Exchange traded funds – the low-cost substitute for mutual funds and other actively managed products — have taken a diversity of forms recently as they encompass more markets and more styles of trading, such as short-selling, and that has some concerned about systemic risks.
In the U.S., for example, most ETFs that track fixed income or equity indexes, physically replicate the index they follow by buying the constituents of the index. In Europe, only half of ETFs do so; the others use synthetic replication strategies.
Zone reason, points out Srichander Ramaswamy, in a working paper for the Bank of International Settlements called “Market structures and systemic risks of exchange-traded funds,” is because of regulatory differences. U.S. regulations require investment funds to hold “80% of their assets in securities that match the fund’s name.” By contrast, European regulations allow investment funds to use exchanged-traded and over-the-counter derivatives.
Derivatives serve as a cheaper way to get access to less liquid markets, he notes. “However, physical replication can be an expensive method for tracking broad market indices such as emerging market equity or fixed income indices, or other less liquid market indices. Including only a subset of the underlying index securities for physical replication can lead to significant deviation in returns between the ETF and the index in volatile market conditions. Furthermore, in less liquid markets the wider bid-ask spreads increase replication costs, particularly when the fund has high turnover.”
One way to implement this is through an unfunded total return swap. Typically, the ETF sponsor swaps cash in return for an index’s return, and receives as collateral a basket of securities that may not be related to the target index. A motivation, Ramaswamy suggests, is “that this structure exploits synergies between banks’ collateral management practices and the funding of their warehoused securities.”
European regulations also introduce some complexities. The collateral need only be 90% of net asset value, leaving the potential for a 10% loss should the swap counterparty default. In addition, the collateral is available for securities lending.
Another, less common, option is a funded total return swap. Here the collateral is transferred to a custodian bank. But the sponsor is not the beneficial owner; the swap counterparty is. The swap is also overcollateralized, by between 10% and 20%.
Ramaswamy sees a problem with synthetic replication. “[T]here is a trade-off: the lower tracking error risk comes at the cost of increased counterparty risk to the swap provider. ETF sponsors tend to emphasise the lower tracking error and downplay the counterparty risk to support the case for synthetic replication schemes, which are also marketed as being cheaper than the alternative method of replicating the index in the cash market. In reality, the increased popularity of ETF products among investors has led to greater competition between ETF sponsors, forcing them to seek alternative replication techniques to optimise their fee structures.”
The competition to reduce costs can lead to another problem. ETF providers become more reliant on their bank sponsors. Thus, there are “synergies arise from the market-making activities of investment banking, which usually require maintaining a large inventory of stocks and bonds that has to be funded. When these stocks and bonds are less liquid, they will have to be funded either in the unsecured markets or in repo markets with deep haircuts. By transferring these stocks and bonds as collateral assets to the ETF provider sponsored by the parent bank, the investment banking activities may benefit from reduced warehousing costs for these assets. Part of this cost savings may then be passed on to the ETF investors through a lower total expense ratio for the fund holdings.”
To take an example, there is an emerging market ETF that tracks the MSCI index. In the collateral basket are securities from mostly advanced countries, such as Japan, Germany and the U.S. There is a similar composition of bonds in the collateral basket, but only 16% are rated AA or better. What is occurring, Ramaswamy suggests, is that banks are taking lower-quality assets off their books to reduce their Basel III capital requirements.
Synthetic structures may thus make for an increased potential for systemic risk. Here, Ramaswamy outlines four transmission mechanisms: “(1) co-mingling tracking error risk with the trading book risk by the swap counterparty could compromise risk management; (2) collateral risk triggering a run on ETFs in periods of heightened counterparty risk; (3) materialisation of funding liquidity risk when there are sudden and large investor withdrawals; and (4) increased product complexity and options on ETFs undermining risk monitoring capacity.”
There is an irony here: ETFs were supposed to make investing simpler and cheaper. Apparently, cost competition has led to greater, not lesser complexity – and with it, more risk.