The low cost of exchange-traded funds relative to traditional actively managed open end funds and index trackers is perhaps their most appealing feature. When looking at the cost savings of ETFs, and comparing one fund to another, most of us just look at the total expense ratio given in prospectuses and other literature. But expense ratios reflect just one of the many costs of ETF ownership. Costs of buying, selling, and potential hidden charges or earnings all affect the ultimate returns that shareholders get from a given ETF. Here, we will piece together all of the costs involved in transacting and owning ETFs to give a holistic view of the total cost of ETF ownership.
Total Expense Ratio (TER)
The total expense ratio is the most explicit cost of ETF ownership. This ratio represents the portion of your investment that will be extracted by the fund on an annual basis for their efforts. This charge covers costs incurred by the fund which can range from trading expenses to marketing budgets and custodial fees to index licensing costs. The average expense ratio across all European ETFs is 0.38%, which compares very favourably to average expense ratios for both traditional actively managed open end funds and open end index trackers. For more on total expense ratios please see our article Low Costs are the Smartest Investment.
Trading Costs
We first discussed the costs of trading ETFs in our article Tricks of the Trade. These costs include brokerage commissions, bid-offer spreads, and market impact (a topic we will touch on in further detail in the future). Furthermore, premiums and discounts to net asset value represent another important consideration (for more on this topic see this article).
Tracking Difference
Tracking difference is an implicit cost of ETF ownership. All else equal, ETFs will by definition produce returns that lag their benchmark by an order of magnitude equal to their TER. For example, assume an investor owns an ETF that tracks the EURO STOXX 50 index. The fund charges a hypothetical TER of 0.25%. In a given year the EURO STOXX 50 rises 10%. The investor’s return—net of fees—would be 9.75% (the 10% gross return to the benchmark less the 0.25% TER). In this hypothetical example, the tracking error for the ETF is equal to the TER.
Tracking Difference = ETF Return – Benchmark Index Return
Expense ratios are the most predictable and readily quantifiable source of the tracking difference between ETFs and their reference indices. It is important to understand that ETF providers can offset some of the effect fees have on ETFs' net performance through what providers term “enhancements,” which usually take the form of income produced through securities lending. In some cases, this level of income has more than offset the level of certain funds’ expenses and allowed them to even perform modestly better than their benchmarks.
Tracking Difference vs. Tracking Error
It is important to distinguish between tracking difference and tracking error. Tracking difference is simply the difference between an ETF's return and that of its benchmark over a given period.
Meanwhile, tracking error measures the short-term volatility of an ETF's return relative to its benchmark. Tracking error is commonly calculated by taking the root mean square error of the difference between a fund's daily or weekly returns and those of its benchmark across a given period.
Sampling in Physical Replication Funds can Create Tracking Issues
Headline expenses and securities lending revenue are not the only cause of tracking difference for ETFs. The manner in which funds seek to replicate their benchmark can also be a substantial source of tracking difference. For instance, a physical replication fund that tracks a benchmark containing a number of smaller, less liquid components (like the MSCI Emerging Markets Index, or fixed-income indices that often contain more than 1,000 component bonds) may use “sampling” techniques to replicate the returns of its reference index. Sampling involves investing in a select basket of only the largest and most liquid components of the benchmark index in an effort to improve the overall liquidity of the fund itself (it makes the ETF shares easier and cheaper for market makers to hedge or create) and to minimise costs. While sampling has some obvious advantages, by virtue of excluding some of the smaller, less liquid components of a fund’s reference index, it creates another potential source of tracking difference as the fund strays from perfectly mirroring its benchmark.
A useful case study in the potential for sampling to create tracking issues is the US-listed iShares MSCI Emerging Markets Index ETF. This fund uses a representative sampling method to track the index, so it doesn't own every security in the benchmark. Rather, it invests directly in the majority of the securities (this fund holds about 625 stocks out of approximately 770 in the index) and then samples a quantitatively representative portion of the smallest and least-liquid equities. Prior to 2010, this fund would hold only about half of the index components (the more-liquid names) in order to facilitate trading. For this reason, as well as the fund's relatively high expense ratio, the ETF’s 2009 net asset value return trailed its benchmark index's return by nearly 7 percentage points. We expect the fund will narrow its tracking difference in 2010, as it now holds a greater number of its reference index’s components.
