Journal of Asset Management (2009) 10, 210–221. doi:10.1057/jam.2009.13

Price volatility and tracking ability of ETFs

Jack W Aber 1. Dan Li 2 and Luc Can 3

management.bu.edu

1 is Professor of Finance and Chair of the Finance Department at Boston University School of Management. Dr Aber’s research interests span financial instruments, institutions and markets.

2 was awarded the PhD in economics from Boston University College of Arts and Sciences in 2008. She is a specialist in economic history and applied econometrics.

3 is Senior Advisor to the Chairman and Senior Vice President at the Bank for Investment and Development of Vietnam. He was recently a senior fellow at Harvard Kennedy School and a Hubert H. Humphrey Fellow at Boston University under a Fulbright exchange program. Dr Can holds an MBA (Finance) and a DBA (Finance) from Monash University, Australia.

Received 15 March 2009; Revised 15 March 2009.

Abstract

In this paper, we examine the price volatility and tracking ability of four iShares™ exchange-traded funds (ETFs). We use three measures: the premium and discount position, daily return, and tracking error, compared with conventional index mutual funds tracking the same index. Our results indicate that the ETFs are more likely to trade at a premium than at a discount, with comparatively large daily price fluctuations; and that both fund types have approximately the same degree of comovement with their benchmarks, but differ slightly in their tracking ability. On average, the Vanguard™ conventional index funds beat their corresponding iShares™ competitors in terms of tracking error.

Keywords:

tracking ability, ETFs, index mutual funds

INTRODUCTION

Exchange-traded funds (ETFs) can be organized as open-end mutual funds or unit investment trusts. Most US ETFs are of the former type. An ETF is typically index-based, in that it consists of a portfolio of securities that is meant to track the investment performance of a specific index. To date, most US ETFs track equity indexes, such as the S&P500 or the Nasdaq-100. Because they are index-based, ETFs are considered potential substitutes for conventional open-end index mutual funds, and are widely viewed as competing for investors in the same markets as do index funds. A significant feature of ETFs is that, as stocks, they can be traded throughout the trading day, unlike mutual funds that can be traded only at the end of the day at their Net Asset Values (NAVs). It is also clear that ETFs’ expense ratios can be, and usually are, lower than those of conventional index mutual funds, which in turn are lower than the expense ratios of actively managed mutual funds. A major reason why ETF expense ratios are lower is that because they are traded on an exchange, they do not need to provide shareholder services, transfer agency services, and so on. Expense ratios of funds analyzed in this study are provided in Table 3. ETFs are also considered to be tax-efficient because the creation and redemption process described below does not create taxable capital gains. In addition, ETF investors do not incur sales loads and 12b-1 distribution fees such as those applied to many actively managed mutual funds ( Doran et al. 2006 ).

Among the many interesting questions about ETFs since they appeared in the United States in 1993 are those that focus on their functional similarity to index funds. Put simply, in what ways are ETFs similar to, and different from, index funds? Additionally, in the same way that questions arise concerning the performance of index mutual funds compared to the indexes they track, the performance of ETFs compared to that of their indexes is a matter of interest. In this paper, we investigate these areas of comparability, with a specific focus on price volatility and tracking error.

The history of the ETF is a story of impressive growth. The original US ETF, the SPDR (‘Spider’) was first offered in 1993, and for several years – roughly through the 1990s – it was not clear that ETFs would capture the interest of large numbers of investors. At the end of 1999, there were 30 US ETFs with US $ 33 billion in assets (see Table 1 ). But this decade tells a different story. Although SPDR remains the leading ETF on the US market, at about $ 55 billion in assets, its brethren now number over 600 funds, and growth through 2007 has lifted asset totals to greater than $ 600 billion. Clearly, investors have embraced ETFs.

Do ETFs pose a competitive threat to conventional mutual funds? The size of the US mutual fund market, at the end of 2007, was approximately $ 12 trillion. The US ETFs thus now have about a 5 per cent market share. But it is growth relative to conventional mutual funds that ETF sponsors are optimistic about, where continuation of their strong trajectory is predicted. Direct comparison of asset growth in the first 15 years, respectively, of ETFs and conventional mutual funds suggests that sponsor optimism may well be justified (see Tables 1 and 2 ). Time, of course, will tell.

ETFs are open-end index mutual funds and as such are passively managed by their respective sponsors. Shares in a given ETF can be created or redeemed at will. Only qualified institutional investors as a practical matter create or redeem shares in blocks of 50   000 or more shares. The existence of these institutional investors makes arbitrage of any price discrepancy possible. To illustrate, if shares of an ETF trade at a premium in excess of the value of the index’s underlying shares, the institution can sell a block of ETF shares and buy the underlying shares. Activity by institutional investors should in theory lead to economically insignificant price discrepancies ( Dellva, 2001 ). However, whether this theory fits the facts is debatable. For example, Jares and Lavin (2004) provide evidence of arbitrage opportunities in Hong Kong and Japanese markets, due in part to non-overlapping trading hours. This issue will be explored in this paper.

