ETFs invest in shares or other assets to track the performance of a stockmarket or bond index, or the price of a commodity such as gold.
Typically this involves the ETF buying each of the securities in the index according to their percentage weight within that index.
So, if BP is 3% of the UK’s FTSE 100 index, a FTSE 100 ETF would invest 3% of its portfolio in BP shares, and so on for the other shares in the FTSE 100.
Having copied its make-up, the ETF then naturally mirrors the ups and downs of the index.
An ETF that buys all the securities in an index according to their weights is described as using full replication.
Some ETFs, particularly, those tracking indexes containing thousands of securities (such as the MSCI All Country World Index) may seek to match the index performance by investing in a selection of securities rather than all of its constituents. This is variously called sampling or partial replication or optimised replication.
All the above are also termed physical ETFs because they are buying actual shares or other assets of the index they are tracking.
Alternatively, with synthetic or swap-based ETFs, the ETF doesn’t directly invest in the index’s securities. Instead, it has an agreement with an investment bank in which the bank promises to pay the ETF the return on the index, in exchange for a fee.
Synthetic ETFs can be useful for tracking markets that are more difficult or costly to invest in directly. However, because there is a swap transaction with a third-party bank, there is a potential downside from so-called counterparty risk.