Providers can use either physical or synthetic replication to ensure their ETFs mimic their designated indices as accurately as possible.
Nick Blake, head of retail at Vanguard, explains physically-backed ETFs follow their strategy by trading the underlying shares, whereas synthetic ETF aim to achieve the same result using derivatives.
Morningstar’s European passive fund analyst team states: “In constructing a physical replication ETF, the provider simply builds a portfolio of all the underlying assets of the index - or a select subset thereof... [known as] ‘optimsed sampling’.
“Synthetic replication is done through the purchase of a swap contract... usually [from] an investment bank. Typically the latter party agrees to provide the index return to the fund in exchange for a fee. To protect the fund from counterparty risk, the fund takes collateral.
“Since physical replication involves buying and selling index components, this strategy is inherently more labour intensive and costly than synthetic replication... [and] may also exhibit larger tracking error.”
Mark Johnson, head of UK sales at iShares, argues physical replication can actually be split into two distinct categories, with full replication strategies distinguished from ‘sampling’.
Full replication is the most common and is seen to be the most reliable approach, but he says it may not be appropriate in larger, global markets where the time, effort and expense involved in conducting all transactions for complete replication could be “disadvantageous to the portfolio”.
‘Sampling’ techniques have the advantage of reducing administrative and transaction costs, but come with the possibility of slightly higher tracking error, as the fund does not hold all the securities in its benchmark.
Ben Thompson, director of business development, listed products and ETF UK for Lyxor, suggests that “purely physical funds” work well for large, liquid markets. However, in the case of very large indices such as the MSCI World index, which has more than 1,600 constituent stocks, “buying all the stocks is inefficient”.
He argues that the difference between physical and synthetic ETFs is in princple not as great as many might say. Both, he says, start with a portfolio of physical assets which is owned by the fund and held in a segregated account - but where one uses the assets for performance, the other uses them for security.
Synthetic ETFs, he says, are based on contractual swap agreements where a counterparty commits to pay the precise daily performance of the benchmark index, including any dividends, to the fund. In return, the ETF pays the counterparty a fee and the performance of the physical assets it holds, including any dividends.
Before taxes and replication costs the daily performance of the ETF is a more precise replication of the benchmark index, but the stratgegy involves additional risk for investors related to the counterparty.
For this method, Mr Johnson said the fund provider uses a swap agreement with an investment bank, which is a deal whereby the latter delivers the returns of the index in exchange for certain assets.