Enhanced view of risk through data validation and enrichment

How Moody’s insurance Analytical Services conducted an extensive data enrichment process for a leading global managing general agent (MGA)

Key takeaways

• Significant improvement in exposure data quality via manual and automated augmentation
• More accurate portfolio loss estimates across all perils
• More effective portfolio and capacity management
• Better claims preparedness
   

The challenge — poor data quality and completeness

A leading global MGA underwriter specializing in an extensive range of specialty insurance products appointed Moody’s insurance services to provide a series of data services across the company’s property and casualty portfolio, including portfolio roll-up and reporting.

While working on client data, the Moody’s Analytical Services team ran a stringent validation algorithm comprising over 800 checks for data quality anomalies. During this check for approximately 500,000 properties, the team noted poor geocoding due to missing address attributes as well as limited and incorrect information on critical building attributes, such as construction type, year built, occupancy, and number of stories.

This shortfall in data quality and completeness reduced the accuracy of modeled results and heighted uncertainty of loss estimates, impacting portfolio management and capacity allocation decisions.
 

The solution — data enrichment services

The Moody’s Analytical Services team flagged these issues with the MGA and conducted a data enrichment process across the portfolio to illustrate the impact on model outcome.

Data quality enhancement is one of the Analytical Services team’s core offerings, and the team’s combination of flexibility, expertise, technology, and process efficiency allows for a quick turnaround time, even when enhancing large volumes of location data. The team applied multiple engineering heuristic validations and augmentation processes to the dataset, including both manual and automated procedures, to establish geocode accuracy, remove invalid information for each location, and introduce missing building attributes.

By combining manual and automated processes, the team achieved extensive data enhancements, including:

• Manual data augmentation measures:
  • 95%-100% enhancement in construction and occupancy data
  • 70%-75% enhancement in year-built and floor-area data

• Automated Data Quality Toolkit (for US exposures):
  • Further 13% enhancement in year-built data
  • Further 11% enhancement in floor-area data
  • 61% enhancement in number of stories data
     

The outcome — improved portfolio management and capacity planning

By improving data accuracy and introducing critical building attributes into its property portfolio, the MGA was able to achieve a more accurate loss analysis. As a result of this exercise, the MGA noted a decline in average annual loss (AAL) for earthquake (16%), windstorm (5%), and severe convective storm (37%), and an increase for winterstorm (12%) compared with modeled losses using the original data.

Establishing this improved modeled loss accuracy helped the MGA adjust the allocation of capacity more effectively. The enhanced data clarity provided a stronger foundation for portfolio management and capacity allocation decisions. It also enhanced claims preparedness through a more accurate event loss analysis.