Meanwhile, Vanguard's US-listed Emerging Markets Stock ETF also tracks the MSCI Emerging Markets Index. In contrast to the iShares fund, the Vanguard ETF employs full replication. In 2009, its total NAV return was just over 76%--much nearer the return of its benchmark.
Index Turnover
Index turnover costs reflect another potential source of tracking problems. While index investing is an inherently passive strategy, the fact of the matter is that indices do experience turnover. Strategies such as fundamental indexation or equal-weighting, or indices that only contain a portion of the market such as value, growth, and high-dividend ETFs, all tend to have even higher turnover. Even among market-weighted indices, bankruptcies and mergers and acquisitions can change an index’s composition and force rebalancing trades. The costs involved in realigning the portfolio of a physical replication fund to reflect these changes may ultimately show up in the form of a larger tracking difference.
Dividends The timing and tax treatment of dividend payments are two additional potential sources of tracking troubles for ETFs. The lag between physical replication distributing funds' ex-dividend and dividend payment dates (typically a few weeks) can potentially weigh on the fund’s performance. During this span the pending distribution amount is segregated from the fund and sits in an interest bearing account. As a result, investors lose benchmark exposure over this span equal to the pending distribution, creating the potential for increased tracking error. It is important to note that accumulating funds and synthetic replication funds are not affected by the timing of distribution payments, as these funds remain 100% invested in their benchmarks at all times--albeit in different ways, one through physical holdings and the other in the form of a total return swap.
Dividend taxation is another important consideration. When funds receive dividends from securities located in multiple tax jurisdictions outside their domicile, the associated foreign dividend withholding taxes can create an additional drag on fund performance. These withholding taxes cannot always be fully reclaimed, and when they are clawed back it is usually at the expense of time and effort. The synthetic replication structure has shown that it can manage the withholding tax issue a bit more nimbly than physical funds, receiving more favourable tax treatment on dividend income.
Commodity Costs
Many exchange traded commodities (ETCs) and exchange traded notes (ETNs) will charge a range of fees in addition to their headline expense ratios. These include custody and storage fees (unique to physical precious metals funds), collateral costs, and index licensing costs. Collateral charges reflect the cost of holding collateral to reduce counterparty risk in synthetic ETC and ETN structures. Deutsche Bank research shows that these charges averaged 0.40% for ETCs and 0.23% for ETNs as of March 11. Index licensing costs are fairly self-explanatory. They represent the payments made to index providers for use of their intellectual property (the benchmark index) in constructing the funds. Deutsche Bank estimates index licensing fees cost ETC providers an average of 0.38% per annum. These fees are rarely included in the total expense ratio, and may not even explicitly appear in the prospectus, so investors in these products should be wary of the hidden costs.
Putting the Total Cost of ETF Ownership in Perspective
While the total cost of ETF ownership extends well beyond the headline expense ratio, ETFs are still an inherently low-cost vehicle relative to the broader investment menu. Especially when considering some of the largest and most liquid funds, those that represent some of the best building blocks for a core asset allocation strategy, ETFs on an all-in cost basis are perhaps the low-cost leader. Take the db x-trackers EURO STOXX 50 (Capitalising) ETF, for example. This synthetic fund has a TER of 0%, is one of the most liquid ETFs on the exchanges where it trades, and has actually generated a positive tracking difference relative to its benchmark since its inception through the end of April (it has outperformed the EURO STOXX 50 index by 46 basis points over that period). So while it is important to keep in mind all of the costs involved in ETF ownership, both explicit and implicit, it is also important to remember that the largest, most liquid funds—which also tend to sport the lowest expense ratios—are still very inexpensive to own on an all-in basis.
Patricia Oey contributed to this article.
