|
Clients,
prospects and consultants are putting increased pressure
on investment firms to provide fixed income attribution
information that is clean, reliable and consistent. At the
same time, there is a trend in portfolio manager/analyst
compensation that is based on quantifiable "excess
return" contributions. This combination of strong external
and internal requirements has made fixed income attribution
systems a necessity at many firms.
Even though equity attribution systems are widely deployed
in most investment firms, these firms have been slow to
deploy fixed income attribution systems for many reasons,
including their theoretical complexity, their voracious
appetite for spotless data, and the lack of standard methodologies
for calculating fixed income attribution. To meet their
demands for fixed income attribution, firms have installed
new systems and others have built their own systems. Most
firms have created a web of spreadsheets.
Once again, it's the data…
Fixed income attribution systems need data - lots of data
- and are especially sensitive to errors. All positions
and transactions must be correct. Systems require detailed
and preferably constituent-level index data. Index prices
and portfolio prices must be comparable. Security descriptions
must be complete and absolutely correct for the internal
pricing models of many systems to yield useful numbers.
Any additional data provided by the user, such as the risk-free
curves for various countries, must also be correct. Inconsistent
pricing and security master data for the same holding is
common (refer to prior CutterEdge publications "Renewed
Focus on Fixed Income Technology" April 2004 and "Reference
Data" February 2003 at http://www.cutterassociates.com/.
Data management is further complicated because managers
need the flexibility to classify bonds and customize the
way the data and calculations are viewed based upon their
investment style and the intuition built from their risk
systems.
New instruments
To get precise attribution calculations, proper attribution
requires daily security and market level calculations and
pricing, the capability to handle derivatives and new and
complex security types, and the ability to include the notion
of "derivatives offsets".
No agreement on methodology
The problem for fixed income attribution is to provide an
explanation for a deviation from the benchmark return. Compounding
the problem for fixed income attribution is the lack of
consensus among vendors and practitioners on algorithms
and methodologies. For example, the duration or interest
rate adjustment varies among performance vendors and even
Treasuries produce different numbers on different systems.
Unfortunately, neither AIMR PPS nor GIPS have provided guidance
or standards for performance attribution and are not expected
to do so until after 2005.
Vendors are scrambling
To meet the demands of investment managers, vendors have
been rushing to enhance their current systems and build
new ones for fixed income attribution. It is interesting
to note that in November 2002, there were three systems
that provided fixed income attribution (duration, shifts
in yield curve, spread changes, credit rating, etc.) as
part of their integrated performance measurement and attribution
systems. Now, for its research on fixed income performance
attribution for The Technology Council, Cutter has identified
over 20 vendors.
Conclusion
There is no simple solution and firms will continue to invest
in attribution systems and enhanced data management capabilities.
Industry leading firms have deployed multiple fixed income
attribution systems combined with custom capabilities to
meet client requirements and internal management demands.
In addition, firms will continue to invest in fixed income
data management systems and data management processes to
insure quality data. It's not surprising that the cost of
staff for acquiring, scrubbing and maintaining data is often
the largest component of the on-going cost of a fixed income
attribution system.
|
