Exchange Traded Funds

Exchange traded funds (ETFs) have recently received a lot of negative press, some such as the comparison to collateralised debt obligations is inaccurate in our view but there are some credible critiques which serve as a reminder of the importance of "looking under the bonnet" before investing in these passive products.

ETFs are funds that can be traded intra-day on exchange and allow investors to achieve a diversified exposure to an asset class through close replication of an index with the minimum tracking error, that is the difference in performance between the relevant index and the ETF. A report by Blackrock suggests that there are close to 2,500 ETFs offered by around 130 sponsors and traded on over 40 exchanges globally as at the end of 2010. The ETF market grew by 27% in 2010 to $1.3 trillion.

There are two techniques used by ETFs to replicate an index: physical replication or synthetic replication via the use of a total return swap. Physical replication occurs where the ETF sponsor buys all of the constituent securities in an underlying index and holds them in the fund. However the costs of doing this can be high and so in markets where trading costs are more elevated e.g. emerging markets, optimisation is often used, where a fund holds a sub-set of the underlying index rather than every constituent. This does mean that there is a greater risk of underperformance and in fact the US-listed iShares MSCI EM ETF uses optimisation and underperformed by 8% in 2009.

Physical replication is not free from risk, the majority of physical ETF providers lend out the securities held to lower the tracking error so creating exposure to counterparty risk. Who are the stock lending counterparties and how are they chosen? The payment of dividends can also cause tracking error as an ETF typically pays out dividends received from the underlying shares it holds on a quarterly basis, but the underlying stocks pay dividends throughout the quarter resulting in some cash drag.

In more illiquid markets, particularly where the target index is broad, there is greater potential for an optimised ETF to underperform and investors may look at using a swap-backed ETF. In a swap-backed product, when an investor sends monies to the ETF sponsor, that sponsor then buys a basket of equities. The ETF then separately enters into a total return swap with an intermediary, using the basket of equities as collateral. The ETF sponsor will pay the returns on the performance of the basket of equities to the investment bank and the investment bank will pay the performance of the target index for the ETF back to the ETF sponsor (see below diagram). In this way the difference in performance between the basket of stocks held and the target index is mitigated. The key issue is you are reducing your tracking error risk but taking on more counterparty risk!

The nature of the swap is important, "unfunded" swaps mean the assets in the collateral basket of equities are directly owned by the ETF on behalf of investors in a segregated, ring fenced account such that if the swap counterparty defaults investors’ assets can be recovered quicker. In a funded swap structure, the collateral is posted into a ring fenced custodian account to which the ETF has legal claims but is not the beneficial owner. This can potentially lead to delays in realising the value of assets if the swap counterparty fails.

The less frequent the reset of the swap, the more exposure to counterparty risk. If the target index has performed better than the basket, upon reset (the netting of both legs of the swap) the investment bank must pay that additional performance into the fund which exposes investors to counterparty risk as they are dependent on the bank’s ability to pay. Some swaps automatically reset when the counterparty risk exceeds 10% of the NAV (UCITS rules) and others reset daily or monthly.

Counterparty risk can be further mitigated by over-collateralisation, meaning the value of collateral backing the swap exceeds the notional value of the ETF. It is not always the case that the collateral behind the swap is derived from the asset class of the benchmark index. Collateral tends to be OECD listed equities so investors wanting exposure to Asia and emerging markets will suffer an asset mismatch.

Some ETF providers have numerous counterparties which is beneficial as swaps can be transferred quickly in the case of insolvency for one of them. However, this adds a layer of complexity from our perspective as it means that in a credit event, we would not know who our swap counterparty was and consequently how much risk we were exposed to.

Many of the issuers of ETFs sit within large investment banks which also have desks that act as market makers in ETFs. Clearly there is a conflict here in terms of determining trading spreads so it is important to check that a bank’s ETF traders are not remunerated more for trading products from a particular provider.

With the exception of the physically backed commodity ETFs (precious metals) all synthetic commodity ETFs are unable to track exactly the performance of the spot price because commodity indices are constructed using futures which need to be rolled forward on a monthly basis, incurring a cost if the curve is upward sloping (in contango) that eats into investors’ returns. The graph below shows an investor in the ETF Securities Oil ETF (the black line) would have underperformed the spot price of oil (the green line) by a massive 73.98%.

Recent criticisms of ETFs in the financial press

There have been numerous new players that have entered the ETF market, many of which are investment banks and it is important to question the rationale for this. The market makers in these banks usually require a large inventory of stocks and bonds that has to be funded, this can be expensive if the securities required are less liquid. By transferring these stocks and bonds to an ETF provider as collateral, the investment bank reduces the warehousing costs of these assets (Bank for International Settlements (BIS) 2011). Part of these costs may then be passed down to the ETF investors through a lower TER.

Regulatory capital charges such as the Basel 3 laws are punitive for lower quality and less liquid assets but they can be significantly reduced by including them in the collateral basket for a swap based ETF and consequently one can see how it is in banks interests to provide these products as they can substantially improve their capital ratios.

There is little transparency as to how swap counterparties replicate index returns to meet their contractual commitment to deliver the total return on the index via the swap. BIS predict that they use a similar technique to their physically replicating peers, primarily through optimisation. However, the risk of underperformance or tracking error could be co-mingled with the rest of the risk on the trading book. This means that assessing the risk of the end product sold to investors, like ourselves, is very difficult.

BIS also highlight that the ability of the swap counterparties to bear the tracking error risk in the event of a sudden and large liquidation of ETF is untested. One cannot ‘gate’ an ETF in the way that we have seen hedge funds do so. It is not difficult to foresee a period where swap-backed ETFs were aggressively sold due to heightened counterparty risk. The fact that in some ETF asset classes there is little overlap between the collateral basket and the target index would further encourage institutional investors to sell quickly. This could cause liquidity to severely decrease. It is worth noting that ETFs only make up 4-5% of the total assets held in mutual funds.

Our conclusion is that synthetic ETFs do have a part to play in minimising tracking error, however, it is imperative to carry out rigorous due diligence and understand the nature of the counterparty and operational risks associated with the particular structure. Some of these structures have been untested in market stress and so we are cautious about their use.

By Mona Shah
Research analyst