The risks of excess in ETFs

Whether or not its inventions are socially useless, it is hard to disagree that the investment banking industry has an extraordinary facility for innovating its way to disaster. The publication last week of a consultation paper on ETFs by the European Securities and Markets Authority (ESMA) is a further reminder of mounting regulatory and central bank concern that ETFs are becoming a case in point. They were invented as an even-lower-cost, open-ended alternative to the indexed mutual fund, with the additional benefit of stock exchange levels of continuous liquidity, no subscription or redemption charges, and (especially in the United States) greater tax-efficiency. Now regulators and central bankers are worried that, in the hands of over-inventive financial engineers, ETFs are in danger of degenerating into a yet another leveraged synthetic-cum-derivative financial instrument with no utility beyond the ability to be devised, sold, valued and traded by investment bankers and their clients.

There is strong prima facie evidence for concern. It is precisely because they are free of the regulatory constraints imposed on mutual funds – notably, limits on leverage and insistence on asset and counterparty diversification – that the financial engineers of the investment banking industry were drawn to ETFs. At the innovative end of the market, ETFs do indeed appear to be developing many of the pathologies associated with any credit-fuelled financial market spiraling out of control. They can be bought on margin, short-sold, leveraged, inverse-leveraged, and capital-guaranteed. They can lend stock, use swaps, and concentrate risk on a single asset or counterparty. They can be used as a substitute for real assets, or to help banks fund poor quality inventory. They can provide routes around capital and liquidity requirements, alter the risks facing investors and even - in perhaps the clearest measure of how far they have strayed from their roots as passive investment products - be managed actively.

True, the innovative ETFs that are the main source of concern account for a fraction of a fraction of the global investment management industry. At the end of last year, the entire global ETF market was worth a mere $1.4 trillion in assets under management – and no more than $1 .6 trillion even if all exchange-traded products, including debt-based notes and commodity funds are added to the total. That is one sixteenth (6.1 per cent) the size of the $23.1 trillion global mutual fund market, and a mere thirty-third (2.95 per cent) of global equity market capitalization of $47.4 trillion at the end of last year. The vast majority of ETFs remain passive, standardized tracking products of the kind the industry was invented to provide. According to a recent report from Deutsche Bank, purely passive products make up 96 per cent of ETFs traded in Europe and the United States. In other words, the leveraged synthetic ETFs that have become the focus of so much regulatory and media attention in recent years even now make up less than a twentieth of the industry. Even in Europe, where most of the synthetic ETFS are being packaged, industry leader BlackRock (owner of i-Shares) says leveraged ETFs represent a mere 3.1 per cent of assets under management.

In other words, the vast majority of ETFs on both side of the Atlantic still use mainly physical replication of the kind the industry has employed since the first ETF was invented in 1989. This means the market makers purchase the physical securities that make up the index the ETF is intended to track (such as the S&P 500) and deliver them to the ETF sponsor (a fund manager) in exchange for listed units in the ETF. Though they are bought and sold throughout the trading day, the prices at which units change hands remain closely linked to the net asset value (NAV) of the basket of securities deposited with the issuer and can only be redeemed at the end-of-day NAV. Even the redemption of units remains physical: market makers approach the fund manager which issued the units and exchange them for an appropriate proportion of the basket of underlying securities. In short, most ETFs are conservatively managed instruments that remain firmly linked to the underlying assets.

The ETF market also remains overwhelmingly institutional, especially in Europe, though half the investors even in the more mature American market are non-retail. For now, distributors to retail investors still prefer the fatter commissions they get from selling mutual funds. At most, one fifth of assets under management in ETFs globally are owned by retail investors, and the real proportion could be as low as a tenth. It follows that the bulk of ETF investors ought to be sophisticated enough to understand the risks that they are running. In addition, although proprietary trading desks and hedge funds are large users of ETFs for hedging, transition management and short-selling purposes - especially in the United States, and not least because the securities are relatively easy to borrow – ETFs are not great drivers of transactional activity either. Last year, ETFs accounted for just 8.5 per cent of equity turnover on European exchanges. Even in August last year, when American market commentators blamed ETFs for the extraordinary volatility in the equity markets that month, they accounted for just over a third of daily value traded. Within that total, leveraged ETFs accounted for an eighth, or just 4 per cent of total value traded.

