As an investment manager with over $533 billion under management, Merrill Lynch Investment Managers (MLIM) has a varied and complex portfolio of investment products and of course an even more varied and complex strategy for managing its risk. Recently the firm decided to take risk management to the next level and is currently in the process of implementing Barra's TotalRisk System for enterprise-wide risk management.
While MLIM had previously been using stand-alone risk models also provided by Barra, the firm decided it needed a single system to bring together risk across portfolios and asset classes for a more efficient and comprehensive view. More specifically, MLIM was using components of the Barra Aegis system for equities and derivatives in addition to components of the Barra Cosmos system for fixed-income risk management before deciding to bring all of its risk management together.
"These were really desktop versions that were used to analyze one portfolio at a time," explains Brian Fullerton, global head of risk at MLIM. "By going to TotalRisk, it allows us to bring all of our portfolios together and do them all at once, as now we can take into account portfolios that would cross each one of the individual Barra models." He adds that if a manager were to balance a portfolio with both bonds and stocks in the previous risk management environment, "it would have to be done with two different pieces," or run twice using different Barra models. Although he would not name them, Fullerton says that MLIM had "looked at a lot of other vendors," mainly when the firm first settled on the Barra stand-alone models. However, some general competitors in this space include Askari, Vestek, Wilshire Associates and Algorithmics.
MLIM is in the process of implementing the TotalRisk system, which will also be integrated with its portfolio management system. So far, the firm has finished the implementation across its U.S. equity products group, has begun deploying the system in the fixed-income area and plans to finish the project across the firm by the end of this year. "We're using the TotalRisk product to generate a report on a daily basis for most of our U.S. equity products right now and we'll just keep on rolling it out," says Fullerton.
"We're delivering this as part of the desktop we're building for our portfolio managers so that it's incorporated into their information," says Fullerton. "We want risk management to be part of how they manage their portfolios, not done afterwards." The MLIM portfolio management system is "more of a collection of systems," he adds. With the addition of TotalRisk, portfolio managers will be able to view and use a collection of tools and reports. While MLIM is first rolling out the system to its U.S. locations desk by desk and portfolio management team after team, in the future Fullerton says, "we'll test that and figure out how we're going to roll it our further to our other locations."
Fullerton explains that one key reason MLIM chose the Barra system for enterprise-wide risk management was to ultimately make portfolio managers more efficient. "We're quite a big asset manager so we need to run through all of our portfolios and it's more effective to relieve the portfolio managers of the responsibility of generating their own numbers and just provide the risk analysis to them," says Fullerton. "By centralizing it we are saving resources and becoming more effective."
In addition, MLIM needed a risk system that was catered towards the buy side and needed a good deal of data to support the system. "The asset management side is a little different in that we're really looking for a product that has a lot of the data already in it," says Fullerton. "And that's what Barra gives us. It has the time series analysis for the volatility of individual stocks or bonds and how they correlate with other holdings."
He adds that in general many risk management systems are geared more toward the sell side in terms of the data provided. "For the broker-dealer side, they have a little bit of a different problem," says Fullerton. "For example, they might have a derivatives desk writing an option and they want to make sure that they price that correctly, so their focus will often be on pricing algorithms." On the other hand, buy-side firms such as MLIM have less of a pricing focus because they get independent securities pricings from custodians. "Because we're not in the business of selling securities to people and we have a longer term horizon and that sort of thing," says Fullerton.
A Closer Look at the TotalRisk Technology
The technology behind Barra's TotalRisk System is based around the consolidation of information from multiple, disparate data sources. The product includes data integration tools and facilities for mapping relational data, text files or real-time data into its object-oriented data model. This aggregation enables the "enterprise-wide" risk view across portfolios, asset classes, countries and currencies. Aamir Sheikh, executive vice president at Barra, explains that aggregation is key for the buy side, "There are individuals and groups that specialize and have a talent for their specialization and need to make reasonably autonomous decisions. But you need to aggregate those decisions and look at the impact on an overall portfolio. So there's aggregation across many managers and there's aggregation across many asset classes."
TotalRisk's complex analytics are run across clusters of computers through parallel-processing technology and underlying risk engines. Limits can be set against multiple risk measures and exposures such as VaR, tracking error and contribution to risk, and can be specified and monitored at all levels of the firm. The TotalRisk system includes an integrated database covering publicly traded securities along with a facility for upgrading its characteristics in real-time. It covers over 35,000 equities across 55 countries, as well as over 1.1 million fixed-income securities in 25 markets. The Barra historical data covers its 25-year history for risk computations and performance attribution. Data also includes mortgage-backed securities, CMOs, asset-backed securities, a set of exchange-traded derivatives and data on a number of popular benchmarks.
"TotalRisk is a flexible tool that allows users to aggregate risk across multiple managers, multiple portfolios and multiple asset classes and then to decompose that risk along many dimensions such as industry, style, managers and countries," says Sheikh. Analytics and models available through the TotalRisk system include multiple-factor risk models covering a range of asset classes and security level data including terms and conditions for fixed-income securities. Also available: fundamental data for equities, pricing models calibrated to the market for VaR, stress testing, integration of proprietary models or other third-party models and optimization capability to provide risk/return profiles incorporating user-defined constraints.
Tech Requirements for Barra TotalRisk
The following are minimum system requirements for Barra TotalRisk: Database Server (256 MB/10 GB RAM + 450 MHz Pentium III processor), Application Server (256 MB/1 GB RAM + 450 MHz Pentium III processor and Client (128 MB/1 GB RAM + 400 MHz Pentium II processor). For more information on Barra and the TotalRisk System log onto www.barra.com.