As discussed in the first part of this series, over the remainder of the summer, I’ll be performing deep dives into the intricacies of fixed income attribution across asset types. I have a simple goal in mind: to improve the quality of our attribution analysis, regardless of the investment mandate, by breaking out commonly used effects, sample reports, and tools for interpretation.
The Credit Quality Rollercoaster
The second stop on our trip takes us on a credit quality rollercoaster from core investment grade to high yield.
Why does the rollercoaster metaphor work here? A few reasons. First, high yield bonds exhibit far greater return volatility than investment grade bonds; if a macro event causes investment grade to sneeze, high yield catches pneumonia. Second, credit events tend to unfold quickly (and sometimes unexpectedly) in the space, like a sudden drop or an unforeseen corkscrew. And finally, anecdotally across our global client base, high yield attribution tends to throw them for a loop!
What I mean with that last bit is that client by client, I do not see one specific approach to attribution. Some prefer a Brinson approach, only looking at allocation and selection while ignoring other drivers of fixed income total return. Others opt to apply the basic investment grade model to maintain continuity across the overall asset class. Still others have adopted more of a hybrid approach. The inclusion of a Duration Times Spread, or DTS, methodology to the mix adds even more choice in trying to explain excess return in the asset class proportionally.
While I believe that there is a place for each of these in high yield attribution, I advocate a hybrid approach alongside a DTS model in terms of value add and explanatory power. That said, as there is no overarching approach, it is worth running through the slate to determine which is the best fit given the investment process and mix of asset types on play.
KISS (Keep It Simple, Sunshine)
One of the most common refrains I hear from high yield managers is something along the lines of “high yield is all a relative value play; interest rate exposure is a knock-on effect; sector allocation and security selection are all I care about.” And, depending on the assets types involved, that can be true.
Case in point, utilizing CDS/CDX for a pure credit play, entering short treasury futures positions to neutralize duration, or investing in bank loans (which are generally accepted to have minimal interest rate risk because of their floating rate provisions), are all great examples where it makes sense to keep it simple and showcase the investment process through a basic Brinson approach. This tells the story of the investment process, is easy to implement, and even easier to interpret.
But, as I will illustrate, even if relative value is the name of the game in high yield, these securities are still impacted by interest rates, and even more importantly, by spreads. With that in mind, why wouldn’t our model take that into account?
Introducing a Hybrid Approach
Tweaking the initial investment grade model allows us to tell the high yield story more impactfully while also taking fixed income return fundamentals into account. As we can see from our total return decomposition below (Figure 1), even high yield is sensitive to the level and movement of rates. While rates may not be a primary driver in the investment process, they remain actionable (I’d argue that duration positioning in high yield is still very much an active management decision) and should be considered. The difference relative to investment grade, though, is that we may not need or desire the same level of detail.
The first tweak is simple; we will simply combine the shift and twist effects into an overall yield curve effect. The second tweak is to expose the spread effect to showcase our spread management acuity. The third tweak is to expose the income effect since, definitionally, we should want to understand how the yield component of security selection impacts our high yield relative performance. From here we maintain the allocation and selection effects or combine those into a single excess return component. Figure 2 below is a great sample view that is easy to use and explain. More importantly, it leverages an out-of-the-box approach and analytics.
Over the years, I have also heard of the desire to utilize sector curves in the credit space. Certainly, there is pride of place for that in the relative value process. The issue I see with that in attribution is that creation, maintenance, and integration of sector curves in a standard model are difficult. Why? Simply put, not every credit quality, sector, and/or tenor have enough securities to construct reliable curves on an ongoing basis. We resolve that by expanding the spread effect to look at sector level as well as security level spreads. In Formula 1 this treats the benchmark as the investable universe, removing the sector curve requirement. Then, in Formula 2, the security spread effect explains how much of the overall spread effect is derived from additional OAS movements not explained by sector spread moves.
While this is an approach that is easy to use, maintain, and interpret across interested parties, I do see a pivot towards an increasingly popular approach, Duration Times Spread, or DTS.
What Can DTS Do for You?
Let’s start with a basic definition. DTS is as simple as it sounds: spread duration times spread. This stems from feedback and observation that spreads tend to change on a relative basis rather than an absolute basis. Larger spreads widen at faster rate than tighter spreads. This introduces a level of comparability across the credit spectrum, where an easy example states that a security with a spread of 400 bps and spread duration of 1 is directly comparable to a security with a spread of 100bps and a spread duration of 4.
In an attribution framework, shown in Figure 3, we can see how easy this approach is to implement and interpret. We still capture the impact of interest rate sensitivity, but we can isolate spread moves and relative value in a somewhat more intuitive way than using the extended investment grade model.
High yield is a unique part of the fixed income world that tends to be viewed apart from more vanilla strategies. Volatility, optionality, and a demand for bespoke views have created fragmented approaches to attribution. By adapting our baseline fixed income attribution model to address these nuances while maintaining laser-like focus on the investment process, we create an environment that dramatically simplifies understanding relative performance in the asset class.
Vice President, Sales Manager, Fixed Income, FI Analytics
Pat Reilly has 15 years of experience within the investment management industry focusing on fixed income. Prior to joining FactSet, Pat started his professional path as a Credit Manager at Wells Fargo and then as an Investment Services Analyst at Pacific Life. He holds a Bachelor of Science, Finance from the Eller College of Management at the University of Arizona and a Master of Business Administration from the Marshall School of Business at the University of Southern California.