Dynamic Data - the next step for risk management systems.

As risk management extends across the enterprise and moves from batch to real-time processes, a sound information management infrastructure becomes ever more critical. Alex Tsigutkin of Axiom Software Laboratories explains how to design a data architecture that meets the demands of modern risk management.

"The difference between the right word and the almost right word,” Mark Twain once mused, “is the difference between lightning and the lightning bug.” The same can be said for the difference between an “almost right” static risk management system and a dynamic system that reacts with lightning-quick rapidity to market changes. In today’s so-called efficient markets, billions can be lost in the blink of an eye. Onsets of market volatility, pressing regulatory concerns and even changes in political or economic climate must all be instantly accounted for and anticipated.

Today’s trader, risk manager or analyst is also hard-put to oversee the constantly expanding world of new financial instruments in a global market. Consequently, risk management technology is a burgeoning area of interest for the finance professional. In this article, we’ll both illustrate the advantages of the dynamic system over traditional static systems and elaborate on how different components of proposed a risk management system should deal with various aspects of data management and risk calculation.

Until quite recently, institutions and vendors relied mainly on traditional approaches when attempting to build their risk management systems. This traditional approach was necessarily static, since it is based on hard-coding system processes and data. Institutions often hired expensive consultants to add to the ranks of dedicated in-house programmers and financial engineers. Thus fortified, technology teams worked in concert to build huge systems following detailed specifications provided by the management. However, what was sufficient for accounting, regulatory and even trading systems performs significantly less successfully in the rapid-fire milieu of risk management.

The problem with this traditional static approach is that it prevents risk management systems from being easily re configurable. They are often unable to accommodate the business or technological changes that occur throughout the business cycle. A static risk management system is designed to perform a particular specified task with a particular specified set of information. Moreover, the system will handle information based on a predefined data structure, a selected risk methodology, a specified combination of instruments and pricing models, and predetermined parameters for analyzing the results of calculations.

System flexibility depends largely on the path taken in the early stages of the firm’s risk management software development process, at the stage of selecting a data management model. Unfortunately, many existent risk management systems have evolved from older front office trading systems, which were originally designed to handle trade entry or transaction processing pegged to a limited number of instruments. Others have originated from a theoretical model without a sound data management backbone. Altering these systems in order to accommodate additional instruments, or even similar instrument types originating from new sources, is a difficult and expensive task.

This situation has led many financial institutions today to develop dynamic and scalable user-controlled risk management technologies. The better a firm is able to control and change its risk technology infrastructure, the better able it is to improve its internal risk analysis and trading strategy.

Flexibility in allowing the user to introduce new instruments and corresponding pricing models into system inventory is one of the most important functions of any risk management system. Yet this commonplace task demands a great effort to be expended on developing and incorporating new information within a static system infrastructure.

For example, a portfolio might initially consist of interest rate swaps, loans, and foreign exchange options. The introduction of a new group of instruments, such as credit derivatives, would have to be dealt with in the front office on the transactional level first, then gradually worked into middle office decision support, where appropriate analytics would have to be coded in and applied in order for the system to reflect the impact of the added instruments on the overall portfolio.

This is time consuming, to say the least. For firms that deal with the introduction of new instruments on almost a daily basis, the ability to integrate new data without losing system accuracy or efficiency becomes a pressing issue.

Another deficiency of static systems is in the way input data is stored and made available for analysis. There are often limits on the amount of transactional data that can be cached and funneled into analytical applications. Many systems, for example, carry only currently outstanding transactional data, making it difficult to effectively perform investigative analyses such as historical trends and stress testing. In addition, limits on the number of attributes (fields) of transactional information may have adverse effects. Drill down analysis often require additional data elements to shed light on risk concentration effects or unusual instrument behavior.

In contrast to the descriptive term static, the definition of dynamic, when applied to a risk management system, denotes a system which is easily configurable and capable of reflecting changes made in a firm’s overall and specific risk and trading strategy. System flexibility also serves a key role in portfolio optimization and capital allocation.

The competitive advantages delivered by a dynamic system stem most significantly from the use of multi-tier architecture. This ensures that different components of the system are able to sustain environment-imposed changes and that the data management subsystem is scaleable, flexible and adaptable to these changes without requiring lengthy alterations to overall system infrastructure.

In enterprise risk management specifically, back and front office systems house complex data models comprising multiple tables, all of which may need to be taken into account during risk assessment. These tables retain contract information, transactional data, settlement information, mark-to-market values and sensitivities, securities master agreements, counter party information, and so on.

If an institution has middle office interfaces in place and changes need to be made (because of development shake-ups in front or back office systems, or mergers and acquisitions processes, or whatever), IT managers find that it is arduous and time-consuming to rebuild hard-coded interfaces to each of these systems.

In our experience of building risk management systems, we have learned that each institution has its own unique business and technical environment. This makes it is very important to take into consideration data logistics, terminology and data elements specific to the institution. This philosophy served as a key guideline for creation of infrastructure products. A dynamic infrastructure can be tailored on the fly to be compatible with the company’s operations policy, making it easier for users and IT to adapt to the new decision support system environment. This can be realized by paying greater attention to data and process administration, as well as focusing on solutions that not only integrate multiple data sources, but also provide full data management and administration functionality.

