Demand for corporate bond and high yield ETFs has surged as investors seek higher returns. But they may also get higher tracking errors.
[This article previously appeared on our sister site, IndexUniverse.eu.]
This year has seen increased demand for fixed income ETFs and record inflows into investment grade and high yield corporate bond products. However, some commentators have been critical about the transaction costs and tracking error associated with these products.
Tracking error measures the divergence of an ETF from its benchmark. But this can be hard to interpret and many observers look instead at the actual under- or outperformance of the fund versus its index. Tracking error is a good measure of the quality of a fund’s replication of the benchmark index.
But though they may not realise it, the amount that investors pay in management fees actually contributes towards a fund’s tracking error. There are also three other important factors that may impact on the tracking error.
The majority of fixed income ETFs replicate index performance by physically buying and selling index securities. However, it is not always feasible or cost effective for a fund manager to physically buy everything in an index. Instead the fund manager will buy enough holdings to faithfully represent the index. This is called sampling or optimisation, and is the most common method of physically replicating bond indices.
Sampling can cause the fund’s performance to veer significantly from the benchmark. In the fixed income market this risk is particularly relevant for corporate bonds, where the pricing has a volatile credit spread component that is driven by corporate news and ratings. Being under or overweight a distressed credit can have a big impact on the tracking error and for this reason it is rare to see aggressive sampling in corporate bond trackers.
With the synthetic replication that is frequently used by issuers in Europe, a fund sources the index performance using over-the-counter swaps, typically with an investment bank. Provided the swap is based on the same underlying benchmark as the ETF, the swap counterparty guarantees to deliver the index return to the fund.
Fixed income indices typically have a higher turnover than comparable equity benchmarks.
Much of the turnover exhibited in bond indices is intrinsic to the bond market itself. Most fixed income ETFs use benchmarks made up of bonds with a fixed maturity date and interest payment. As these fixed interest bonds get close to their scheduled maturity date they become less sensitive to interest rate changes and most benchmarks exclude bonds below a minimum maturity. The supply of new bonds can vary widely with market conditions but frequent borrowers typically issue bonds with a range of maturities in order to spread their capital repayments and to take maximum advantage of differing funding opportunities.
At each index rebalancing (usually monthly) new bond issues that meet the eligibility requirements enter the index and bonds below the minimum maturity leave. Corporate bond benchmarks also have rules based on credit ratings, which are also applied during rebalancing and these, particularly changes from investment grade to junk or vice versa, can have a major impact on index composition.
High index turnover is conventionally associated with high tracking error for a fund. As discussed in a recent IndexUniverse.eu article (What’s Inside Your Index), transaction costs are not fully reflected in indices and will therefore contribute to the fund’s tracking error. While some indices attempt to capture trading costs by taking into account the bid/offer spread on entry and exit from the index, others completely ignore costs. Slippage (the difference between the price used in the index and the price at which a fund can trade) also contributes to fund tracking error.
Pimco and Source launched a new ETF this week that attempts to deal with this issue by allowing the benchmark to hold bonds to maturity. The Pimco Short-Term High Yield Corporate Bond Index Source ETF (STHY) is based on the Bank of America Merrill Lynch 0-5 Year US High Yield Constrained Index. Instead of selling bonds prior to maturity and suffering transaction costs the index allows bonds to mature, when it receives maturing bonds’ principal and final coupon payment.
While the technicalities of index replication and the level of index turnover both influence tracking error, the biggest risk for a fund is in not being able to access all of the bonds included in the index.
Traditional bond investors take a “buy and hold” approach, buying new bond issues when they come to the market and holding them until they mature. Small cap issues are hard to source as they quickly become locked up in such portfolios. Bonds that have been in the market for a long time also tend to suffer from this problem. Illiquid bonds often have high bid/offer spreads and poor pricing, but are still included in broad benchmarks for completeness. As ETFs depend on being fully tradeable, an ETF provider is unlikely to be able to include these issues in the fund portfolio.
Many fixed income ETFs use so-called tradeable indices to address this issue. Designed to deliver comparable returns to a broad benchmark but with a more efficient/smaller portfolio, tradeable indices utilise a sampling approach to reduce the benchmark to a small, often fixed, number of highly liquid bonds. The risk with these indices is that over time they may become unrepresentative of the broader market. A case in point is the iBoxx $ Liquid High Yield Index, which started life as the Goldman Sachs $ HYTop Index, a basket of 50 of the most liquid high yield bonds. By 2009 the dollar high yield market had doubled in size and Markit, the new owner of the index, changed the index rules to follow a more traditional large cap based approach.
Tradeable indices can also be too concentrated to support investment demand. If an index becomes too popular fund managers may struggle to source enough of each index bond to match index weights without moving the market. To counteract this, some fund managers use an oversampling approach where they match the underlying index exposure but spread it across more bonds. The iShares Markit iBoxx Euro Corporate Bond ETF (IBCX) is based on the Markit iBoxx Euro Liquid Corporate Bond Index, which comprises 40 bonds, but it uses between two and three times the number of holdings of the underlying index. Oversampling also carries the risk of introducing tracking error. Concentrated indices also tend to have a higher turnover than broad indices.
Tradeable indices are heavily used by providers of synthetic ETFs where index size and ease of replication is particularly important. In Europe Deutsche Bank, Amundi and Lyxor all offer synthetic ETFs benchmarked to tradeable versions of the Markit corporate bond index family, while iShares, ETFLab and UBS all provide physically-replicated ETFs based on the same family.
Some ETF providers, however, make a virtue out of using a broad benchmark. As well as offering ETFs based on tradeable indices, iShares also offers a parallel range of ETFs using broader-based traditional bond benchmarks from Barclays Capital.
Alex Claringbull, senior fixed income portfolio manager at BlackRock, parent company of iShares, says: “The first generation of fixed income ETFs were a product of their time, and a large number of providers tried to change investor perceptions about what they wanted, often because the products were unable to offer more flexibility. More recently providers have changed the way their products can be accessed and responded to investor demand for greater choice. We have the index tracking experience to offer well-known benchmarks in iShares format, for example, but there is room for both types of products, such as the iShares Euro Corporate Bond iBoxx Liquid ETF and the iShares Broad BarCap Euro Corporate Bond ETF, on the shelf.”
Another fund using a broad benchmark is the newly launched Pimco Source High Yield ETF (STHY) mentioned above. Its benchmark covers over 800 securities, from which Pimco select bonds using an internal rating system that attempts to avoid illiquid issues and issuers who are most at risk of default.
Where tracking error is concerned it seems that there are no silver bullets. While tradeable indices seem to offer a partial solution for corporate bond ETFs, they run the risk of becoming unrepresentative of the underlying market and can also suffer higher turnover as a result of concentration. Broader indices are more representative but are harder to replicate and rely more on the skills of a fund’s manager to keep tracking error to a minimum. It seems that there is no escape from the old adage of “know your index and know your manager”.