Clients continue to face an unprecedented environment of low bond yields in 2012. This environment has created challenges in terms of managing investor expectations and in reaching for yield, where possible. Our clients are also under pressure to find ways to operate more efficiently with existing resources. In this challenging environment we remain committed to helping our customers, as we have for the past 32 years, be successful by providing them with insight and automation in the management of fixed income portfolios.
In this issue of the BondEdge customer newsletter we highlight progress we are making and keys issues that we believe to be important to you, our clients. We continue to improve the BondEdge product and have summarized the new capabilities included in BondEdge version 3.4. We introduced a new Yield Curve and Rates data feed service earlier this year. This new service is powering new client facing fixed income portals providing current and historical information on market rates and spreads.
Last November we held BondEdge user training and certification programming in Boston which was well attended by both new and longstanding BondEdge end users. We will continue to offer this training and certification program twice a year in alternating locations. The next session will be offered in Newport Beach, CA on March 7th to 9th. This is an excellent opportunity for new users, and for experienced users looking to deepen their skills in specific areas, and a great way to network with fixed income professionals.
Throughout 2011 we have been focused on improving system performance, product stability, and scalability. One of the ways we are improving system performance is by enhancing the BondEdge software to take advantage of the processor technology more fully. We describe how we are doing this in our article on software parallelization and the associated system performance benefits. Our new BondEdge OnDemand service is another way that we are helping clients outsource both IT support and automation of the BondEdge system. We have many clients that have moved to BondEdge OnDemand to save IT cost, increase system performance, and simplify their internal support of the BondEdge system.
As always, we greatly appreciate your business and look forward to continuing to serve you.
Best Regards,
Keith Webster
Managing Director
BondEdge Solutions
In this Winter issue, we introduce you to Chris Pedersen, CFA, Senior Consultant. Mr. Pedersen helps clients to maximize their usage of BondEdge and assist in providing solutions for their portfolio analytic and management needs. Mr. Pedersen has over 15 years of experience in the financial services field of which 8 were in the fixed income industry.
|
Q. | When did Chris Pedersen join BondEdge Solutions? |
|---|---|
| A. | Chris joined BondEdge Solutions in August of 2006. |
| Q. | Where did Chris work prior to joining BondEdge Solutions? |
| A. | Chris worked at Check Capital assisting with the management of client’s fixed income portfolios. Previous to Check Capital he was with Charles Schwab & Co in trading and client services. |
| Q. | Where did Chris receive his education degree? |
| A. | He holds a BA from UCSB in Mathematics/Economics, an MA also from UCSB in Economics and holds the CFA designation. |
| Q. | What does Chris likes to do during his spare time? |
| A. | He enjoys drag-racing his ’63 Ford Falcon, surfing and enjoys watching his kids play soccer, ice-skate and playing t-ball. |
BondEdge Version 3.4 has been released and includes enhancements to overall system performance, multi-portfolio what-if analysis, portfolio importing, multi-factor term structure modeling, and user experience.
Multi-Portfolio “What-If” Blotter Matrix
BondEdge Version 3.4 includes a “What-If” Blotter Matrix that allows clients to more efficiently analyze the rebalancing of a group of portfolios in a single session. The blotter matrix displays all selected portfolios and currently held securities at the same time. Clients can then have portfolios displayed as either rows or columns, with the security listings on the other axis. For each security, clients can view a series of attributes including current/original par, % of portfolio, effective duration, and contribution to duration. Clients can then propose changes to holdings amounts or duration contributions and review the effects of such proposed changes on portfolio measures and allocations.
Multi-Portfolio (Policy) Reporting
Effective with Version 3.4, clients can create multi-portfolio (policy) reports which access pre-processed portfolio averages and distribution allocations. This enhancement significantly decreases the time needed to generate multi-portfolio summary and distribution reports with these pre-processed fields. Clients can specify that a range of portfolio averages and distribution allocations be stored during a portfolio import or re-price in order to then quickly retrieve this information when running multi-portfolio reports. Alternatively, clients can specify that portfolio averages and distributions are calculated and stored independently of the import and re-price process.
With this release, we have also added the capability to generate single portfolio reports from within the multi-portfolio report, allowing clients to more easily drill down into more portfolio detail as desired without needing to exit the multi-portfolio report. The selected reports will be added as new tabs to the main report tab control.