So, if ETFs are a small market in terms of value and turnover, and consist almost entirely of conservatively managed funds sold to institutional investors, why are regulators so exercised about ETFs? Exercised they most certainly are. In February last year the Financial Services Authority (FSA) expressed concern that retail investors in the United Kingdom were being exposed to specific collateral and counter-party risks through swap-based ETFs, and to unquantifiable and often inadequately regulated offshore risks through the proliferation of Exchange Traded Notes (ETNs) that focus on asset-backed bonds and Exchange Traded Commodities (ETCs) that attempt to replicate investment in a commodity, such as gold, mainly through the use of futures or forward contracts. The European Commission has also placed ETFs at the forefront of its concerns about investor protection. Because almost all European ETFs have adopted a UCITS structure, they are subject to wider concerns about lack of disclosure of risks and conflicts of interest within the organizations designing and promoting all forms of UCITS funds, and especially those that make use of complex instruments or structures. The paper published last week by ESMA is actually its second consultation on UCITS-wrapped ETFs in the last six months.

National and international central bankers are also voicing concern. In April 2011, the Financial Stability Board (FSB) published a paper on ETFs that expressed disquiet at the implications for financial stability of the diversification of ETFs from equities into illiquid rates, credit, commodity and emerging market asset classes. It argued that this was liable to increase price volatility and make it harder for investors to redeem, and expressed concern about the growing use of swaps and securities lending by ETFs to cut costs, enhance returns and bypass regulations. That same month, the Bank for International Settlements (BIS) published a working paper pointing to a variety of ways in which ETFs were creating systemic risk. It itemized incomprehensible and unmanageable complexity, embedded but hidden leverage, and illiquidity risks in both the financing and collateral markets. Last but not least, the systemic dangers posed by ETFs have in the last year become a staple feature of the Global Financial Stability Reports issued by the International Monetary Fund (IMF), and of both the Bank of England Financial Stability Report and the Financial Stability Review published by the European Central Bank.

ETF insiders profess to be puzzled why so much regulatory and central bank energy is concentrated on a sub-section of a small industry which is already heavily regulated, not only under the UCITS regime in Europe but under the 1940 Investment Company Act in the United States as well. Yet it is not hard to understand why regulators are concerned. There is no ETF-specific set of regulations. A fast-growing industry, which investment bankers have identified as amenable to unconstrained financial engineering, is stirring uncomfortable memories of recent systemic crises and demonstrable failings in investor protection. As the history in the structured credit markets between 2003 and 2007 proves, investment bankers tend to focus on the immediate fees and spreads, while taxpayers collect the bill when liquidity dries up and leverage starts to have its reverse effects. The FSA and ESMA can scarcely be criticized for worrying it might happen again. Indeed, investment bankers grumbling that regulators are misinformed and misunderstand the nature of innovation in ETFs would do well to remember that they said exactly the same thing about structured credit as that market shifted from the physical realm to the synthetic.

At one level, the FSA and ESMA are seeking nothing unusual. They want to eliminate regulatory arbitrage (in which tinkering with ETFs allows unsuitable products to be sold to investors), impose full disclosure on promoters about swaps and stock loan and tracking error (so investors know exactly what is being sold to them) and prevent widows and orphans being sold unsuitable investments. The regulators want ETF managers to tell investors how they track an index (whether it is done by full physical replication, sampling or a total return swap), how the details of the methodology (such as dividend reinvestment, tax enhancement and securities lending) influence the scale of tracking error in either direction, and what risks are being incurred on their behalf (in the taking of collateral against swap or securities lending counterparties, for example).This is what regulators charged with protecting investors are meant to do, whatever anybody thinks of the usefulness of such work, and in this sense they are treating ETFs no differently from any other investment product. But at another level the risks regulators are highlighting for investors in ETFs are part of an equally familiar narrative: the continuing propensity of an unrepentant investment banking industry to manufacture and distribute what Warren Buffett famously called “weapons of mass financial destruction.”

