Editor's Note: This blog has been updated to include more recent data. It was originally published in February 2018..
This post was written with contributions from Patrick Bresnehan, Managing Director and North America Head of Fixed Income Beta Solutions Group.
As I discussed in previous blogs, index tracking methodology for equities can be constructed using either a full replication (where an ETF holds all of the securities that make up the index in their respective weights) or optimization (where an ETF holds a subset of securities that closely resemble the index.) Index tracking methodology for fixed income ETFs can be full replication as well, but it can also work a little differently by using a sampling strategy.
To learn more about State Street Global Advisors’ approach to sampling, I met with Patrick Bresnehan, Managing Director and North America Head of Fixed Income Beta Solutions Group.
Below are excerpts from our conversation:
Matt: Can you provide an overview of fixed income ETF portfolio sampling?
Patrick: Sampling can be the most efficient technique for constructing portfolios as many broad fixed income indices include a large number of securities, but not all of those securities can be purchased. Couple this with potentially high transaction costs to access illiquid bonds and full replication isn’t always possible or practical. With a sampling approach, rather than owning every security in an index, we seek to build a portfolio with the same characteristics as the index.
Matt: With sampling, what are some of the most important considerations to ensure the portfolio is able to efficiently track its underlying benchmark?
Patrick: From a high level, we generally take two approaches to ensure our tracking error, as a result of exposure differences, is limited:
- Top-down approach: We seek to align the common factors of the ETF to the index. Common factors include:
- Duration: how do we match on key rate duration exposures
- Credit spread: examining differences between option adjusted spread as well as other metrics like option adjusted spread duration
- Sector exposures: looking at the sector and industry compositions to manage macro impacts
- Ratings: how are we allocated at the credit rating level
These factors are typically the key variables that drive market beta.
- Bottom-up approach: This is often used in markets such as high yield or convertible bonds where we typically find more price volatility. In a bottom-up approach, we try to identify large or outsized idiosyncratic risks and mitigate them. An example of this is making the decision to purchase one bond vs. another from a company based on its position in the credit curve, a factor that can impact single bond volatility.
Matt: Are there specific constraints that need to be met from a geographic, sector, credit rating or duration perspective?
Patrick: It’s not necessarily constraints, but rather ensuring the portfolio has the appropriate exposures for its market beta. We need to ensure major exposures are in line with the benchmark by taking a top-down approach as well as taking a bottom-up approach to align idiosyncratic risks.
We also look at cost per exposure (similar to a bid-ask spread or transaction cost) when prioritizing these various betas. A good example in the high yield space is a bond that is trading yield-to-call. If that bond comes into the index, we will weigh the cost of adding that bond to the portfolio given its potential short holding period.
Similarly, if we hold a bond that does get called, the index will generally carry the bond’s market weight exposure to month-end when the index rebalances. The ETF will appear to be underweight that sector despite the fact that the bond no longer exists. In this instance, we would not buy a similar bond to trim that underweight. We know the sector weighting is overstated because the bond has been called, and it will smooth out during the next month’s rebalance when positions are re-aligned. This approach helps minimize transactions costs, a primary factor in tracking error—an index doesn’t have to incur trading costs, but the ETF portfolio does.
Matt: Does the type of fixed income exposure dictate your approach? It sounds like there’s a pretty stark difference between high yield and other areas.
Patrick: Yes. Market inefficiencies and lack of liquidity make it challenging to obtain particular exposures. Rather than fully replicate a particular portfolio to match the index, we trade the portfolio within the market context. This may result in minor mis-weightings of the ETF relative to the index at the security level, but when we look at index tracking through the cost of exposures lens, we’ve tried to manage transaction costs and limit the performance friction (expense ratio + transaction costs) between the index and the ETF.
In other asset classes, such as those in the Bloomberg Barclays U.S. Aggregate Bond Index (the Agg), there have tended to be fewer market inefficiencies. These asset classes have historically exhibited strong liquidity and volume, so exposures have been obtained relatively efficiently.
Matt: How does a fund’s AUM (assets under management) affect the sampling process?
Patrick: It’s two-fold. There’s very small AUM, but there’s also small AUM relative to the index. If it’s small AUM on a broad index, the ETF would typically be highly sampled in an effort to get the best execution.
An example would be an ETF with $100 million of assets tracking the Bloomberg Barclays Global Aggregate Bond Index, a benchmark with 24,051 securities1 in it. That would be an area where we would just have to sample. On the other hand, for an ETF with $100 million of assets tracking the Bloomberg Barclays US Treasury Inflation Notes 1-10 Year Index—a benchmark with 29 securities in it—we would more likely than not fully replicate it.
Matt: Can you invest in securities not included in the index?
