ETFs carry substantial holding and transaction costs that are not immediately apparent. Learn more.
ETFs can have significant costs that aren’t entirely evident in expense ratios. From transaction to holding costs to ETF composition, the total costs of ETFs can be a significant drag on returns, which are coming under the microscope, as are the robo-advisors that typically use them.
Morningstar’s director of exchange-traded fund research penned an interesting article April 12 on the total cost of ETF ownership, which we also wrote about in a recent article examining the latest academic research on the active-passive debate and the competitive threat that growing ranks of robo-advisors supposedly represent to mortal financial advisors. The investment research firm’s Ben Johnson, breaks down the total costs of ETFs into two categories: holdings and transactions costs. The relative impact of each, of course, depends on the investor’s time horizon, with holdings costs, as fees, growing in importance the longer an ETF is held. But it’s Johnson’s review of ETF sampling that caught Market Insights’ eye.
“The manner in which funds seek to replicate their benchmarks can also be a substantial source of implicit holding costs,” Johnson points out. “For instance, a fund tracking a benchmark that contains a number of smaller, less liquid components like an emerging market or fixed income fund may use ‘sampling’ techniques to replicate the returns of its index.” So an ETF manufacturer might pick the larger-cap, more liquid index constituents “to improve the overall liquidity of the fund itself—it makes the creation and redemption process simpler and cheaper for market makers—and to minimize costs,” he notes, adding that some of the “obvious advantages” of skirting smaller-cap, more illiquid constituents creates potential divergence from index returns.
No kidding. Before 2010, the iShares MSCI Emerging Markets (EEM) ETF just held roughly half the index stocks for the sake of efficient trading. But in 2009, that composition differential translated into EEM’s huge tracking divergence from its benchmark: EEM’s total return that year lagged that of the index by ten percentage points, clearly driven principally by the composition differences, though a “relatively high expense ratio” didn’t help. Index replication differentials may have subsequently decreased, but even lower differentials, coupled with the ETF’s trading and holding costs, have still meaningfully dragged on EEM’s returns over time. For example, the total cumulative return of the MSCI EM Index from the end of the first quarter 2012 to the end of March 2017 amounted to 5.9%, more than four percentage points versus EEM’s 1.66% return in the period.
Other costs associated with ETFs include changes to the “market portfolio” or index constituents, including inclusion of new constituents and removal of those replaced; secondary share offers or company share repurchases; mergers and acquisitions; bankruptcies. The timing and treatment of taxes on constituent dividends also impact ETF returns, Johnson notes.
ETFs are a valuable tool for investors, as are the robo-advisors that typically use them in their portfolio allocation models. But they aren’t without their own risks and implicit costs, which impact returns. As for the robos, investment “automaticity” is clearly a positive development, but just as with active portfolio managers, performance among robo-advisors also varies. And the degree to which they can manage the behavioral elements of their clients—from market timing to performance chasing—may be quite limited when paired against a seasoned human financial advisor who presumably knows how to manage his client’s goals, greed and fear.