Since total return swaps were chief among the “weapons of mass financial destruction” identified by Buffett back in 2002, it is worth noting that they are the instrument used in synthetic ETFs. Thanks to regulation, the swap-based ETF is at present a largely European phenomenon. European ETFs are almost invariably structured as UCITS funds, and UCITS funds have since the passage of the UCITS III directive been able to make use of derivatives. In the United States, any ETF wanting to use derivatives has to seek exemption from the provisions of the 1940 Investment Company Act, and in March 2011 the Securities and Exchange Commission (SEC) made plain its view of the use of swaps in funds sold to retail investors by suspending its review of applications for exemption. But in Europe, synthetic ETFs now make up nearly half the market. They use total return swaps to provide investors with a pay-off at the end of a pre-defined period, based on the return from an index or asset or portfolio of assets. The advantage is that they offer exposure to markets and asset classes where physical replication is expensive, difficult or impossible, such as emerging markets. That is also, of course, their danger, because they enable ETFs to expose investors to illiquid asset classes.

But potential illiquidity is not the only danger they represent. The swaps also alter the nature of the risks. They can take a funded form (in which the ETF passes to the swap counterparty the cash subscription monies in return for securities as collateral to secure the cash and the return on the chosen index) or an unfunded form (in which the ETF gives the cash subscription monies to the swap counterparty in exchange for a basket of securities, and pays the return on those securities to the swap counterparty in return for the return on the chosen index). While either approach is always cheaper than physical replication, the savings are bought at the cost of exposure to risks other than the underlying assets. The first is the swap counterparty (where the exposure is reset every day) and the second is the collateral (at least in the case of a funded swap) which may not bear any relationship at all to the return sought by the ETF, or be readily realizable if the counterparty defaults. Interestingly, and in contradistinction to the view taken by the SEC, the ESMA paper published last week concludes that these risks are not sufficient to make synthetic ETFs “complex” products of the kind that can be sold to investors on the basis of advice from experts only.

Yet swaps are also one of the tools used by ETF managers – alongside short selling and futures - to create the increasingly notorious leveraged and inverse leveraged ETFs. These instruments, which aim to deliver a multiple or inverse multiple of the return on a chosen index or asset, have become the main focus of regulatory anxiety that the ETF model is being taken beyond its original, standardized, beta-driven personality. Clearly, leverage in ETFs carries the same risk as leverage does in any asset class. In other words, there is a strong possibility that a large gap will develop between the value of what is owed to investors (the fund) and the value of what is owned (the underlying index or assets). Losses incurred by investors in leveraged and inverse leveraged ETFs have already prompted regulatory interest in the United States as far back as 2009, chiefly on grounds that ETF distributors had not properly explained to investors the risks and rewards. In particular, investors did not grasp that leveraged ETFs deliver double the daily return, not double the annual return. The fact that leveraged ETFs cannot possibly deliver multiples of annual returns, but only daily leveraged returns, is still not well or widely understood.

Although leveraged ETFs are not yet a large part of the ETF market - $40 billion, or 3 per cent of outstandings, according to the Bank for International Settlements (BIS) in its paper of April 2011 - they are actively traded. The Bank of England has estimated that leveraged ETFs account for a fifth of total turnover in ETF assets. This high level of activity reflects the fact that a leveraged ETF, unlike its orthodox counterparts, is not a passive buy-and-hold investment, but trades more like an option. A leveraged ETF has to be rebalanced every day, in the sense of buying index assets in which it is underweight assets and selling index assets in which it is overweight, so inevitably trades the underlying assets more often. The more leveraged the ETF, the greater the required rebalancing. Over time, the compounding of daily returns guarantees the development of a divergence between the daily re-balancing and the leveraged performance of the target index, with divergence becoming especially large when prices are volatile. In other words, leveraged ETFs have travelled a long way from the beta returns with low transaction costs which ETFs were invented to supply.

Yet it is increasingly plain that even the most orthodox of ETFs incur risks which are not well understood, thanks to their appetite for lending the underlying securities.  ETF managers like to lend stock, for exactly the same reason other indexed funds like to lend stock: they have to hold the assets, and lending them can enhance returns, and at worst offset running costs. Securities lending is also an important source of revenue in a business where management fees are only 40-50 basis points on standard products, against the 150 basis points they can earn from explicit management charges alone on a mutual fund. Lending the assets of the fund can increase that 40-50 basis points by at least a half, so the manager (and any agent lender they use) is strongly incentivized to maximize the lending of the portfolio, irrespective of the risks this poses to investors. In fact, securities lending by physical replication ETFs is a far greater source of counterparty and collateral risk than the swaps used in synthetic or leveraged ETFs because - unlike swaps in a UCITS fund, which are constrained to single counterparty exposures of between 5 and 10 per cent of the NAV of the fund, according to the nature of the counterparty - the exposures are not subject to any regulatory limits.