Portfolio Scans
The Portfolio Scan in BondEdge has been expanded to include portfolio average measures and distribution allocations. This new capability allows clients to search through a list of portfolios to determine which satisfy user-specified criteria for overall portfolio averages (e.g. effective duration greater than 5) and distributions (e.g. as cash allocation greater than 5%) and is available for portfolios which have been selected for the pre-processing option described above for multi-portfolio reports. The portfolios which are found in the scan can then be automatically transferred to the single or multi-portfolio What-If tool within BondEdge.
Expansion of Multi-Factor Term Structure Model
The BondEdge Multi-Factor Term Structure Model, added to BondEdge in Version 3.2 for RMBS and floating rate securities, is now available for fixed rate government and corporate bonds with embedded call and put options. This model is accessible for securities in both the BondEdge and local (client modeled) databases. Clients can specify whether they wish to use the multi-factor or single factor model within a new “Models” tab located in the Preferences window.
Reporting and Charting
Several new chart types have been added for portfolio and benchmark reporting in Version 3.4. These new chart types include: horizontal bar charts, 3D stacked bar charts, and security holdings scatter charts which include sector icons. Additional graph customization is also available, including x-axis label display control, color palette selection, and scatter chart zooming. We have also made user experience enhancements for charting such as the ability to change chart types on-the-fly, and to customize graph window field order.
We have also implemented additional report, print, and export formatting control in Version 3.4, which provides clients with more flexibility in formatting report books in Excel®. This is available via a new “Template for Excel” specification in the “Other Settings” tab within each report setup. This feature provides easier control over margin, header/footer, and sheet settings, and also supports the addition of logos to BondEdge reports.
System Navigation
A number of modifications to further enhance the system navigation in BondEdge are included with this release, such as:
Portfolio Import
Version 3.4 includes a Portfolio Import Summary pop-up window that appears at the conclusion of the import process. This window includes information such as the import start and finish time, the number of import files, portfolios, and security records processed, as well as information on any exceptions. We have also added an email notification capability which includes the data in the pop-up window. The email notification can be configured within the General Preference window in the Tools menu.
A summary import log file has also been added to Version 3.4. This comma-delimited log file is created once per month and includes a summary of portfolio import processes occurring during the current month. The log file is saved automatically in the LOGS folder in the BondEdge user directory.
Email Notifications
In addition to the email notifications described in the portfolio imports section above, Version 3.4 also includes email notifications related to the start and completion of other critical work flow processes. These activities include monthly, weekly, or daily BondEdge database updates, and automated report book production.
BondEdge Database
Effective with Version 3.4, the BondEdge database includes floating rate securities with call option features and/or amortizing principal payments. These bonds are sourced from the Extended Database via an internet connection (i.e. the same method that is used for pool specific mortgages).
With today’s interest rate uncertainty, financial institutions have a responsibility to monitor the yields associated with various fixed income asset classes and credit quality. While the rates of bonds with different maturities and industry sectors behave independently of each other, it is critical to have insight into the overall pattern of term structure movement. By providing yield curves and rates service, a new capability which provides end of day yield curves and mortgage rate levels on a daily basis, BondEdge gives firms the benefit of a curve construction methodology developed by the industry leader in fixed income portfolio analysis.
This new service includes US taxable and tax-exempt yield curves as well as fixed and adjustable mortgage rates. Included in the initial release are:
Each yield curve includes the 1, 2, 3, 5, 7, 10, 20 and 30 year points. The Treasury curves also includes a 6 month point. The treasury and corporate yield curves will also include the rates for the following historical points in time: 1 day, 1 week, 1 month, 6 months, 1 year, 1 year high and 1 year low.
BondEdge generates these curves through a proprietary curve fitting model which captures yield levels across the maturity spectrum. This methodology enables the generation of a smooth curve capable of manifesting a rich variety of shapes, while maintaining yield stability from day to day. Additionally, the municipal curves are constructed using option adjusted analytics of a diverse population of over 10,000 tax exempt municipal securities. The curves are delivered daily in XML file format via FTP.
