Blog Posts — September 19, 2022

How to construct high-yield index tracking portfolios  

by Joseph Au-Yeung, Hassan Ennadifi, and Jean-Baptiste Solanet

For fixed income portfolio managers, tracking an index to create high-yield portfolios that are the same or outperform the benchmark can be challenging without the right portfolio construction tools. That’s why we created the Axioma Credit Spread Factor Risk Model. The new model significantly improves risk forecasting thanks to its enhanced factor granularity and coverage of credit default swaps.  

In a new research paper, we showcase how a portfolio manager can replicate fixed income indices using the Axioma Portfolio Optimizer and the new Axioma Credit Factor Model to create optimal US high- yield portfolios. In the workflow, we replicate a US high yield index with liquid bonds and a set of derivatives and test the solution from a risk perspective. The end goal is to create a set of portfolios that is more cost efficient, as (or more) liquid and as diversified as the index.  

We explored two dimensions in this replication problem – liquidity and the percentage allowed to cash bonds investment with the intention to get a replicating portfolio as liquid as possible with a minimum cash commitment. Those two dimensions require a tradeoff being made – higher liquidity implies higher tracking error as we shift the universe of eligible bonds towards the IG territory. After rebalancing, we test the solution by looking at key analytics on those portfolios to understand which one(s) are the most adequate. 

Stress testing the replicating portfolios 

We start by comparing risk with a granular approach using the Axioma Credit Spread Curve Risk Model and different time periods for a full repricing historical simulation. The granular approach is closely aligned to the pricing factors and is based on the granular issuer or cluster curves. Comparing the two risk forecasts is a good way to assess the accuracy and relevance of the Axioma Credit Factor Model. We highlight two portfolios that achieved close to 50bps tracking error (which is just under 10% of the benchmark volatility): 

  • Portfolio 1: 60% Bond Allowance with liquidity > 5.5 
  • Portfolio 2: 70% Bond Allowance but with liquidity > 6 

We conducted a few stress tests based on historical events as well as two transitive stress tests (equity-based but propagating the credit/rate curves).  

We plot the stressed PnL for the benchmark as well as the portfolio optimized with less constraint along with our two portfolios. We also plot the derivative-only portfolio. 

We notice that under stressed conditions portfolio 2 behaves quite similarly to the benchmark, and the optimized portfolio under fewer constraints matches the benchmark extremely closely, though still providing more liquidity than the HY index components. 

Sovereign crisis shows rates going down and credit going up, overall resulting in a minimal impact in July 2011. Equities went down overall 15% hence the “no bond” solution being more impacted as it is more exposed to equity. 

Source: Qontigo

Download the whitepaper here