For an index-tracking fund like an ETF, there are three main choices to make when it comes to how a fund manager is going to copy (replicate) the particular index being tracked:
- Physical – Full replication
- Physical – Stratified sampling
- Synthetic – Using derivatives
There are a number of reasons why a passive fund manager will choose each of these options. Let’s look at them briefly in turn.
1. Full physical replication
Full physical replication is a 100% copy of the index being tracked. It is the simplest method, but it is sometimes costly.
Full physical replication of an index means the following: If there are 100 stocks in an index like the FTSE 100, then an index ETF using a full physical replication method will buy all 100 stocks in the underlying index in exactly the correct quantities required to mimic the weightings of each stock in the overall index. For a benchmark index like the FTSE 100, Euro STOXX 50 or US S&P 500, this is the method that is usually used.
Most transparent, easiest to understand
It is very WYSIWYG, to use a technology acronym. That is, What You See Is What You Get. It is thus the most transparent and easy to understand way for an index fund to copy the underlying index it is tracking. It also results in performance (net of management fees) that is closest to the performance of the actual index, as calculated by the official index provider (i.e. FTSE, STOXX or S&P Dow Jones Indices).
Potentially it is more costly for the client
However, it can prove to be more expensive than the other two methods of copying an index, given that the amount of buying and selling is potentially the greatest, and clients can thus be charged higher management fees as a result. As the ETF has to exactly replicate the constituents of the index it is tracking, when that index is rebalanced or reconstituted (in the case of the FTSE 100 this is every three months) it has to adjust its holdings accordingly.
There is also the issue of “slippage”, which is the cost of buying and selling a share when there is not that much liquidity. Effectively, the spread between the buying and selling price for a stock can be quite wide due to a lack of liquidity and the size of the buy or sell order – this amounts to an extra cost for the fund and thus for the final client.
2. Physical stratified sampling
So, when the stock index to be copied has a very large number of index member stocks in it, not 100 but perhaps 2000 or more, or when some of the members of the underlying index are not very liquid, a passive fund manager may instead opt to use stratified sampling to construct the index fund.
Copy the largest members, sample the rest
Stratified sampling means that instead of buying the exact quantities of each share as per the weightings in the underlying index, the fund manager decides only to exactly copy the largest index members in the fund, and then chooses a reduced, representative sample of the remaining, typically smaller, companies to buy. A sophisticated computer program is used to make sure that the resulting smaller list of companies held in the fund nevertheless copies the performance of the overall, full index.
The benefit of this is cost: there are fewer companies to buy or sell each time the index is changed or rebalanced, thus lower trading costs. There is also reduced loss of fund performance through “slippage”, as typically this stratified sampling technique will favour the more liquid, and thus cheaper to trade, companies.
Downsides of stratified sampling
The risk is that this smaller group of companies does not perfectly replicate the performance of the full index: there may be more “tracking error”, that is to say a difference over time between the fund’s performance and the performance of the full index. So this method of copying an index is not as transparent as full replication and has a risk that the fund’s performance deviates from that of the index it is meant to be following.
That said, with stock index funds this tracking error is usually not that significant. However, when we look at index ETFs in other asset classes where liquidity is not as high, such as in corporate bonds, then this tracking error can be quite significant. So for these types of index ETFs, it is certainly worth bearing this risk in mind.
3. Synthetic replication
With synthetic replication, the fund manager does not try to buy the underlying members of the index being copied in physical form: instead, a derivative based on the underlying index is bought, typically from an investment bank, which is a contract where the investment bank promises to pay the fund manager the exact performance of that underlying index.
So in a simple example case, if the FTSE 100 gains 15% including dividends over the course of a calendar year, then the investment bank concerned will pay the fund manager that amount in cash according to the derivative contract between the fund and the investment bank.
In this case there are no issues of having to buy and sell lots of different company shares every time the index is rebalanced, because there is just one trade that needs to be made – the derivative contract between the fund manager and the bank for the precise amount needed for the fund to have 100% exposure to the underlying index.
Sometimes, synthetic replication is the only way
For stock index ETFs, a fund has a choice of which index replication method to use. However, in other asset classes like commodities, there is often no choice but to use synthetic replication, as there is no underlying physical asset that you can reasonably buy.
This is particularly the case with crude oil, for instance. How would a fund manager realistically store potentially millions of barrels of oil without incurring huge storage costs? So a crude oil ETF will inevitably use synthetic replication via buying a crude oil future derivatives contract, as there is no other economically viable way to get economic exposure to crude oil.
No tracking error or slippage, but counterparty risk
With synthetic replication, there are no issues of tracking error or of trading costs to the fund from slippage, which can make synthetic replication a much cheaper way to copy the performance of the underlying index.
But using synthetic replication a fund may be exposed to a risk called “counterparty risk”. This refers to the risk that the fund may not receive the performance of the index from the bank as per the derivatives contract, if that bank goes bankrupt. It is a small, but non-negligible risk to be aware of. So it is very important with synthetic replication that the bank selling the derivatives contract to the index fund has very solid profitability and a robust balance sheet, so that this counterparty risk is close to zero.