Untapped Possibilities: Unlocking Better Risk Management Through Price Modeling

By Harris Clarke, Vice President of Operations at PEMCO Mutual Insurance

Harris Clarke, Vice President of Operations at PEMCO Mutual Insurance

With the explosion of connected devices, technologies collecting data and machine learning, the insurance industry is able to tap into new, dynamic factors for more accurate risk assessment and price modeling. Providers now have an unparalleled opportunity to leverage this data to make a risk assessment and management more objective, which has historically been a recurring challenge. Price algorithms can now depict the risk that is assumed with more precision by referencing larger and more diverse data sources, which have unlocked new variables.

In-car telematics using a device plugged into the OBDII port has allowed companies to develop auto rating based on driving behavior with moderate customer engagement.  The next generation of telematics will focus on mobile phones.  Coupling mobile applications with driver identification algorithms mean that carriers can have higher confidence in knowing who is driving and the risk they represent.  Mobile applications have the added benefit of enabling an easy way to provide feedback to drivers interested in improving their profile.  These applications allow carriers to measure the impact of phone use on vehicle accident rates – and influence their customers to eliminate phone distractions while driving.

Accurate price modeling will help reduce the frictional costs and inefficiencies that negatively impact profitability. As we learn more information about specific assets or customers, some opportunities may be identified as too risky to insure. To avoid jeopardizing the business by assuming too much risk, insurers will now have data to support cutting ties with individual customers or specific assets until new parameters can be put in place to ensure risk is priced and pooled correctly.

"The next generation of telematics will focus on mobile phones"

Another new way to get information is crowd-sourced data.  Weather networks where individuals contribute weather data from their own home-based stations can provide localized, historical weather data.  This data, when overlaid with storm loss data, will allow insurers to more accurately price weather-driven events. 

The emergence of data sets that are a product of machine learning enables insurers to better assess the risks they are taking on and, in some situations, use for rate development.   Emerging analytics firms are using aerial imagery with machine learning to index images into classes that provide carriers with information like defensible space around homes, roof condition, and the presence of swimming pools and trampolines.  Currently, insurers can only ascertain this information through expensive home inspections.

Algorithms are automated and as a result, presumed objective.  However, pricing methods, underwriting rules, and rating algorithms are created by humans.  Inevitably, not only can these work products contain subjective biases in their construction, but they also do not always contemplate every possible case.  As a result, there should be a human-centered form of adjudication and review.

We are just beginning to grasp the positive financial and social potential of these tools and technologies. However, these benefits can only be realized if insurance companies start taking proactive steps to fine-tune pricing models and more accurately categorize the risk posed by customers and assets now.

Read Also

Ensuring your Seat on the Leadership Table

Ensuring your Seat on the Leadership Table

David Otte, CAO and Former CIO, Bingham Greenebaum Doll LLP
How to Make a Notoriously Reactive Industry Proactive

How to Make a Notoriously Reactive Industry Proactive

Mike Gulla, Senior Director of Underwriting, Hippo Insurance

Weekly Brief

Top 10 Risk Management Tech Companies - 2018

Risk Management Special