Patrick: We can and we do for certain ETFs. This goes back to minimizing transaction costs: sometimes we use out-of-index securities as a proxy for index positions. These bonds may be from the same issuer, but they aren’t included in the index because of various rules (e.g., they are below the index’s minimum size requirement).
We are also active in the primary market buying new issues because they typically come with a discount to create new buyer demand. We know those new issue bonds will be added to the index at the next rebalance. This means we minimize our transaction costs by avoiding having to pay a higher price for those bonds at the end of the month when they roll into the index.
Matt: What are some of the potential drawbacks to sampling?
Patrick: It can create short-term variations in an ETF versus the index it seeks to track, but those variations should equal out over time. For example, if a portfolio has moderate drift it can be course corrected during the creation/redemption process or by on-going maintenance used to smooth out exposure differences, and therefore potential variations.
Matt: If an investor is looking at fixed income ETFs, are there any specific considerations if one fund is fully replicated while another is sampled?
Patrick: In markets with liquidity challenges, it makes more sense to sample to reduce tracking error rather than fully replicate the index—again, focusing on the cost per exposure to mitigate transaction costs the funds have to incur, but the index doesn’t. Overall, however, as long as a fund is able to maintain minimal tracking error the concern of whether or not a fund is fully replicated or sampled should be abated. Nonetheless, it is an important part of the due diligence process, and I hope our dialogue has provided insights into how we, at State Street Global Advisors, apply different techniques as we seek to deliver efficient beta exposure.
To learn more about the mechanics of ETFs, as well as how to evaluate and implement them in portfolios, sign up to follow SPDR® Blog.
1Bloomberg Finance L.P., as of 8/22/2019
Authorized Participants (APs)
Authorized participants, or APs, are US registered, self-clearing broker-dealers who meet certain criteria and sign a participant agreement with a particular ETF sponsor or distributor to become “authorized participants” of the fund. APs are highly scrutinized for their integrity and operational competence as they are the only parties who transact directly with the ETF.
A single unit of at least 15 stocks that are used in program trading, index fund management and currency portfolio management. Baskets are traded on both the NYSE and the CBOE for institutions and index arbitrageurs.
The difference between the highest price a buyer is willing to pay for an asset and the lowest price the seller will accept. Bid/ask spreads are a key measure of the liquidity of an asset or security.
Bloomberg Barclays Global Aggregate Bond Index A broad-based measure of the global investment-grade fixed income markets. The three major components of this index are the U.S. Aggregate, the Pan-European Aggregate, and the Asian-Pacific Aggregate Indices. The index also includes Eurodollar and Euro-Yen corporate bonds, Canadian government, agency and corporate securities, and USD investment grade 144A securities.
Bloomberg Barclays US Aggregate Bond Index (the Agg)
A market-weighted index, meaning the securities in the index are weighted according to the market size of each bond type. Most US traded investment grade bonds are represented. Municipal bonds and Treasury Inflation-Protected Securities are excluded, due to tax treatment issues. The index includes Treasury, Government agency bonds, Mortgage-backed bonds, Corporate bonds and a small amount of foreign bonds traded in the US.
Bloomberg Barclays US Treasury Inflation Notes 1-10 Year Index A benchmark designed to measure the performance of the inflation protected public obligations of the US Treasury commonly known as "TIPS" that have a remaining maturity greater than or equal to 1 year and less than 10 years. TIPS are securities issued by the US Treasury that are designed to provide inflation protection to investors. The index includes publicly issued, TIPS that have at least 1 year remaining to maturity and less than 10 years on index rebalancing date, with an issue size equal to or in excess of $500 million.
Hybrid investments that combine characteristics of stocks and bonds. Convertible securities can be exchanged, at the option of the holder, for a specific number of shares of the issuer's preferred stock or common stock or even cash.
Credit curve is essentially the spread over treasuries of various maturities for a single bond issuer. The greater the difference between the front end (near term maturities) versus the long end (longer maturities) determines the steepness of the credit curve.
The sensitivity of the price of a bond to a 100 basis point change to its option-adjusted spread.
A risk that is endemic to a particular asset such as a stock and not a whole investment portfolio.
Credit spreads are the difference in yield between any type of bond, and a US treasury of the same maturity. Corporate bonds, which carry a risk of default, yield more than US Treasury Bonds, which carry no risk of default.
Option-adjusted spread (OAS) is the yield spread which has to be added to a benchmark yield curve to discount a security's payments to match its market price, using a dynamic pricing model that accounts for embedded options. OAS is hence model-dependent. ... OAS is usually measured in basis points (bp, or 0.01%).
Tracking error is a measure of how consistent a portfolio’s return is with that of its benchmark. In reality, no indexing strategy can perfectly match the performance of the index or benchmark, and the tracking error quantifies the degree to which the strategy differed from the index or benchmark, by measuring the standard deviation between the two values, annualized.