In theory, the overwhelming majority of ETFs are very simple creatures. They are passive vehicles, designed to replicate the return of a diversified index of securities, whether that be equities, bonds or something more outlandish.
In practice, though, the more closely one looks under the hood, the more complex these beasts really are.
As a result, an ETF will rarely deliver precisely the return of its target index. In most cases it will underperform its benchmark, although outperformance is not unheard of. The divergence is known as the tracking difference.
Most obviously, managers of ETFs levy an annual management charge, although this will typically be much lower than for actively managed products, and a handful are actually free.
This, together with other small charges such as auditing, legal and custodian fees and the cost of licensing the underlying index, comprises the total expense ratio, the “price tag” of the ETF.
If the TER is 0.3 per cent, a novice investor might expect an ETF to undershoot its target by 30 basis points every year.
However, matters are not that simple because, despite its name, the total expense ratio does not include all expenses.
Every time the target index is rebalanced, such as with the quarterly rejigs of the UK’s FTSE 100 share index, an ETF needs to buy and sell securities to maintain accurate replication. All this activity inevitably incurs transaction costs,
“This will cause a drag on investors’ returns, since the index performance does not reflect these costs,” said Anaelle Ubaldino, a quantitative financial adviser at Koris International, a French investment advisory firm linked to TrackInsight, the FT’s partner for the ETF Hub.
So-called “smart beta” ETFs, which are not based on market capitalisation-weighted indices, will have to trade more often, incurring higher transaction costs. For instance, an equal-weighted ETF will need periodically to rebalance, selling stocks that have risen in value and buying those that have fallen.
Tracking difference can also result from discrepancies between the holdings of the ETF and its target index.
This can result from “sampling”, which occurs when there are so many securities in an index it is unrealistic for the ETF to hold them all. Instead, the manager will create a portfolio designed to be representative of the underlying index but which may differ in performance to some extent. This divergence can be either positive or negative.
Sampling is often seen in bond funds, which can contain thousands of instruments, some of them lightly traded.
Other factors can also cause a divergence between the underlying holdings of the index and the ETF.
“When the index holdings are spread across time zones and currencies, or when they are not readily available to trade on the secondary market, it is harder for the ETF issuer to replicate the index accurately,” said Ms Ubaldino.
Cash drag is another potential source of a discrepancy in returns between the ETF and its benchmark. This occurs when dividend or coupon payments are reinvested later than the index accounts for.
This results in the ETF holding cash and being under-leveraged for a period, something that is beneficial for investors when the underlying index is falling but will lead to a degree of underperformance when it is rising.
One practice that is almost always a positive is securities lending, a fairly common practice among ETF issuers. The ETF manager is often able to lend some of the underlying holdings, for a fee, to a hedge fund or other institution that wishes to borrow them in order to short the security in question.
In theory this entails a small degree of risk for the ETF, of the borrower being unable to return the security at the agreed date and the collateral it lodged with the ETF being insufficient to cover the cost of repurchasing the holding in the market. However, the risk is slight and ETF providers instead reap a small, but steady, income stream to offset against other costs.
The importance of tracking difference may at first seem minimal. However, small deviations in this measure can add up over time.
The example above shows the performance of two ETFs and their benchmark index. Based on expense ratio only, investors would be indifferent between the two funds, as they both stand at 50 basis points, or 0.5 per cent, a year. However, looking at their tracking difference, investors would find that ETF A has generated better overall performance than ETF B.
“Tracking difference is critically important when selecting passive ETFs,” said Ms Ubaldino. “It paints the full picture of how well an ETF tracks its index and should thus be investors’ preferred metric for ETF comparison.”
Get alerts on Exchange traded funds when a new story is published