“Fixed Income investors and analysts are increasingly seeking timely and robust yield curve and rate level data in order to make informed investment decisions. By leveraging BondEdge’s 30 years of fixed income expertise and robust, institutional quality analytics, we are able to provide a yield curve service that will become the standard in the industry.”, says Keith Webster Managing Director, BondEdge Solutions.
We are pleased to announce an enhanced interface between the Dynamic Asset Cash Flow functionality in BondEdge and the Asset Delphi series of PolySystems’ Delphi liability projection software via an application programming interface (API).
The movement towards principles based reserves and IFRS has led many life insurance companies to increase the number of interest rate scenarios used for projecting asset cash flows. For many years, firms that have used both PolySystems and BondEdge have relied upon a multi-step process to develop cash flow projections for their asset holdings and then have imported those cash flows, by scenario, into Asset Delphi. While the process has worked well, it typically requires coordination between the actuarial department and investment department, and may not be easily scalable for increasingly large interest rate scenario counts.
The newly released cash flow interface is available to joint clients via a capability within Asset Delphi called “Project BondEdge Assets”. The inputs into BondEdge are a file with assets from all portfolios and an interest rate scenario file. “Project BondEdge Assets” passes the scenario data and security identifiers to BondEdge, which then calculates and returns the projected cash flows and projected market values. Other measures, such as amortized values, are computed from the cash flow data. Finally, it creates the Imported Asset Information table, multiple Asset Portfolio Files, and a newly defined file that holds all assets across all scenarios.
For more information about this new capability which enhances the efficiency of your cash flow testing processes, please contact your client representative or email us at fia.info@interactivedata.com. For more information about PolySystems Asset Delphi, please contact Bob Keating at (312) 332-5670 or via email at bkeating@PolySystems.com.
Persistently low interest rates in the US have provided fixed income investors with multiple challenges. A few of these challenges include the difficulty in achieving a threshold of positive performance (in both absolute terms and versus inflation), low levels of income return, and meeting guarantees on annuities.
Exhibit 1 illustrates the trending towards lower interest rates by charting the yield to worst for the BondEdge proxy of the Barclays Capital Government Credit Index from 1995 to October 2011. Average yields for this benchmark fell below 4% during 2008, dropped below 3% towards the end of 2010, and have settled near 2% in 2011, with occasional recordings of average yield below 2%.
US fixed income investors have explored a number of ways to enhance the yield in their portfolios, including:
BondEdge provides a number of tools for investors and analysts to scan the fixed income security universe and many fixed income benchmarks to determine which securities or sub-investment grade categories have relatively high yield or spread within a specified credit rating, maturity bucket, or sector classification. These tools include scanning fixed income indices (including investment and non-investment grade benchmarks) and scanning the BondEdge security database.
Clients can create and save report templates which compute detailed portfolio versus benchmark (or just benchmark) analysis that displays the yield or spread of the benchmark by user-specified categories, such as sector, industry, country of issuance, or maturity buckets. Customized reports can be created for benchmarks by selecting the “issue level” detail box in the benchmark report tab (please see an example in Exhibit 2). An example of an analysis for a sample portfolio versus the Barclays Capital All Credit Index is included in Exhibit 3.
The Bond Scan facility can be used to search the BondEdge database to find securities which satisfy client specified criteria, such as rating, sector/industry, amount outstanding, maturity, and currency. Exhibit 4 displays a sample filter created in the Bond Scan which searches for bonds with the following criteria:
As of October 31, 2011, 29 securities satisfied this criteria (please see Exhibit 5). Any or all of these bonds can then be saved into a new or existing portfolio for further analysis. In addition, the identifiers of the bonds found can be transferred to the BondEdge “What-If” module via cut/paste.
If you have any questions on these BondEdge capabilities, please contact your consultant or the Client Services Group.
Exhibit 1
Exhibit 2
Exhibit 3
Exhibit 4
Exhibit 5
Gouthaman Balaraman Ph.D. and Chris Ruan Ph.D.
Computers with multi-core processors have become the de facto standard since 2005 when the first multi-core processors were released. Today, it is not uncommon to find desktop computers with 2, 4 or even 8 cores. However, many software applications sold today are built to be sequential (or serial) in nature, and cannot fully utilize the available computational resources. To take full advantage of multi-core processors, the application itself needs to be rewritten to execute in a parallel fashion. In this article, we outline an ongoing effort to parallelize the BondEdge portfolio management platform – the flagship product of BondEdge Solutions. Particular emphasis is being placed on improving the processing speed of complex analytical calculations while producing the same analytic measures.
