Risk Insider: Ernie Feirer

Predictive Modeling and Small Commercial Risk

By: | March 2, 2018

Ernie Feirer, CPCU, is Vice President and General Manager, Commercial Insurance, at LexisNexis Risk Solutions, where he is responsible for developing a suite of solutions for the commercial insurance market. He can be reached at [email protected].

The small commercial insurance sector has been relatively slow to adopt predictive modeling despite its proven successes in other segments. Often it is due to a lack of resources. Other times it’s because an insurer doesn’t know how to build an effective model. Or there may be concerns about engaging the organization in the predictive modeling process.

The good news is that there are simple best practices businesses can use to benefit from predictive modeling and reduce risk vulnerability.

Leveraging from the product development life cycle

Creating and using an effective predictive model can be likened to following a four-stage product development lifecycle process: ideation, design and development, implementation, and monitoring. Following this process can help integrate predictive modeling into a workflow to better predict risk and improve business outcomes.

Step 1: Ideation

The starting place is identifying a problem that needs solving and determining whether or not a predictive model can help. The critical first steps are garnering strong executive sponsorship for the effort and defining a committed cross-functional team that can help bring the idea to reality.

Step 2: Design and development

In the small commercial segment, there’s a growing movement to use predictive modeling for risk assessment and pricing through building insurance scores that order risks in terms of loss propensity. Designing and developing such a model is a very iterative process, which begins with data exploration, followed by training and validating the model, and finally ensuring it meets regulatory compliance.

During data exploration, the team members evaluate data sources. There are many third-party data sources to consider. These include commercial credit from the big credit bureaus, or business owner consumer credit for micro businesses. Public records on the business or business owner are also good sources for assessing risk. Additionally, many carriers choose to integrate prior loss or geospatial data into their models.

An insurance score approach can streamline underwriting and improve pricing based upon the risk associated with the account. With proper segmentation, an insurance score assists underwriting automation to potentially decline, refer, or accept business without the intervention of an underwriter.

Step 3: Implementation

Once the model has been designed and proven, it’s time to implement it within the workflow. This stage requires careful planning. Implementation impacts many parts of the organization and requires thoughtful decision-making. Questions to ask include: Will the score only be used for discretionary pricing or will it be incorporated more broadly? Which, if any, underwriting rules and procedures will change?

To ensure the model is successful you need to work with IT to implement the final model, modify the application workflow to use the model’s score, define customer dispute resolution processes if applicable, and deploy stakeholder training.

Step 4: Monitoring

A model is only as good as the results it produces. To make sure your model is working the way you want it to, it’s critical that you track ongoing performance, make any necessary tweaks, and monitor its efficacy.

For example, scores should be tracked both when they are used and when they are overridden. When they are overridden, who overrode the score and why? Allowing for and documenting score overrides provides valuable insight into score limitations and how the score and its implementation should be improved in the future.

You should periodically monitor the efficacy of the model to determine if it’s achieving the desired results. If not, a deep dive into the underlying causes is required. You might need to periodically recalibrate or rebuild your models to ensure their performance. You may also want to incorporate the score into your underwriting dashboard and business intelligence reports.

Putting it all Together

Aligning a predictive modeling integration with the product development lifecycle process is a methodology any carrier can follow.  It enables commercial insurers to realize the full benefits of predictive modeling for small commercial risk assessment and pricing. Keys to success include executive sponsorship, a competent and engaged cross-functional project team, and a four stage life-cycle process of ideation, design and development, implementation, and monitoring to steer the process.

Mathew Stordy, Director of Commercial Insurance for LexisNexis Risk Solutions, also contributed to this article.

More from Risk & Insurance