Financial institutions and insurance companies serve clients first and foremost, but they have also developed methods for transferring financial assets to investors through structured financial products. As the recent economic crisis has shown, however, a huge gap still exists in the risk-management and valuation aspects of these capital markets, especially in the complex financial-products business.

To survive, institutions need to better evaluate their risk and understand how that risk stacks up against their bottom line. This can be accomplished by utilizing a data warehouse environment that supports a flexible, granular assessment of complex financial products along with an integrated approach to storing and analyzing data.

Such storage and analysis environments can be easily realized with the powerful capabilities of enterprise data warehouses (EDWs) and business intelligence (BI) software. Yet few financial institutions have implemented these concepts and tools, as achieving this state may require more than just technical changes. For companies to succeed, they must embrace data management as an organizational discipline as much as a technical one.

Strengthen the foundation
Financial and insurance industries exist to manage risk. It's a mandate that hasn't always been successful. Over the past year, established organizations have suffered unanticipated and crippling losses from peripheral business units, and the ratings on various financial products have dropped like stones. The deterioration began some time ago, but many financial institutions were unable to detect the extent of their exposure with the available data and tools.

The financial environment is filled with complex assets like derivatives, which are aggregations of other financial contracts based on the value of other financial assets, such as interest from mortgage loans. Their purpose is to minimize risk exposure through diversification, as with a collateralized debt obligation—a common type of derivative that consists of groups of mortgage-backed securities that are based on groups of tens of thousands of mortgages.

Tracking this product's risk means not only following the collateralized debt obligation but also monitoring the various mortgage-backed securities within these groups. Gone are the days when an institution could loan $1 billion and be confident that the possible downside was $1 billion. Now, it could be exposed to a loss of $1 billion or $100 billion. The question is, how does it know?

The analysis necessary to understand the complexity of these structured products and adequately predict losses requires a tremendous richness in the data environment, as well as an infrastructure that can deliver that data up to date and in real time. This type of infrastructure is available through the capabilities of an EDW. Superior data management calls for a willingness to focus and invest resources in this critical area of risk management.

Tools at work
The financial and insurance industries assess risk using multiple methods, including analytical modeling and mathematical simulations, as well as credit indexes, stress testing and ratings-agency evaluations. (See figure 1.) To be effective and robust, the models must be based on input parameters that are broadly supported by data. In other words, a model is only as good as the quality of the data on which it is based.

Quality data needs to be traceable, valid, secure and properly formatted. The data should also be multidimensional in order to make sense of complex interconnections. Effectively monitoring a collateralized debt obligation requires following not just the groups of derivatives and the mortgage-backed securities upon which they're based but the underlying mortgages themselves, as well as geographical trends that might affect them. (See figure 2, below.) Data such as interest-rate swaps, credit enhancements or linkages between securities—whether two collateralized debt obligations share a manager, and the attendant risk, for instance—can provide crucial insights. Essentially, the data must capture all exposures across the organization, down to details like customized deals held on a trader's spreadsheet.

Consistency is another crucial characteristic of quality data. Standards must be followed to ensure names are listed the same way, for instance, or account numbers always start with zeros. Above all, to allow modeling, ensure diversification and avoid dependencies, the data must be highly granular, listing everything from loan-to-value ratio to the outstanding balance on the individual mortgages.

Of course, data alone, even if very high quality, is not sufficient. The repository and associated analysis software need to be sophisticated enough to allow the data to be used for effective risk assessment. Integrating the data enables the system to link information about insurer-wrapped groups of securities to information about the insurers themselves.

In these cases, details like current ratings information should be easily accessible. The format should permit users to analyze and view data as desired, filtering by geography, credit rating and more. Such flexibility is crucial in the "test and learn" environment that is essential for these instruments. To that end, the data repository should also be extensible, accommodate new data types easily and be updated as frequently as required—this might be weekly for a regional consumer bank or in a trickle feed for a multinational organization.

Identify a roadmap
Some financial institutions that have so far avoided these risk exposures and weathered the crisis reasonably well possess these accessible and extensible data repositories and analytic tools. For companies that are without these systems and functionality, achieving this kind of data environment requires change.

An institution can start by assessing its current data environment, paying close attention not only to software and hardware architecture, data quality and data management but also to aspects like user access tools; data availability, recoverability and protection; business governance procedures as they encourage management support; and, especially, common types of workload queries. An environment that is responsive to critical activities of modeling, reporting, stress testing and so forth requires the creation of a process for handling the analytic results, the ability to interact with external systems, established service level agreements and proper management.

Next, the organization should determine where it wants to be. It can accomplish this by designing an ideal data environment and identifying the changes needed, whether that is replacing a series of data marts with a data warehouse or re-architecting an existing data warehouse. (See figure 3, below.) The proposed new system should be tested using real data and data types to ensure robustness and effectiveness. With the goal defined, a roadmap can be developed and implemented.

Collaborative effort
The key to successful roadmapping and implementation—and the heart of the challenge—is acquiring buy-in and cooperation across the entire organization, from upper management to IT staff.

Integrating data into one depository will expose the differences in how various departments think about information. For example, if a financial analyst is unsure whether a loan balance used to calculate exposure is the net of syndications or hedges, then he or she might not be able to make a clear determination of the residual risk to the corporation. Any differences should be settled before the system can be considered trustworthy. It's a metadata problem more than a data problem, but it should be taken into account.

A first step in moving toward integration is to define how to format and store the data. If the data is formatted inconsistently, it is much more difficult for the appropriate departments to receive the data they need; consequently, the team that assesses risk management will waste time sifting through and cleansing raw data. Data needs to be properly entered, stored and managed to be optimally accessible and useful.

How to effect this change depends on the organization. Strong leadership from management is key to encouraging cooperative participation from the business and technology sides and in establishing an organization-wide focus on creating the kind of data environment needed. With proper prioritization and allocation of resources, a solution exists, and the payoff is better risk management—vitally important in these uncertain economic times.

Stabilize risk
In an increasingly complex global financial environment, managing risk exposure means developing the right models and the right data repositories to support them. Institutions benefit immensely when they define a quality data environment as the goal, provide the resources and, most of all, focus all levels of the organization on working together. With this kind of approach to data, risk can be maintained at acceptable levels. T

Dilip Krishna leads Teradata's Enterprise Risk Management and Capital Markets practice in North America.

Dr. Robert Mark is CEO of Black Diamond, which provides corporate governance, risk management and transaction services.

Photography by Corbis

Teradata Magazine-December 2008