In an effort to parallelize the analytics in BondEdge, we researched on a handful of parallel computing technologies. After rigorous testing and evaluations, we picked OpenMP, an open standard for parallel programming, as a suitable choice for BondEdge because of its ease of implementation, superior performance, and compatibility with our Microsoft Visual Studio development environment.
Parallelizing Monte Carlo
Fixed Rate Mortgages (FRM), Collateralized Mortgage Obligations (CMO), Asset Backed Securities (ABS) and Adjustable Rate Mortgages (ARM), owing to their inherent complexity, are some of the most computationally demanding securities to analyze. BondEdge employs a Monte Carlo method to calculate probabilistic future cash flows under various interest rate scenarios, which are then discounted and averaged to obtain dynamic risk measures such as option-adjusted spread, duration and convexity.
Typically, the Monte Carlo algorithm accounts for 70% to 90% of the computation time associated with the risk measure calculations for FRMs, CMOs, ABS, and ARMs. A parallel version of the Monte Carlo method has been developed and is released in BondEdge (version 3.32 and higher) to use the multi-core processors more efficiently.
Figure 1: An illustration of the Monte Carlo algorithm in a) serial computing and b) parallel computing approaches. In parallel computing, the processing of four paths is split between two cores equally.
As an example, suppose 32 interest rate paths are used within the Monte Carlo valuation process. The idea behind the parallel algorithm is to split the calculations among all available cores. In particular, the calculations associated with the 32 paths are split such that 16 paths will be executed on each core of a dual-core, or 8 paths will be executed on each core of a quad-core, and so on. A schematic representation comparing the serial and parallel Monte Carlo algorithms is shown in Figure 1. The parallel implementation is expected to reduce the processing time spent on the Monte Carlo routine by approximately half on a dual-core processor and by three-quarters on a quad-core processor. While the task of generating the probabilistic futures cashflows for 32 paths is split between multiple cores, the actual cashflow on each path at each cashflow date remains the same as the serial approach, so does the discounted present value, hence the valuated price of the bond.
Results
The computational efficiency of a parallel algorithm can be measured by computing the speedup, the ratio of time taken by the serial and parallel algorithms. In an ideal scenario, we would expect the speedup on a dual core machine to be a factor of two. This means that it would take the parallel algorithm half the time of the serial case. In practice, the observed speedup is often less than the ideal. This is due to the presence of non-parallelizable bottlenecks in the overall calculation, and a small amount of computational overhead in the parallelization process. As an analogy, suppose one is shopping for 100 items at a grocery store. The task can be accomplished in a shorter amount of time by splitting the task among two people each responsible for finding 50 items independently. While there may be some negligible overhead in coordinating the effort, significant time should be saved. However, having two people will not speedup the drive to the grocery store and back!
Figure 2: Observed speedup (compared to serial processing) of key rate duration parallel calculation using a multi-factor term structure model, for various portfolios on dual and quad-core machines.
The impact of parallelization can be seen from speedup reported in Figure 2 for key rate duration calculations employing a multi-factor term structure model. The observed speedup shows that the parallel implementation in BondEdge will cut the computation time for key rate duration calculations by close to half on dual-core and by about two-thirds on a quad-core. Apart from key rate duration calculations, we observed consistent performance gains for other BondEdge reports such as aged parallel simulations, and portfolio re-pricing as well. The observed speedup will vary depending on the number of cores available, and the type of report generated.
Moving Forward
Encouraged by the computational speedup from our initial effort, we have successfully parallelized the Monte Carlo calculations for both single-factor and multi-factor term structure models for MBS, CMO, ARM and ABS (in version 3.40) within BondEdge. We are also making a concerted effort to parallelize other computationally intensive algorithms by monitoring performance bottlenecks in various reports that are frequently used by our clients. We expect our clients to benefit significantly from faster BondEdge processing times.
