Generally speaking there are two ways how an ETF can replicate its underlying index. The first is the so-called physical replication, where the ETF holds the actual securities in its portfolio. The second is the so-called synthetic replication where the ETF normally holds a total return swap and a set of in-kind securities as collateral. Both replication methods have their strengths and weaknesses which will be unveiled in this article.
ETFs that use physical replication can do so in two different ways:
Full Replication: The ETF holds all securities of the underlying index in the exact proportions as the index.
Sampling (or Partial Replication): The ETF holds a representative sample of the securities in the index with the aim to replicate the risk/return profile of the index as close as possible by using the lowest possible number of securities.
While full replication is used for indices with a smaller number of highly liquid securities, sampling is used for indices with a high number of securities and/or indices which contain securities with a limit liquidity.
Conversely, ETFs that use synthetic replication normally execute a total return swap based on the respective index and a portfolio of in-kind securities as collateral for the swap. In-kind securities are securities from the same asset type as the underlying index, but they don’t have to be in the index at all. Most ETFs hold a portfolio of very liquid securities as collateral for the swap. In some cases, futures or other derivative contracts are used instead of a total return swap.
In addition to this, there are also some ETFs who can use one or the other and in rare cases even a mix of replication techniques to achieve their goals as per their fund prospectus.
Advantages:
ETFs using full replication are easy to understand since they are highly transparent and are very simple products since they hold the securities of the index with the same weightings as the index. Conversely, ETFs using sampling are partly losing this advantage since the portfolio is made from a sample of securities which may not match the securities and their weightings in the respective index.
On the other side, ETFs using synthetic replication may have a low tracking error since the swap replicates the risk/return profile of the index. Synthetic ETFs should also be more cost efficient since there are no transaction costs for the ETF when the index is rebalanced. In addition, synthetic ETFs can provide investors with access to asset classes which can’t be accessed with physical replication such as (soft) commodities or some more exotic securities markets.
Disadvantages:
While the low transaction costs for the ETF are seen as an advantage for synthetic ETFs, they are a disadvantage for ETFs using physical replication. In addition, the tracking error for ETFs using physical replication, especially for those using sampling, can be significantly higher than for synthetic ETFs.
The high transparency and simplicity are a clear advantage for ETFs using physical replication, while the complexity of derivatives within their portfolio is seen as a disadvantage for synthetic ETFs.
Even as the counterparty risk from the swap contract is often seen as a disadvantage for synthetic ETFs—investors need to be aware that ETFs using physical replication often utilize securities lending programs to enhance the returns of the portfolio, which bear the same kind of risk.
With regards to the above, it can be concluded that all common replication techniques used by ETFs are suitable for replicating the respective indices as accurately as possible. Nevertheless, it should be noted that all replication techniques have their pros and cons and it is the duty of the investor to chose the replication methodology which suits his needs best.
You can find an overview of the use of the different replication methodologies in our ETF Yearbook 2024 (click here)
This article is for information purposes only and does not constitute any investment advice.
The views expressed are the views of the author, not necessarily those of Lipper or LSEG.