While abundant literature exists on conventional mutual funds, there is limited academic research on ETFs, including those which compare conventional mutual funds with ETFs, mostly due to the short history of ETFs and lack of availability of related data ( Agapova, 2006 ; Guedj and Huang, 2008 ).

Previous ETF studies mainly describe specific features of this innovative investment instrument (see, for example, Gastineau, 2002 ; Mussavian and Hirsch, 2002 ); or examine their return performance and trading characteristics (see Rompotis, 2006 ). Rompotis (2007a). in his examination of seasonal effects on return and volatility of various types of ETFs, points out a significant November effect.

A handful of other studies compare ETFs and conventional mutual funds in terms of their costs, performance and risk. Dellva (2001) compares annual expenses of the Vanguard Index 500, iShare S & P 500 Fund, and SPDR; and finds that ETFs have a significant cost advantage. Poterba and Shoven (2002). in their comparison of the pre-tax and post-tax returns on the Vanguard Index 500 and SPDR, find that their returns are quite similar. Elton et al (2002). on the other hand, document that the SPDR underperforms both its underlying index and conventional S&P 500 index funds. In addition, they document its minimal premium or discount relative to its NAV. Gastineau (2004) addresses a similar issue, and claims that ETFs underperform their index fund competitors. Specifically, Gastineau suggests that, at least in part, the deficiency in an ETF’s tracking ability is due to the ETF managers’ reluctance to adjust the portfolio before the official moment of the index adjustment.

Rompotis (2005) examines the performance of 16 ETFs and index funds in pairs tracking the same index, and reports that they substantially produce similar returns and tracking errors. Agapova (2006) investigates the substitutability of ETFs and index mutual funds, and finds that their coexistence can be partially explained by a clientele effect, and also that ETFs have smaller tracking errors and lower fund expenses. In a similar vein, Guedj and Huang (2008) compare these two types of funds, and argue that, though being similar investment vehicles, ETFs offer new options to track more volatile and less liquid indexes that are not cost effective in the open-end fund structure.

Overall, in tandem with the rapid growth of ETFs, there has been a surge in the literature on ETFs. Empirical evidence on comparative performance and risk of ETFs and conventional index mutual funds, however, is inclusive. ETFs have certain cost advantages, but do not outperform (and in some cases, underperform) their index mutual fund competitors. To our knowledge, no studies have examined the price volatility and tracking ability of ETFs in both the short- and long term. In this paper we contribute to the limited literature on ETFs by applying statistical approaches to investigate the price volatility and tracking ability of ETFs, compared with conventional index mutual funds. Our objective is to demonstrate how different comparisons may be made, rather than to undertake a comprehensive empirical study.

The paper is organized as follows. Data are described in the next section. The subsequent section contains four subsections, in each of which a methodology is introduced, followed by empirical results. The final section presents our conclusions.

DATA

We have selected for analysis four actively traded iShares™ ETFs, of which three are domestic funds and one is an international fund (see Table 3 ). The domestic ETFs (IVV, IWF and IWM) track the S&P500 index, the Russell 1000 growth index and the MSCI US Small Cap index, respectively. The international fund (EFA) tracks the MSCI EAFE index. This index includes a selection of stocks from 21 developed markets, but excludes those from the United States and Canada. It is the most commonly used benchmark for non-US stock funds. For ETFs based on non-US stocks, the underlying portfolio stocks and the ETFs in US exchanges are not synchronously traded, which can be a source of observable tracking differences from their domestic counterparts.

Additionally, for comparative analysis we have selected four Vanguard conventional open-end mutual funds, which track the same indexes as their corresponding iShares competitors. By comparing the tracking performance of the ETFs to that of the conventional mutual funds, we expect to observe the relative efficiency / inefficiency of the ETFs in terms of how well they track their underlying indexes. Table 3 summarizes the data set. The sample size in other studies has varied from one fund to 20. 1 Approximately 11 pairs of iShares ETFs and Vanguard index funds track the same benchmark, of which we have chosen the four identified above. 2

The ETF data include closing prices, the daily high and low prices, and NAVs beginning with each fund’s inception date and ending on 14 December 2006. NAVs are calculated once at 1600 hours Eastern time on each trading day. The NAV data and inception dates for the iShares™ ETFs were obtained from the iShares™ official website, www.ishares.com. All price data were obtained from www.yahoo.finance. 3

Most of an ETF’s trades on a given day occur neither at the closing price, nor at the daily high or low prices. Instead, they are more likely to occur at a price around the midpoint price of the day. For example, Figure 1 shows the trading volume (the horizontal bar) by price (the line) for IWM on 4 May 2007, during which the closing price, daily high and low prices were $ 80.44, $ 80.81 and $ 80.35, respectively, and the largest trading volume occurred around the price of $ 80.65. We would argue that the price at which the largest volumes are traded may matter more to investors than the closing price because it more closely reflects the likely direction of the price movement, such that, for example, large trading volumes at low prices are likely to indicate that the price will decline. Moreover, the question arises whether the price (at which the largest trading volume takes place) deviates substantially from the underlying index. If it does, we cannot confidently call the ETF under investigation a good index tracker simply based on its closing price.