After reviewing the historical fall out from the credit crisis by sector and the current government debt, the focus turned to stress testing securities, possible scenarios, and a review of alternative asset classes. Stress testing securities in each sector illustrated that there are interesting and new alternatives, however one must be aware of the details to correctly asses the risks. Stress testing was done based on a number of plausible interest rate and credit changing scenarios, which was eye opening to all attendees.
The presentation was modified specifically for depository institutions as we joined BBVA Compass Global Capital Markets in a series of presentations around the country. Dr. Bill Burns spoke in Atlanta, followed by Dr. Tristan Egualada, Senior Quantitative Analyst at BondEdge Solutions, who continued the tour by speaking in both Houston and Miami.
Since this topic is so relevant, the presentation will also be given via a Web Conference on Thursday, January 12th at 2pm eastern/11am pacific time. Please go to Stress Testing Securities in a Post QE2 Market for more information.
In November, BondEdge Solutions hosted its 2011 Fixed Income Portfolio Management Training and BondEdge Certification Program in Boston with participants from the asset management, insurance and consulting industries.
The 1 ½ day training and certification program began with a continental breakfast followed by training on topics that concern today’s fixed income investment management professional. At the end of the training, 63% of attendees completed our training course assessment, a pre-requisite to receiving a BondEdge Acknowledgement of Completion Certificate and participation in our 2011-2012 program.
Event participants gave overwhelmingly positive reviews. All agreeing that the training material met their requirements and 96% would recommend the training services to colleagues or industry contacts.
Sonia Dixon, EVP Analytic Support & Customer Operations and Dave Lampert, CFA, Vice President, Director of Consulting presented on managing and integrating the portfolio investment process with BondEdge.
Lou Gehring, Senior Vice President, BondEdge Product Manager, demonstrated the many ways that financial institutions can utilize BondEdge to help address real world fixed income challenges. He covered various portfolio management applications including: benchmark risk comparison in total return framework, liability-driven investing analysis, and fixed-income performance attribution methodologies and applications.
William Burns, Ph.D., Senior Vice President, Director of Quantitative Research covered topics on the importance of measuring fixed income risk and key differentiators in structured finance.
For those clients that missed us in Boston, BondEdge Solutions is repeating the training in Newport Beach, California on March 7-9, 2012. Clients interested in the program, please send us an email at be.certification@interactivedata.com.
How can I see the percentage held (%) for my AAA (Moody’s) quality bonds that are Treasurys and Agencies in my portfolio?
Clients can view the sector and quality breakdown of their portfolio in the “BE Sector /Quality Matrix” report. This can also be compared to a benchmark.
What’s the difference between Credit Trigger and Loss Modeling for a RMBS?
Credit Trigger and Loss Modeling for RMBS works independently from each other. Credit Trigger alters the payment allocation and cash flows enabled by a credit event while Loss Modeling causes realized losses for the collateral. With Credit Trigger turned on, the cash flow payments will be altered according to the prospectus in this event and for Loss Modeling, losses will be allocated accordingly to the tranches dependent on the prospectus where loss of certain principal payments will be viewed (usually seen on the junior/subordinate tranches).
Limitations
This document is provided for informational purposes only. The information contained in this document is subject to change without notice and does not constitute any form of warranty, representation, or undertaking. Nothing herein should in any way be deemed to alter the legal rights and obligations contained in agreements between BondEdge Solutions and its clients relating to any products or services described herein. Nothing herein is intended to constitute legal, tax or other professional advice and is not an offer of advisory services or investment advice. BondEdge Solutions makes no warranties whatsoever, either express or implied, as to merchantability, fitness for a particular purpose, or any other matter. Without limiting the foregoing, BondEdge Solutions makes no representation or warranty that any data or information supplied to or by it are complete or free from errors, omissions, or defects.
Interactive DataSM and the Interactive Data logo are either registered service marks or service marks of Interactive Data Corporation in the United States or other countries. BondEdge® is either a registered trademark or a trademark of Interactive Data Corporation in the United States or other countries. Pricing, evaluations and reference data are provided in the U.S. through Interactive Data Pricing and Reference Data, Inc. and internationally through Interactive Data (Europe) Ltd. and Interactive Data (Australia) Pty Ltd. Other products, services, or company names mentioned herein are the property of, and may be the service mark or trademark of, their respective owners.
BondEdge Solutions, an Interactive Data comapny
