Exchange traded funds represent an opportunity for both retail and institutional investors to access movements in global asset markets at relatively low cost. Nonetheless, they have come in for major criticism from some market professionals who believe that they distort asset values away from market fundamentals and have become a tool for institutional investors to game the system to their own advantage. Whilst the criticisms are based on some truth, the benefits appear to outweigh the costs. In particular, any institutional investors who use instruments such as inverse ETFs to move the market to their advantage are likely to come off worst.
The ETF industry is heavily concentrated, with three companies accounting for 80% of US ETF assets under management (AUM). One of those companies is The Vanguard Group, founded by Jack Bogle who died earlier this year. Mr Bogle can lay claim to being one of the most influential people in the history of finance, since it was he who devised the index fund – a portfolio designed to match the components of a broad market index such as the S&P 500. ETFs were the next logical step in the product development chain, being differentiated from index funds in terms of their cost structure (ETFs are generally cheaper). Yet, Mr Bogle was less than enamoured of ETFs, calling people who dabbled in them 'fruitcakes, nut cases and the lunatic fringe'.
What concerned him so much was the fact that whereas index funds are marked to market only at the end of each trading day, ETFs can be traded throughout the trading day which subjects them to general intraday moves and potential liquidity squeezes. Moreover, he believed that the ETF industry had been taken over by speculators, and in his final book, published last year, he wrote, 'the arithmetic suggests that only about one-sixth of ETF assets are held by investors with a focus largely on the long-term'. Bogle’s concern was ETFs had become such a popular instrument for hedging the market positions of institutional investors that they had become increasingly ill-suited to the needs of the buy-and-hold investors that he had spent his career supporting.
Whilst there might be some truth in these criticisms, we need to disentangle fact from fiction. Market volatility is a fact of life but over the last decade key measures of market volatility such as the VIX have traded near to all-time lows. Moreover, the very fact that ETFs have to replicate the structure of the market means that equity prices do not fall because ETFs are selling. Rather, the value of the ETF falls because investors are selling the equities which constitute the index. It is also questionable whether long-term investors account for such a small share of ETF assets: industry professionals reckon that the majority of ETF investors are buy-and-hold investors. Nonetheless, ETFs appear to have had some unintended consequences for the overall operation of financial markets.
To understand the issues, consider how ETFs work. They are hybrid investment products with many of the investment features of mutual funds combined with the trading features of equities. What makes them unique is the creation/redemption feature that improves their ability to trade close to their true net asset value (NAV) throughout the trading session. ETFs disclose their holding at the start of each trading day. So-called authorised participants (APs) have the authority to buy the underlying securities in the ETF in relation to their overall index weight and deliver the basket to the ETF manager at the end of the day. Similarly, if an AP wants to sell securities as part of the rebalancing process, they present the shares for redemption to the ETF issuer and receive the basket of underlying securities which they can sell on the open market. Precisely because the AP knows the composition of the basket they will have to deliver at the end of the day, they can buy (sell) ETF shares and sell (buy) a basket of underlying securities in the knowledge that they will be able to swap their holdings at the end of the session. This process of constant creation and redemption is the critical factor which keeps the price of an ETF in a tight range around the NAV of the securities portfolio.
Chart 1 shows the tracking error of the S&P 500 SPDR Index compared with the underlying S&P 500 Index. In the period since 2010 the error has averaged 0.01 percentage points per day which annualises to 1.9% per year. This deviation is determined by a variety of factors, including ETF fees, the quality of the ETF manager and the bid-offer spread on the underlying assets, which in turn is determined by the volatility of market conditions. For a long-term buy-and-hold investor there is nothing to do but sit back and enjoy the ride. But those who dip in and out of the market on a regular basis have to be aware that sometimes the ETF will outperform the underlying and sometimes it will not (hence the volatility evident in Chart 1).
Daily change in S&P 500, less change in SPDR
It is impossible to reduce the tracking error to zero. This is particularly true in the case of equity indices, where reweighting occurs due to the fact that companies enter and exit the index which in turn requires the ETF to be reweighted. In such a case, the trading costs associated with such moves, and the lagged response of the ETF to index changes, opens up the prospect of daily deviations in NAV performance. In addition, during periods of high market volatility, the bid-ask spreads on underlying securities will widen, which raises the range of traded prices around the NAV. Nonetheless, the overall performance of ETFs tracks the underlying asset sufficiently closely, and offers significant advantages in terms of liquidity compared to standard index funds, that to all intents and purposes they offer investors a cheap and convenient way to trade the wider market.
Although the settlement process for ETFs is the same as for any other listed security, there are some quirks which might suggest they are used predominantly as an instrument to short the equity market. This would certainly be one explanation for Jack Bogle’s concerns regarding their use as a speculative instrument.
Data of the Securities and Exchange Commission (SEC) on shorted securities, for example, often show that ETFs have sold more shares short than apparently exist. Of course, this is not possible. One explanation for this phenomenon might be the so-called cascading shorts problem. Imagine an AP wants to sell shares to an institutional investor but does so by borrowing from another market maker. In accounting terms, the ETF is 100% short. This problem is further magnified if the buyer lends these shares to a third party, with the result that the ETF is now 200% short. But since the chain can be swiftly unwound without anyone suffering any losses and is fully collateralised, the ETF is not really short. To the extent that the accounts show the current gross position of the ETF and not the ultimate net position, the ETF’s short position is overstated.
However, one curious feature of the ETF market is that the put/call ratio on the SPDR Index has consistently exceeded unity (Chart 2). To the extent that a ratio above one is a bearish signal as it indicates the number of put options exceeds the number of call options, it is difficult to see why this should be the case unless investors are indeed using ETF options as a hedge against moves in the S&P 500.