The only limit on exposure is the combination of counter-party concentration limit set by the manager, and operated by the agent lender, and the concentration, quality and liquidity of the cash or securities accepted as collateral for the loan of the securities. In other words, investors are exposed to the skills and experience of the collateral manager and his agent, who set and operate the collateral eligibility criteria, haircuts and concentration limits. To the extent that the collateral manager is part of the same group as the fund manager, or the agent lender, this job is likely to be ill-done, because there is a conflict of interest. In a universal banking group, for example, the asset management arm may instruct the custodian banking arm to lend securities in one of its funds to the investment banking arm of the same group, which then on-lends them to a hedge fund, which may also be part of the same group. Conflicts of interest of this kind are far from improbable in securities lending by ETFs, since many of the leading issuers are part of universal banking groups that span asset management, investment banking and custody services.

But there are more immediate risks in securities lending than exploitation by associated entities. The principal risk is that the borrower of the stock defaults, and fails to return the stock. There is a further counterparty exposure when cash is reinvested in money market instruments issued by third parties, and collateral risk if it proves hard to realize in an event of default. Investors in ETFs run the further risk that they end up holding collateral which has nothing to do with the index or asset class in which they were originally invested. This is far from absurd – after all, the collateral is being held to mitigate a risk, not meet an investment objective – but the higher the degree of inconsistency between the collateral received and the constituents of the index, the greater the risk of irrecoverable loss. Securities lent might also be re-hypothecated by the borrower to third parties, and prove irretrievable. But the ETF manager is indifferent to all of these risks, because they are ultimately borne by the investors in the ETF.

This is the most remarkable aspect of the entire process. Despite the fact investors in the ETF bear all of the risks of lending securities, they do not enjoy all of the revenue. In fact, it is because they too can profit from lending the portfolio that almost every physical replication ETF manager lends stock. And managers retain an astonishingly high degree of discretion over how the returns from securities lending are allocated between the investors in the fund, the managers of the fund (they typically get 30-40 per cent), and any agent lenders (at least 20 per cent) used. It follows that the owners of the assets (the investors) take all of the risk of securities lending, but typically end up with only a 40-50 per cent share in the return. As Deutsche Bank pointed out in its study of the ETF market last summer, while taking a fee for administering a securities lending programme is not unreasonable, splitting half of the revenues with the agent lender is “a bit excessive.”

As listed securities, ETF shares can of course also be lent themselves. Being linked to particular indices and asset classes, shorting ETFs has become an attractive hedging tool. An ETF can be used, for example, to short an index (such as the S&P 500) or a sector (such as retailing) in a single trade, which is much more efficient than taking out dozens of short positions. This is one reason why ETF shares crop up in securities borrowing data published by Data Explorers, and why trading of ETFs is often high in relation to the underlying shares. This has fuelled the suspicion that many ETFs are being manufactured not for end-investors but for secondary market trading purposes only, in much the same way that convertible bonds and especially CDOs were issued for the convenience of speculative traders at the height of the 2003-07 credit boom.

But nothing stirs more uncomfortable memories of the structured credit disaster than Exchange Traded Notes (ETNs). These are tradable debt instruments which promise at maturity a pay-off based on the performance of an underlying index or asset class (such as commodities, real estate or currencies) or even an individual asset (such as gold). Since the pay-off is guaranteed by the issuer, and not by ownership of the underlying assets, ETN returns are totally dependent on the creditworthiness of the organization which issues the shares. Being unregulated - they are not eligible as UCITS funds, for example - they can be issued by SPVs (remember them?) and are not required to hold investment monies at a third party custodian bank either. Which is why ETF insiders insist that conflating ETNs with ETFs – as the Financial Stability Board does, describing them as “close substitutes” - is unfair, since they are not funds but notes, and not equity but debt obligations of a single issuer that do not entitle investors to ownership of underlying assets or offer them the additional security of collateralization.

