In our previous blog post we discussed the use of attribution analysis and fixed income attribution methodologies to analyze bond portfolios. Surprisingly, attribution analysis is not widely used for bond portfolios; and of those using fixed income attribution, many are dissatisfied to some extent with the results it produces. The reasons often given are that it’s hard to understand and it’s hard to implement.
Why is fixed income attribution hard to implement?
Two major factors contribute to the complexity of the fixed income implementations in the industry.
First, the lack of industry standardization for fixed income methodologies which then introduces a level of complexity within the teams of an organization.
A second major factor contributing to the difficulty of the implementation of a fixed income solution is data. The sensitivity of fixed income attribution solutions to data sourcing and data integrity means that implementations can face a number of data management challenges.
Data Sourcing challenges occur because of the additional data demands of fixed income attribution when compared with Equity attribution. From obtaining additional characteristics for modelling fixed income instruments, to sourcing complete sets of index data, yield curves and analytics.
Data Integrity challenges occur due to the sensitivity of the attribution results to the integrity of the data used; from the integrity within the data source to the integrity between data sources e.g. price alignment of analytics and accounting data. It is particularly important to ensure consistency of inputs between analytics and yield curves.
Instrument Modelling can be a challenge for fixed income portfolios due to the wide range of complex instruments. From bonds with prepayment events, default events, inflation adjustments, underlying instruments etc. The accuracy and value of the results of fixed income attribution hinges on the ability to model these events accurately. This task is complicated by the fact that different exchanges and different accounting systems are not consistent in terms of how these events are treated.
How can we make it easier to implement?
There are three steps that BI-SAM uses to simplify fixed income implementations:
Address the complexity, we take a pragmatic approach to how the portfolio is modelled. The beauty and complexity of fixed income solutions lies in the granularity and detail that can be provided about the drivers of return; leading to a temptation to calculate all of the effects available in the model. Our recommendation is to focus initially on getting the major drivers of return accurately modelled. How is a portfolio affected by changes in the market: currency changes, passive return, changes in the curve and spread? By simplifying the attribution requirements, more focus can be directed to the effects that have the biggest impact. Once experience, sophistication and data capabilities have developed, then the second order effects can be addressed.
Address the data challenges, we focus early on identifying data issues. This is a powerful way of increasing the ability of the implementation team to plan and control how to address those issues and ultimately reduce project risk. To facilitate the early identification of these issues, at BI-SAM we recommend modelling “pilot” portfolios being used early in the implementation process to help identify data issues. We have a series of data readiness tools and processes to check for common issues, for example: checking the consistency of an instrument’s yield against the spread; analytics against absolute and relative thresholds; consistency of key rate durations against durations; spikes within data sources.
This theme remains a key component throughout our implementation lifecycle with more emphasis on data integrity analysis as the project progresses. This typically has been one of the strengths of the BI-SAM B-One solution, and as we evolve the offering and our methodologies, we are increasingly focused on fixed income specific checks. For example we have recently introduced missing data checks and checks for data spikes specifically for fixed income.
Finally, having identified these issues, it is left to the fixed income solution to be flexible enough to cope with the data limitations without adverse impact on the attribution story. For example, some instruments e.g. Structured Products may not have the data required to model the performance available. BI-SAM’s B-One solution provides the ability to implement custom treatments for complex products. Other times, dexterity may be required to handle certain events; for example Bond-default events. BI-SAM provides the ability to implement custom rules for specific events.
Do not be afraid of implementing fixed income solutions
Fixed income implementations can be daunting; but with the right approach consisting of simplifying the process (phasing and targeting the analysis of results and the complexity of the implementation) and by focusing on data integrity issues and gaps earlier, these implementations can be manageable. The benefit of having a good fixed income attribution to analyze bond portfolios can certainly justify the effort.
Multi-Asset Class Implementation and Support
The BI-SAM Professional Services & Production Support teams, backed by years of financial services and technology experience, partners with our clients to provide expertise and product knowledge across the BI-SAM suite of solutions. The team offers a wide range of services to best configure, integrate and support BI-SAM’s products for your business. Contact your BI-SAM account manager or visit https://www.bi-sam.com/implementation-customer-support to learn more.