Are Flows Costly to ETF Investors? (Digest Summary)

By design, exchange-traded funds (ETFs) are liquid and transparent, enabling precise monitoring of investment performance and benchmark tracking. The authors find evidence of investor behavior that suggests a pursuit of yield as implied by return patterns on the indexes tracked.

A financial market innovation that is only 21 years old, ETFs have facilitated access to a myriad of market exposures. Unique to the instrument is its provision of liquidity in both the primary and the secondary markets, which allows for more targeted asset class exposure and accurate tracking of investment returns. ETFs trade like individual securities, permitting intraday sale as well as purchase. Liquidity is supported by the use of authorized participants, which are quantitative trading intermediaries that both create and redeem shares to create a balance between supply and demand. It is as though the investment vehicle itself has a built-in market-making and market-clearing device.

Innovation in response to demand has moved the ETF beyond traditional exposures into more opaque and complex asset classes. This shift has necessitated a dual approach to share creation and redemption at the primary market level. In-kind exchanges have been and continue to be the norm for most ETFs. To create shares, the authorized participant purchases a basket of underlying securities—those that compose the ETF and their quantity—to address excess demand. The process is reversed to enable share redemption. In contrast, cash exchanges require that the fund itself rather than the authorized participant transact in the underlying securities. This latter approach to share creation and redemption often takes place in funds that offer more unique and perhaps modified exposures, such as international emerging markets.

Using a simple market-timing strategy and a statistical model, the authors examine the extent to which the share creation and redemption mechanism of the ETF primary market affects its performance and ability to track a benchmark. The strategy considers a fund’s ability to predict returns and time the market. The model focuses on the possible impact on performance of ETF shares that use cash creations and redemptions because these transactions incur greater costs. The authors uncover return-chasing behavior; they find that share redemptions track negative index returns and share creations track positive returns but do not anticipate them. The quantitative model indicates that cash creation as a result of time lags may affect performance, particularly for international exposures.

This study is useful to traders and portfolio managers who want to create exposures and better understand the volatility underlying those exposures.

The sample includes all publicly traded ETFs that are listed on US exchanges identified in Morningstar Direct as of 1 July 2012. The sample period is from 2006 to the end of June 2012. To control for infrequently traded funds and to reduce noise in the data, the authors exclude ETFs with less than $25 million and average daily turnover of less than one-tenth of 1% of shares outstanding. Additionally, the funds must have had at least six months of daily data available. Funds are grouped by the asset class that makes up the benchmark index and their geography (domestic or foreign) and primary market transaction type (cash or in-kind exchange). Any ETFs tracking commodity, leveraged, or currency indexes are excluded. Data collected for these funds include historical daily closing share prices, number of outstanding shares, distribution dates, and amounts.

Cash exchanges are more common among international and fixed-income funds. Fund creation and redemption is indirectly observable by proxy in the monitoring of daily changes in the number of shares outstanding. Analysis is able to measure the flows, net creation, and redemption that take place each day.

The authors propose a market-timing strategy to test ETF susceptibility to this process by looking for patterns in benchmark index returns around fund flows. The strategy entails the purchase of ETF shares at the close of business if S&P 500 Index returns are positive; otherwise, the strategy invests in the risk-free rate. They find that ETF investors pursue returns as flows respond to them. ETF creations and redemptions follow returns; they neither precede nor coincide with them.

Additionally, the authors describe and implement a quantitative model to investigate a possible link between ETF performance and costly flows that they attempt to replicate with proxies. The authors observe somewhat higher spreads for ETFs that use cash creation and redemption, which was more prevalent during 2007–2009.

Innovation in financial markets has gone wrong at times despite its positive intentions. So far, that does not appear to be the case with the ETF. A fairly new creation, it has allowed for greater precision in portfolio management design and implementation processes. As the product evolves to address the challenges and dynamics of markets, a more precise understanding of not only what it captures from those markets but also what effects it may exert on them is essential to its continued success. What the ETF does or fails to do may influence how investors will use it.