Exchange Traded Commodities (ETCs) have run into similar complaints. They also tend to be uncollateralized debt instruments, issued by SPVs, whose redemption value is linked to a basket of commodities or a single commodity such as gold. ETF apologists point out that UCITS funds, the vehicle chosen by almost every European ETF, are not allowed to trade commodity futures, because they require physical delivery. But this has not stopped investment banks devising ingenious UCITS-compliant structures in which they trade commodity derivatives on behalf of managers to replicate commodity investment strategies. In fact, ETFs are now at the heart of the alleged `financialization’ of commodity markets, in which yield-hunting financial intermediaries with no interest in the underlying physical commodities are said to have alighted on grain, oil and gold as potential sources of non-correlated returns. Gold ETFs, for example, have brought to the gold market a range of new investors, and their buying and selling activity is now regarded as a major influence over the price of gold. In fact, central bankers now argue that ETFs are a sufficiently important influence over the price of gold to pose a systemic risk if they withdrew from the market precipitously.

The ETF industry says talk of systemic risk is wildly overblown, given that the entire global ETF industry has assets under management of just $1.4 trillion, and that no ETF has ever failed. However, it is to forget how fragile the sources of liquidity in the ETF markets actually are. The liquidity of any ETF depends on the willingness of market makers to quote two way prices to investors throughout the trading day. This means that they take the risk of closing out their positions at a loss when selling units back to the fund or buying additional units from the fund at the net asset value (NAV) at the end of the day. As ETFs have diversified into less liquid asset classes (notably commodities, but also emerging market equities) the risk that market makers will incur large losses has naturally increased. Where an asset is liquid, high frequency trading of that asset tends to keep the price of the ETF and the price of its components for NAV purposes aligned throughout the trading day. Where high frequency trading activity of this kind is absent, as in the less liquid asset classes, the risk that market makers will cut their losses by refusing to deal with investors is much higher. After all, market makers in ETFs are not obliged to provide continuous liquidity, and market makers as a group have a long and dishonourable record of withdrawing liquidity in volatile markets.

This matters in a market that guarantees liquidity on demand. Unlike hedge funds, which can always “gate” investors, shares in ETFs are redeemable daily. If there was an unexpectedly large rash of redemptions, ETFs would have to source assets. If they encountered mismanaged  sampling, or imperfect hedging of swap, or had stock out on loan, or had assets that were re-hypothecated, or held collateral far removed from the target index, or found the collateral was proving illiquid, or a bankruptcy administrator had made it impossible to retrieve assets held by or pledged to third parties, investors might be unable to redeem. ETFs may not be alone in incurring these risks, but they are incurring all of them. Any one of them might spark a run on an ETF, creating an atmosphere of widespread concern about counterparty risk that reverberated throughout the financial system as stocks were recalled, margin calls made, collateral sold, swaps unwound and positions closed. The parallels with the pressures on money market funds in 2007-08, when institutional investors redeemed units en masse in response to counterparty credit and collateral risk concerns, are certainly suggestive.

It will be objected that all of these systemic risks are purely theoretical, even speculative, when the real counterparty and collateral risks being run by ETFs are so small. One recent study estimated that synthetic ETFs are equivalent in value to a mere 2.8 per cent of the total outstanding value of OTC and exchange-traded equity derivatives, and that ETF assets on loan amount to only 2.3 per cent of total assets on loan. But size is a poor guide to the systemic importance of a financial market, particularly if it is fast-growing. The CDS market, for example, grew by 809 per cent in just three years between the end of 2004 and the end of 2007. Sales of CDOs doubled every year in the same period. A market which began by re-packaging physical securities quickly developed leveraged synthetic products backed by swaps, in which collateralization was trumpeted as the fail-safe mechanism. In the end, the underlying assets proved to be of poor quality, and even over-collateralization proved inadequate to protect investors from losses. The sale of assets at fire-sale prices was not just an unfortunate episode, but the precipitating cause of the prolonged financial crisis which began in 2007.

