By Paul Amery
An interesting article in the most recent iShares monthly newsletter caught my attention. Its authors make the point that comparing ETFs and ETNs or ETCs simply on the basis of the total expense ratio may not give a true picture of the real cost to an investor when choosing a particular tracker.
The iShares authors suggest paying attention to various additional factors when evaluating exchange-traded securities, and in particular:
- Trading costs—bid/offer spreads and brokerage commissions
- Rebalance costs—from the inflow and outflow of securities during regular rebalancings of the underlying index
- Lending revenues earned or foregone—from securities lending within an ETF, and from lending of the ETF units themselves
These costs can all be quantified, more or less. Trading costs can be measured by looking at the liquidity of the assets underlying the exchange-traded security, and most leading European exchanges publish data showing historical ETF bid/offer spreads in standardized trading sizes.
Rebalance costs are harder to quantify, but one can look at the stability of different versions of indexes and estimate how many index components are likely to change, how frequently and what the likely cost is. Clearly, some indexes are likely to change composition to a greater extent than others—market-cap-weighted indexes tend to have lower turnover than indexes with alternative and/or equal-weighting structures.
In addition, potential ETF lending revenues—which would reduce the cost of ownership to an investor—can be assessed by comparing issuers' policies on sharing revenues with investors from lending within their funds, as well as by checking the likely returns to be generated from lending the ETF units themselves (if you are an institutional investor).
A typical calculation might run something like this, according to the iShares report.
click to enlarge
In other words, a difference in the headline total expense ratio between two ETFs may easily be compensated for by differences in the three other factors mentioned above. In the given example, ETF 1 has a higher total expense ratio than ETF 2, but makes up for it via lower trading costs and greater stock lending revenues.
This all seems valid stuff, and is certainly helpful advice for an investor, even if it means that making a sensible choice about the right fund is going to mean some digging around for information.
But that's not the end of it. Further cost differences are implicit in the differing varieties of ETF themselves, and between ETFs and other exchange-traded securities such as ETNs and ETCs. To get an idea of what these might be, let's look at another table from the iShares report.
*Swap-based ETFs generally require all Primary Market trades to be routed to one internally designated swap counterparty.
**If the ETF tracks a Price Return Index, the dividends will be periodically distributed. If the ETF tracks a Total Return Index, the dividends will be reinvested.
There are various additional and quite subtle differences, all affecting an investor's potential "value" from an ETF/ETC/ETN, that could be highlighted here - from UCITS compliance or noncompliance, to the number and depth of authorised participant relationships, to dividend reinvestment or distribution policies.
But one line in the table does stand out as a real, and also a quantifiable, additional cost: Counterparty risk. And, as we will see below, the market is currently telling us that this cost is potentially a multiple of those we discussed earlier.
To clarify the difference between the different types of tracker, let's review how they work.
A traditional ETF holds the underlying securities that track the index, or at least a representative sample of them. The attraction of this arrangement, as with most mutual-fund-type structures, is that in the case of failure of the fund issuer, the investor has recourse directly to the pool of underlying shares or bonds. As the Borsa Italiana ETF site explains, for example, with regard to UCITS-compliant ETFs:
"It is a known fact that UCITS have segregated assets with respect to those of the companies, which take care of their creation, management, administration and marketing activities. Therefore, ETFs are not subject to the insolvency risk in the event of default of the above-mentioned companies."
If the issuer goes bust, the investor still has access to the basic securities.
There is, however, another kind of ETF that does not offer this same relationship—a swap-based ETF. A swap-based ETF depends on derivative contracts, often written by third parties, to provide exposure to the market. This creates the potential for a new kind of risk—that of the swap writer failing to fulfill its obligations.
In the case of a UCITS-compliant swap-based ETF, the counterparty exposure to the swap writer is capped at 10% of the fund's net asset value. This exposure can be reduced to a smaller amount, either by a policy of exchanging collateral between the fund and the counterparty to cover unrealised gains or losses, or by resetting the swap to zero (i.e., settling accounts, and starting again) on a regular basis. Here, one needs to check the relevant ETF prospectus to see what policy the manager follows. In a worst case scenario, an ETF owner could in theory be left 10% out of pocket if the swap counterparty failed to pay up.