It is also possible for investors to short the market outright using an inverse ETF. Such funds use instruments (mainly derivatives) to produce a daily performance that is the mirror image of an underlying index. It is important to note that inverse ETFs rebalance on a daily basis in order to maintain a constant leverage ratio. As a result, the holding period is only one day and each day’s performance is treated as a separate event. This is important because it has a major impact on investor returns.
To see why this is the case consider an example. Suppose an investor places USD 10,000 into a fund which inversely varies 1:1 with a particular market index. If the underlying market index declines by 2% on the first day, the investor return is 2%. If it subsequently declines by another 3%, the investors’ return compounds to 5.06% (1.02 x 1.03), meaning that they are now sitting on a fund worth USD 10,506. But if the index rebounds to its initial starting point on the third day, thus recouping all its losses, it jumps by 5.2% (1/((1–.02) x (1–.03)). The investor thus loses 5.2% of the fund they have accumulated so far, which takes its value to USD 9,960 – a loss of 0.4% compared to the original stake. In contrast, if the investor simply took an open short position on the index, their initial losses over the first two days would have been exactly offset by the gain on the third day.
In some cases the fund is leveraged, with one extreme example being the Direxion Daily Small Cap Bear 3X Shares (TZA) Index which is designed to produce returns three times the inverse of daily gains in the Russell 2000 Index. Thus, if the Russell 2000 records a daily fall of 1%, the TZA returns a 3% gain. Using the numerical example outlined above, after two days the investor is sitting on a gain of 15.5% but the decline on the third day leaves their fund 2.5% below its initial value (Chart 3).
As Chart 4 shows, during times when equity markets are falling, inverse ETFs can generate decent gains. An investor concerned about prospects for the S&P 500 who purchased a simple inverse ETF in the form of the ProShares Short S&P 500 at the start of 2007, would have generated a gain of 81% by March 2009. But by late-2010 the long equity position would have matched the performance of the inverse ETF and as the S&P 500 continued to rally, the inverse ETF returns would have dwindled sharply. These impacts are magnified by leveraged indices. We cannot replicate the calculations for the TZA since data only exist from November 2008 but the strength of the equity rally from March 2009 would very quickly have made it unattractive to hold long positions in the leveraged inverse ETF.
1/1/2007 = 100
Basic arithmetic mitigates against inverse ETFs. As the index goes lower, an increase of any given magnitude results in a larger percentage increase (a larger percentage loss for the investor). This is compounded by the fact that the index has to be rebalanced each day. In an environment where equities tend to rise it is impossible to generate long-term returns with an inverse ETF. They do, however, have their use in an environment where investors are concerned about short-term market dips and can be used as a short-term hedging instrument.
One of the concerns expressed by investors over the years is that they have not yet been tested during a major market rout which could exacerbate the magnitude of any sell-off. Although US ETF assets today account for a far higher proportion of the S&P 500 market cap compared to a decade ago, a 15% share is still relatively small, though it is big enough to be a potential problem (Chart 5). An academic study published in 20171 concluded that the rise of ETFs means that passive investors derive ‘lower benefits from information acquisition’ which reduces the efficiency with which investment decisions are taken and raises the risk that market swings may be larger than would otherwise happen in the event of a more efficient market.
It is likely that this effect is overblown. After all, equity-based ETFs respond to changes in the overall market rather than the other way round. However, one incident in August 2015, when the NAV of ETFs based on US equities deviated significantly from their underlying indices, highlighted that there are occasions when the market does not always operate as intended. Back in 2010, following the flash crash of 6 May, the SEC had put in place a series of circuit breakers designed to prevent indices from moving by more than a specified amount within a specified time frame. On 24 August 2015, when the US equity market was 5% limit down at the market opening, it took 27 minutes before market makers were able to price ETFs, resulting in significant pricing volatility (Chart 6).
One of the underlying causes was that the circuit breakers meant willing buyers and sellers were unable to come into the market to square their positions which resulted in liquidity drying up. Matters were exacerbated by a huge spike in trading activity which was likely the result of algorithmic trading unloading positions without due regard to pricing or fair value. Consequently, the problem may not so much be ETFs but rather the algo trading programs, which is a wider problem that has also caused disruption in other asset markets. With regard to the liquidity problems, the SEC responded by changing the methodology for determining reference prices and modified the reopening process after a stock is halted in a bid to smooth out calls for liquidity after trading has been halted. So far we have not had any repetition of the 2015 experience which gives us some confidence that the problem has largely been resolved (famous last words).
ETFs are still a new and evolving asset class. Undoubtedly, some of the concerns expressed by Jack Bogle are based on grains of truth. Whilst it does appear that investors use equity ETFs to hedge against widespread market corrections, there is no evidence that they are unduly distorting the market.
Instruments such as inverse ETFs could be used by speculators who wish to nudge the market in a particular direction in order to make gains, but they are inherently dangerous because investors have to time their entry and exit in order to make any gains, and leveraged inverse ETFs need to be used very sparingly. As for concerns that ETFs might exacerbate market corrections, this appears to be based on the notion that there is a two-way causality running between equity markets and ETFs whereas in reality it runs only from markets to ETFs. This might be an overly sanguine view, but we will find out whether or not ETFs have been swimming naked when the equity bull tide finally goes out.
1 D. Israeli, C. Lee and S. Sridhar an (2017), ‘Is There a Dark Side to Exchange Traded Funds? An Information Perspective’, Review of Accounting Studies (22), pp. 1048–1083.