True, ETFs are not growing as fast as the structured credit markets were between 2003 and 2007. But assets under management by the global ETF industry have grown at an annual compound rate of 24 per cent a year since 2003, and the number of individual ETFs has grown even faster, at 29 per cent compound. These may even be under-estimates of the rate of growth. The Financial Stability Board estimated last year that the ETF market had grown at 40 per cent a year since 2001. What is undoubtedly true is that the only major obstacle to even faster growth is the fatness of the fees fund managers are still enjoying on traditional products. So it is important to note how profitable issuers of ETFs are finding the market, despite its reputation for being the low cost alternative. Deutsche Bank estimated last summer that the European ETF industry alone was generating revenues of €2 billion, of which €1.2 billion was being translated straight to the bottom line. Businesses as profitable as that are bound to attract imitators, which need innovation (and leverage) in order to differentiate themselves

The ETF industry which the new entrants confront is concentrated, though relatively open. There are 35 ETF providers in Europe, for example, but the six largest control 80 per cent of the market: iShares, State Street Global Advisors, Vanguard, Lyxor, db x-trackers and Power Shares. The failure of a fund run by one of these dominant providers might well create unfortunate spillover effects as investors lost confidence not only in one firm, but in the industry as a whole. The growing interconnection between ETF managers and investment banks is another consideration that is starting to worry the regulators. Two of the leading providers (db x-trackers and Lyxor) are controlled by universal banks with investment banking arms, and two of the largest providers outside the top six (Credit Suisse and UBS) are also linked to large investment banks. Four other investment banks – Bank of America Merrill Lynch, Goldman Sachs, Morgan Stanley and Nomura – have clubbed together to create Source Exchange Traded Products.

In other words, ETFs – like the fund management industry in general - is no longer a pure agency business in which disinterested fund managers look after the assets of investors, and use only third party service providers. Regulators are mindful that the investment banks which created the structured credit industry also created the investors in the structured credit industry, in the shape of the CDOs that became the main buyers of the lower quality tranches of mortgage-backed securities. In the ETF industry there are intimate connections between fund managers, investment banks and custodian banks, even within the same institution. Assets held by an ETF could be lent by an in-house asset manager through an in-house lending agent to an in-house investment bank, which can on-lend them to hedge fund clients at an attractive margin. Even if a synthetic ETF does not originate within the asset management arm of a major banking group, the opportunity for the investment bank to act as the swap counterparty has the potential to distort the decision-making process.

Potential conflicts of interest do not stop there. In fact, regulators are drawing to the attention of investors in ETFs a potentially sinister reason why investment banks like synthetic ETFs. This is that they can in theory provide a convenient way to fund their inventory. Investment banks, like every kind of bank, are highly leveraged. Only a fraction of what they own is funded through the issue of equity or long term debt. The majority of their assets are funded in the repo market, usually as short as overnight, because they aim to profit from a normal yield curve. The repo market can be expensive, especially for assets other than general collateral, where the haircuts demanded by cash providers are deep. Using a swap as an opportunity to transfer to an ETF collateral which is hard or expensive to finance elsewhere is an obvious temptation to an investment bank. It will be even more tempting if the ETF is issued by the asset management arm of the same investment bank, since lower quality assets can be used to collateralize the cash posted by the swap counterparty. As a means of funding lower quality assets, synthetic ETFs can also help investment banks circumvent the heavier capital and liquidity requirements now being imposed by regulators

If this is happening – and papers issued by regulators and central bankers have certainly alluded to the possibility – it marks the point at which a vehicle designed for the convenience of investors has instead become an instrument for financing the balance sheets of issuers. Whether or not the questions being asked of the ETF industry are fair, the recent development of the industry has brought to the surface conflicts of interest that have existed in the fund management industry since it was invented, and raised issues that ask extremely awkward questions of the structure of universal banking groups. For this benefit, the greater transparency of the ETF industry – at least by comparison with its mutual fund cousin – can be thanked. But there is also an irony at work. This is that the growing temptation to leverage ETFs, to adopt synthetic structures, and to link ETFs to exotic classes of assets, reflects the low rates of interest that have prevailed since the financial crisis started in 2007. By ensuring that yields on conventional products remain low, orthodox monetary policy has sparked a hunt for yield of exactly the kind that regulators now blame for causing the crisis in the first place.