It makes sense to understand whether your ETF is a traditional "in-specie"-based ETF that holds the underlying asset, or a swaps-based ETF that comes with counterparty risk.
Can we quantify the cost of the counterparty risk for a swaps-based ETF? In theory, yes, if we know who the swap counterparty is, and if we know whether there is any money due to the ETF from the swap counterparty that has not been covered by collateral. We could then work out how much it would cost to remove the counterparty risk by buying a credit default swap ("CDS")—an insurance policy to hedge against the risk of the counterparty failing to perform.
In practice, however, it would be very difficult to obtain the necessary information to do this. Most ETF managers only offer an annual snapshot of fund collateral, for example, and may not reveal the names of swap counterparties at all. The risk is indeterminate, if real.
ETNs And ETCs
With ETNs and ETCs, things are different. Here, the investor has direct counterparty exposure to the issuer (in the case of ETNs) or to third parties guaranteeing the securities' performance (Shell Treasury for ETF Securities' energy ETCs, and AIG for most other ETCs—with the exception of some precious metal ETCs which are backed by physical holdings).
The potential exposure is also larger, as the whole value of the ETN or ETC is tied to the financial security of the swap provider.
Fortunately, we can more easily quantify the "cost" to an investor of owning the underlying investment via an ETN/ETC by checking how much one would have to pay to remove the issuer's or third party guarantor's credit risk from the equation, using credit derivatives.
For ETN investors, the results may come as a shock. The cost of insuring against major investment banks' nonperformance—the cost of hedging out the risk that the ETN or ETC provider may go bust—has soared over the last year, reaching several percentage points per annum in some cases. See the chart below, courtesy of Credit Derivatives Research LLC, which shows the average five-year CDS spread in basis points for 15 major banks, both US and European.
After peaking near 2.5% following the Bear Stearns debacle, the cost to insure receded a bit. But as of mid-July, it was still well above 1.5%.
Of course, different banks carry different risks, and it pays to look at the specific underwriter for a specific ETN. But suffice it to say that these represent real (and large) costs, which should be added back to the total expense ratio when evaluating an ETN.
Now of course one advantage of ETNs is that they offer the ability to track areas of the market where physically holding the underlying asset may be difficult or expensive—commodities, for example—so this must also be taken into account. But no comparison of ETNs to ETFs should omit the question of issuer risk for the former, and how much it costs to insure against the counterparty's nonperformance.
What about the swap-based ETCs from ETF Securities, which are backed by third-party contracts? Again, we should check how much it would cost to remove the performance risk of Shell Treasury, or AIG, depending on which ETC we're looking at.
I've been unable to track down CDS spread data for Shell, but the spread chart of AIG (NYSE:AIG) is given below, with data from the beginning of last year. The cost of insuring against default by the US insurance company has surged by 25 times—from 10 basis points to 2.5 percent per annum—since January last year. Again, this must be regarded as a hidden "cost" of investing via the ETC—since, if an investor wanted to remove this counterparty risk, this is what they would have to pay to do so—and its magnitude dwarfs that of the costs discussed earlier in this article.
Now this may not be the end of the story—there could well be other, offsetting considerations for an investor, such as the difficulty in obtaining commodity exposure (especially to those raw materials incurring large storage costs) by other means. But the counterparty risk component of the AIG-guaranteed ETCs has jumped hugely over the last year, making the true cost to an investor substantially higher than the total expense ratio alone.
Unfortunately, most financial market product design continues to rely on ratings from the main credit rating agencies, which have proved almost useless in predicting many recent events of credit distress, and which have fallen into disrepute. Perhaps the next generation of exchange-traded products will pay more attention to what the credit derivatives market is telling us, or at least offer investors a way of hedging such counterparty exposures.
So my brief survey of costs has suggested that the largest one in some tracking instruments is one that is hardly mentioned by financial institutions—counterparty risk. The advent of the credit derivatives market has made it relatively easy to quantify, and no comparison between the ETFs, ETNs and ETCs should ignore it.
 "Selecting the right ETF - not just a question of cost," iShares monthly, July 2008.