It's Time for a Better Data Model
The
Limitations of Existing Data Standards
The Exposure Data Module (EDM) and Results Data Module (RDM) have been widely used for data exchange across the insurance marketplace for over 20 years. These data specifications were developed by Risk Management Solutions (RMS) to support RMS catastrophe models, and they are well-documented, allowing users to dig into the databases with their own queries and extract data. Other similar data formats from other providers exist, and these are also used for data exchange and analysis (though to a lesser extent), but all existing formats share key shortcomings that are blockers to improved efficiency, automation, better models, and the analysis of new types of risk.
The existing data standards have these key limitations:
-
Rigid—The EDM and other formats were built to support specific model approaches and algorithms. An effective industry standard should have the flexibility to accommodate any modeling approach. Current and future models must be able to handle larger data sets, unique and complex contract forms, and the underlying technology that make new modeling approaches possible.
- Loss of Information—Today’s commonly used data exchange formats do not capture all of the information required to show a counter party, or even another user in the same company, the complete, unambiguous picture of how exposure and coverage interact, and how models were used to produce analysis results. This lack of completeness forces skilled analysts to do forensic analysis to understand a submission or past analysis. A modern standard should contain all of the information required to understand an analysis, from exposure and coverage through results.
- Property-focused—Current data formats were created to support property risks. Over time, models have expanded beyond property, but the data formats have either constrained modeling options, or completely new formats have been required to support the new models. A new data model must have the flexibility to cover any line of business as well as emerging or unknown risks that can spring from an evolving economy and present new demand and opportunities in risk financing.
Design Goals of the Risk Data Object
The
- Represent exposure and coverage clearly, with the flexibility to handle any new or complex coverage that is needed in the real world, as well as the coverage interaction and business structure that has generated specific financial perspectives.
- Express exposures in a way that allows new lines of business or types of risk to be included in the schema.
- Provide a complete, repeatable, and auditable data set that provides all of the information needed to understand how exposure was analyzed, including the models, settings, and reference data that were used to produce results. This includes changes to model data that comprise a bespoke view of risk.
- Use a modular design that simplifies modifications in the future.
RDOS Includes Complete Information
The EDM contains exposure and contract terms. The RDM contains modeling results with limited information about how models are used to produce the results. The RDM contains only limited information about the links between the results and the exposure (EDM) that produced the results. Typical client environments have hundreds of unlinked EDMs and RDMs.
In contrast, the RDOS contains both exposure and results and adds more complete information about model settings, links between exposure and results, and other
RDOS Facilitates Data Exchange
RDOS is a unified data model that consolidates exposure, contract coverage,
Exchange of RDOS is easy, efficient, and flexible. The default exchange format of RDOS includes the schema definitions along with the data so that the schema is available regardless of choice of data processing framework or programming language.
External systems can consume RDOS directly or use transformations to consume RDOS in other formats. When RDOS is received by a consuming system, the system can determine the schema and hierarchy without any additional external information. This facilitates easy exchange for any type of risk extended by different parties. During exchange, you can map a counter-party's entity definitions to your own, or you can choose to accept only the attributes you have in common.
RDOS Provides Extensibility
Insurers and reinsurers increasingly seek to apply enhanced pricing and exposure analytics to lines of business beyond property catastrophe. The EDM and RDM data models are rigid and difficult to extend to new classes of risk and associated coverage.
RDOS schema extensibility empowers open modeling and the ability to customize a personal view of risk. The RiskItem entity, which already includes the sub-type RealProperty, can be extended to new sub-types that represent new lines of business, like Aviation. Existing attributes, like RealProperty attributes, can be extended to capture new ones.
RDOS Defines Unified Exposure Entities
In the EDM, property characteristics and insurance/reinsurance structures are spread across multiple tables, sometimes redundantly. Many EDM entities combine both the description of risk and the financial/coverage terms. This means that you have to structure joins across multiple tables to derive data insights.
In contrast, the RDOS is modeled with the following well-defined entities to address specific domain concepts.
Contract—The central component of the data model is the contract, which is a binding agreement between the insurance provider and insured customer.
Risk—Every contract has a single subject, which for primary insurance can be a single risk or a schedule. A risk is a collection of primary exposures, and a schedule is a collection of risks.
Risk Item—A risk item represents something that can be damaged by events. Risk items can be of multiple types, such as real property, contained property, time element, and so on. Each of the sub-types is a separate risk item. The level of risk associated with risk items is the risk exposure, which indicates to what extent the risk items are covered by the insuring party for different coverage types.
Risk Exposure—The Risk Exposure entity provides the link between a risk, risk item, and its member exposures, such as building, content, or business interruption exposure. It stores the insurable interest, which refers to the ownership share. For some lines of business, like offshore platform, it could be less than 100% because these can have shared ownership. Insurable interest scales the valuation before the ground-up loss takes effect.
Portfolio—A grouping of contracts based on some arbitrary criteria is a portfolio. The portfolio may be based on geographical region, business unit, or some other criteria that makes sense within the organization. A contract can be a member of multiple portfolios.
Structure—A structure is a container for financial positions and treaties used to define business hierarchies and contract interactions.
See Exposure Schema for more information.
Implement RDOS to Suit your Organization
Risk Data Object is a logical data model, meaning that it provides schema specifications and business rules to standardize the data model independent of a specific database technology. The database technology is your choice depending on your own access patterns, performance requirements, and scaling needs. Select from the following technology options:
- Relational—Implement the same RDOS specification as a traditional relational database using SQL, Postgres, MySql, Access, and others.
- Interchange format—Keep the RDOS specification in simple, light-weight interchange format using JSON or CSV.
To help you adopt the Risk Data Object, the Steering Committee is providing scripts that implement the schema as SQL tables to simplify relational querying and user understanding.
RDOS Provides an Unambiguous Picture of Financial Perspectives
Contract Definition Language (CDL) enables the expression and evaluation of many types of insurance and reinsurance contracts, but there are two important aspects of coverage that contract terms alone cannot describe:
- Forming a precise subject of a treaty, e.g., the all-perils treaty benefits from the Florida Hurricane Cat Fund
- Specifying and evaluating financial positions, e.g., Gross, Ceded, and Net-of-Retro for each treaty
Structure Definition Language (SDL) enables industry professionals to solve these problems quickly, accurately, and transparently. A Structure written in SDL models the flow of quantities such as risk, loss, premium, and expenses as they pass through CDL contracts and other transformations.
Read next: RDOS Concepts