- Most investors in ETFs are familiar with the costs of index tracking where a fund uses physical or cash-based replication (owning the underlying index securities). These costs include possible divergence from the index return as a result of sampling techniques (owning an imperfect replica of the index), the costs of turnover when index constituents change or when cash flows enter and exit the fund, and “cash drag” when dividend payments arrive later than the index calculators assume, to give just three examples.
But what about synthetic or swap-based replication? This technique offers the promise of better index tracking. Issues such as index turnover costs and the timing of dividend payments become the concern of the counterparty writing the swap that promises the ETF its performance, rather than affecting the fund directly. There are also no sampling concerns to worry about. In theory, the fund should track its benchmark with only the management fee causing divergence.
In practice, however, things are not as simple as that. Although it’s generally accepted that swap-based ETFs tend to produce better tracking than cash-based funds, synthetic ETFs do not perform exactly as might be expected; that is, achieving the index return less the management fee.
db x-trackers, for example, recently pointed out that its average fund had underperformed its benchmark by 9.3 basis points in 2009, substantially less than the average management fee charged by the firm (the average db x-trackers international equity ETF has an all-in fee of 35 basis points, while the firm’s average fixed income ETF charges 0.18% per annum). The average fund outperformed its benchmark, in other words, before fees. Meanwhile, some other swap-based funds underperform on the same measure.
db x-trackers explains the kind of outperformance shown by its range in 2009 as the result of “enhancements” such as dividend arbitrage (crediting dividend income to a fund with a smaller deduction for withholding taxes on income than would otherwise occur, something that can be achieved via the bank’s ability to receive the dividends in a more favourable jurisdiction) and income from securities lending.
However, it turns out that there are various moving parts in the engine of a swap-based ETF, all of which can contribute to or detract from the fund’s performance against its benchmark. Chief amongst these are the composition of the collateral basket, the interest rates on the two “legs” of the swap, and the way the index is calculated. Specifically, the tax assumptions built into a total return index can have a major effect.
As one equity derivatives specialist notes, it’s very important to value the collateral basket correctly (ETFs using synthetic replication own a selection of assets, typically bonds and shares, as collateral for the swap they contract with the fund’s swap counterparty).
“A synthetic ETF structure gains exposure to its underlying index by loaning its subscriptions via a collateralised repurchase agreement (“repo”) to a counterparty, usually the sponsor investment bank. The ETF then swaps the LIBOR-based yield on the loan for the total return of the underlying index. Where the collateral for the loan is of lower quality, say Japanese equities instead of G7 sovereign bonds, the return on the loan should be higher. If collateral credit quality is not reflected in the repo pricing this is a cost to the fund which is not disclosed in the expense ratio, and it is an additional profit for the sponsor,” said the specialist.
“Although this type of ETF is usually set up by the sponsor investment bank using an in-house asset manager, it has a separate fiduciary structure in place, usually an authorised corporate director (ACD), or a board of directors. The ACD or board should ensure that the manager obtains competitive quotes for both parts of the swap transaction: the collateralised repo and the total return swap. It may be the case that executing both sides of the swap with one counterparty is the most efficient way to do things, or that the frequency of re-margining offsets the disparity between different types of repo collateral from a credit perspective, but the manager should have to demonstrate that it is acting in the fund’s best interests,” he added.