To achieve such system and business transparency, you need the capability to register existing data models as input source data for the risk management system, as well as ability to extend such data models dynamically during the analytical process. Another important guideline is to keep the data component and the processes uncoupled.

At bottom, the dynamic approach to system design is really based on optimizing the ability of different components of the risk management infrastructure to operate independently of the others, freeing system components from rigid parts of a monolithic structure. In order to keep pace with the markets, institutions continually seek out new ways to upgrade their risk management infrastructure.

Theoretically, if there were common standards covering every single aspect of risk management, it would be simple to build, implement, and modify risk management systems. In reality, financial markets are in a constant state of flux. New instruments, models, and risk methodologies come to the fore and there are always new systems to interface with.

Let’s take a look at the advantage a dynamic system brings to the process of incorporating new entries into company’s roster of financial instruments. Credit derivatives, while still relatively new, have been gaining appeal. In fact, the active use of credit derivatives has risen incrementally over the past few years. Their structure is highly dependent on the ability of the user to identify different types of underlying assets already existent in an institution, and to provide the means for the user to trade on these assets.

Structured debt products, another type of instrument, are also dependent on a multitude of factors that are all equally important: every one of the factors must be taken into account. In addition, the ability to define and maintain various internal and external hierarchies of the portfolio and relationships among the distinct data entities such as counter party information, organizational structure, regulatory reports etc, is a critical ingredient in the decision process. It is best maintained by the business users directly.

One way of equipping business users with such capabilities is to use a collection of logical conditions, known as business rules, that act as a management interface between the analytical and data management components of the system. These rules give users control over the enormous amount of information and allow them to formulate and implement business management strategies in a transparent and direct manner.

Users no longer have to wait for software upgrades, but have the flexibility to add new functionality as the need arises. The rules let users who may lack programming language proficiency apply common business terms and their insight. The system automatically translates these into complex business content messages. Messaging technology allows these rules to be available in a real-time Web-enabled environment, distributed to all decision support service areas of a modern enterprise.

A dynamic system is an invaluable resource during a merger or acquisition. In this environment, participating entities must simultaneously undergo system reconstruction due to the merging of diversified business units. The system must be able to handle the consolidation of multiple information sources, distinct organizational developments and changes in data structure. Dynamic systems are very flexible in this respect and easy to reconfigure. They can practically self-adjust to the new environment.

Let’s touch upon the different types of input data source types. Data is, after all, the lifeblood of a risk management system. It includes transaction information from front and back office sources, market data, and control (or legacy) data.

Market data is used primarily for the pricing of instruments and for providing key statistical components related to portfolio risk analysis – e.g. volatilities and correlations that are either calculated by the system or formatted and delivered by commercial data providers.

It must be kept in mind that market data is often derived from internal sources. The most specific example of this is information regarding certain energy markets: the data is not delivered by the market exchanges, but is derived from internal trading operations. The system has to be able to accommodate on-demand time series data flowing from internal sources, just as readily as it assimilates information provided by commercial vendors.

The quality of risk calculation is highly dependent on the quality of market data received. Reliable sources of market data are useful caches of historical information; they also serve an inimitable function as real-time “litmus tests”, able to help pinpoint the state of global financial markets at any point in time.

A truly dynamic system will accommodate multiple sources of market data, since it is entirely possible to derive slightly different variance/covariance figures from different market data sources. Any inconsistency in market data will cause a system segue right into erroneous risk estimates. The ability of the dynamic system to switch from one market data source to another, without losing data integrity, prevents this.

Every institution has its own way of looking across proprietary records in different parts of the company. Different branch offices, for instance, may not use the same types of systems for tracking customer information, general ledger information, information on the corporate structure, or specific business-related information. These data types are all brought in under the umbrella term “legacy” or reference data.

Some refer to this package of information as “static data”, which is a bit of a misnomer. Although this information, typically, does not change as rapidly as market data or transaction-related data, it is in fact constantly evolving, as new information is added about new clients, business units, security masters or changes in corporate structure and policy. The dynamic system allows seamless accommodation of the latest legacy data through the ability to adopt already existing operational processes, and where necessary to enhance these process through the use of business rules.

One caveat – the use of the most cutting-edge system building blocks in system design does not guarantee that the result will be a dynamic system. It is possible to build a static system with the same tools used to create a dynamic system. Multi-tier architecture, for example, is the system cornerstone in the initial configuration of a dynamic system, but some static systems can also be said to be multi-tier.

However, there is a difference. Usually, in the latter case, the term “multi-tier structure” simply describes a particular structural choice that allows a firm to manage hardware or software resources properly. But the system is not dynamic from a business standpoint. A system might be user-driven and object-oriented, but the result is still a static system, if the system design does not deal efficiently with variations and alterations in data sources and management. Nor will the system function dynamically if it is unable to handle changes in portfolio composition, or if the problem of introducing data stemming from new kinds of risk exposure is not addressed.

The dynamic approach to system development answers the challenges of the business and technology requirements of today’s and future’s firm wide risk management. The dynamic system provides the middle office with the solution for optimal measure, control, and consolidation of all types of risk and answers to the unique needs of the enterprise’s risk strategy, its risk philosophy, and the specific risks inherent within each business unit. It quickly adapts to the existing technology infrastructure and provides for a continued business and technological advancement and extensibility.

Reprinted with the permission from Risk Professional • September 2000