviated from the performance of its underlying index over a given period. Because data providers may use different sign conventions, investors should check the definition used by the relevant provider.

Tracking error answers a different question. It measures how much the ongoing differences between the ETF and the index fluctuate over time. A small average difference and stable tracking are separate characteristics and are therefore measured using different metrics.

Both figures are based on historical data. They can provide evidence of past replication quality and the effective cost of tracking, but they do not guarantee future results. Comparisons are most meaningful between ETFs that track the same index.

Withholding tax and other structural effects can also influence replication. In a synthetic structure, the index performance delivered through the swap may, under certain conditions, differ from the outcome of a fund that holds the shares directly. This can improve tracking difference in some cases, but it is not guaranteed and cannot be generalised across every index, structure or period.

The underlying market remains important. Physical replication may be relatively straightforward for liquid shares in large companies in developed markets. In hard-to-access emerging markets, less liquid bond markets or specialised market segments, synthetic or partly synthetic implementation may offer advantages. These are possible relationships, not universal decision rules.

The replication method should therefore always be considered in the context of the specific index. Relevant factors include the breadth of the index, the liquidity of its constituents, any trading or currency restrictions and the costs of practical replication.

The necessary information is available in the product documentation. The factsheet, Key Information Document and prospectus should show whether the ETF uses full physical replication, sampling, synthetic replication or a hybrid approach. For synthetic ETFs, investors should also review information on the swap structure, counterparties and collateral. For physical ETFs, they should check whether securities lending is permitted.

Checking once before investing may not be enough. The replication method or other implementation features can change during the life of the fund. It is therefore sensible to review the latest documents periodically.

Assessing an ETF requires the full picture: TER, tracking difference, tracking error, replication method, risks, collateral, potential securities lending and the transparency of the documentation. "Physical" or "synthetic" is an important starting point, but not a final judgement on quality.

There is no single replication method that is better for every ETF. Physical replication is often easier to understand, but it may involve full replication, sampling and securities lending. Synthetic replication can provide efficient access to an index, but it also creates a contractual relationship with a swap counterparty.

What matters is how accurately and cost-effectively the ETF tracks its index and which risks are assumed and mitigated in the process. Investors should therefore not rely solely on the TER or on the labels "physical" and "synthetic".

A sound assessment compares several factors: the index, the market, tracking difference, tracking error, costs, collateral, securities lending and product documentation. Only this overall picture shows which replication method offers the more convincing implementation for the particular index.