In a recent report, analysts from Morgan Stanley note that the average tracking difference for US ETFs rose from 52 basis points in 2008 to 125 basis points in 2009. When weighted for fund size, tracking difference rose from 39 basis points to 113 basis points across the US exchange-traded fund universe.

The proportion of funds exhibiting meaningful divergence from their benchmarks – defined as an annual performance difference of over 1% – rose from 13.5% in 2008 to 37.5% in 2009, according to the report’s authors.

[IU.eu note: we use the most commonly accepted definitions of “tracking difference” as the absolute difference between the total return of a fund and that of the index it is tracking, and “tracking error” as the variability of fund return vs. index. As such, a fund exhibiting a straight-line divergence in performance against its benchmark will have tracking difference but no tracking error.]

Beyond the most widely accepted sources of tracking difference – fund fees and expenses, divergences resulting from the use of portfolio optimisation techniques (when a fund holds a sample of index securities rather than all of them), and costs relating to index changes – there are some more idiosyncratic reasons for the rise in tracking problems, according to Morgan Stanley.

SEC-mandated diversification requirements have caused difficulties in replicating certain indices, they argue, while the push by exchange-traded funds into more exotic overseas equity markets has also made it more difficult for some ETFs to replicate their benchmarks.

More broadly, Morgan Stanley says, “the high level of new issuance over the past few years has broadened the range and complexity of ETFs. Some ETFs follow newer or less well-known indices that may be harder to track.”

In Europe, the large number of swap-based ETFs should, in principle, result in a lower overall average tracking difference than in the US market, suggests one ETF market observer. “ETFs in the US use physical replication, which naturally causes higher tracking differences, whereas over here in Europe swap-based providers guarantee the index return (before fees), so you would expect the tracking error to be smaller.”

Based on a sample of European ETFs’ performance, tracking difference certainly appears to be lower in Europe.

In a universe of around 200 ETFs, provided to IndexUniverse.eu by issuers iShares, db x-trackers, Amundi, Credit Suisse and ETFlab, the average (unweighted) tracking difference between fund and benchmark was 40 basis points in 2009, an increase from an outperformance of 2 basis points in 2008 (i.e., this sample of funds produced a higher-than-benchmark return, on average, in that year). Both annual averages are significantly lower than the equivalent figures for the US market.

Incidentally, there is an often-overlooked reason for the increase in tracking differences in strongly positive years for the market (such as 2009), according to Eric Wiegand of db x-trackers.

As an ETF’s manager deducts the management fee from the underlying assets daily by making a percentage adjustment to the fund’s net asset value, a significant rise in that value will make the absolute deduction larger. This, in turn, means that the actual fee deduction over the course of the year will be larger than implied when multiplying the stated expense ratio by the value of the fund at the start of the period. In turn, tracking differences will also seem larger when compared to the starting point, he explains.

Wiegand gives the example of the db x-trackers’ MSCI Taiwan total return ETF, which returned 78.14% in 2009, compared to an index return of 79.21%. On an absolute basis the tracking difference was 107 basis points, but on a relative basis it was just over 60 basis points (expressed mathematically, 1.7921/1.7814=1.0060).

Equally, Wiegand says, when measured on an absolute basis, tracking differences will be understated in years when markets fall.

However, is the apparently superior tracking ability of European ETFs when compared to their US counterparts simply down to the replication technique?

In 2009, the unweighted average tracking difference on a sample of 45 iShares funds, all using physical replication, was 58 basis points. The same figure for 68 db x-trackers funds using swap-based replication was 53 basis points. For Credit Suisse’s Xmtch ETFs, which use physical replication, the average 2009 tracking difference was 69 basis points.

Amundi’s swap-based ETF range showed an unweighted average tracking difference of 7 basis points in 2009, whereas German issuer ETFlab, which uses physical replication across its range, produced an unweighted average outperformance of 10 basis points in a sample of ten funds.

1 | 2 |