The makeup of ERM: extreme computing and big data are the DNA of enterprise risk management systems.

Enterprise risk management systems are testing traditional risk management models, as well as the sensibilities of the insurance industry. A new paradigm of risk management and performance measurement has challenged the business consciousness of the financial services industry with the growing regulatory importance of COSO (Committee of Sponsoring Organizations of the Treadway Commission), Basel II, Solvency H developed by the European CEA (Comite Europeen des Assurances), Sarbanes-Oxley and credit agency standing.

Risk-aware business cultures are establishing risk programs, measurement and management of risks across the enterprise landscape. From the board room to all responsible employees, the ERM framework includes policies, methodologies and infrastructure. ERM crosses functions at the corporate level, business unit management level, frontline service levels and across all business operations.

According to Dr. Bob Mark of Black Diamond consulting, it is important to have great risk-related policies and methodologies, but an insurer is doomed to failure without the necessary data management infrastructure to implement them.

Within the ERM framework, a number of risk types have been broadly identified and defined, and examples include market, credit, operational, business, reputation and strategic risk. Each of the risk types may be analyzed as a single instance, but the analysis is only meaningful if placed within a much broader context that considers how the various types of risk interrelate. For example, interest rate analysis is interesting if associated with the complexities of trading in debt instruments or foreign currencies. To achieve an understanding of risk that can be translated into action, there's a fundamental reliance on data from across all of these risk types and dimensions of business.

Risks are also managed with increased sophistication in valuation, pricing, financing and control. Statistical models are used to determine loss probability based on observation and experience. Risk identification and measurement of probable loss are cornerstones of insurance and actuarial science. Blended with financial models for capital valuation and allocation, business decisions are being made based on the cost of needed capital, or economic capital, and the risk adjusted rate of return (the future value of discounted cash flows and the cost of money). It's conceivable that every financial decision made on behalf of the business will be based on risk-adjusted return models delivered on the desktop. What does this mean? If you make strategic business decisions--create a new product, manage a portfolio, underwrite a new piece of business or upgrade to a new technology--each one of those decisions will be based on its inherent risk and the probability of loss or ruin and the rate of return it will generate adjusted for that risk.

The application systems used for risk analysis, valuation and measurement push technology to the extreme limits of computing. For example, a single large Monte Carlo simulation or "what if" scenario can contain 40,000 variables, 20 time periods and 50,000 paths and crunch through 300 gigabytes of numbers--the equivalent of processing five terabytes of data within a matter of minutes. Running scenarios for catastrophic losses, such as a pandemic, could include a number of variables such as geography, level of penetration into a specific geography, type of exposure and duration of illness. The new super computers with distributed massively parallel processors are required to perform these types of tasks. Today, the power of computer processing makes complex modeling, simulations and queries possible. Big data volumes, mixed workloads that include simple and complex queries, and multiple concurrent users are all integral parts of the computer processing power.

The data that will drive both computational needs and decision support requirements are formidable. Statistical files that have been the traditional means of actuarial analysis are no longer adequate and are unable to support an enterprise requirement to strategically manage risk. To successfully respond to the demands placed on the insurance industry, it is now necessary to deploy a single, shared data repository with a single view of the business, which provides transparency, consistency and auditability. The enterprise data warehouse is the optimum platform for the task.

Extreme computing and big data are the DNA of the ERM system. The enterprise that does it right will have a better chance of minimizing risk, maximizing returns, and funding the balance sheet, and in the process will create a persuasive competitive advantage for a very long time to come.

Contributors: Dale Strobel is senior consultant of finance, risk and performance management solution, global consulting services, and Dilip Krishna is director of enterprise risk management center of expertise, both of Teradata, a division of NCR Corp. They can be reached at [e-mail].