Predictive modeling in specialty insurance
How specialty insurers can speed adoption of predictive models
Matt Edwards | May 20, 2015
The popularity of predictive modeling in insurance continues to grow, particularly in the personal and small commercial insurance segments. Traditionally, these segments have used predictive models to assist with everything from underwriting decisions to policy pricing to potential claims expenses.
Today, there’s more opportunity than ever for mid- and large-size commercial and specialty insurance companies to also apply predictive modeling across their segments to improve underwriting decision making, pricing, and reserving—while simultaneously decreasing cycle times and expenses. A recent Towers Watson survey of property and casualty insurance executives shows that predictive modeling helped 98% of respondents improve rate accuracy. Additionally, 91% reported a positive impact on loss ratios, and 87% reported a positive impact on profitability.
It’s become standard operating procedure for most insurers to incorporate predictive models into their underwriting decision making and rating and pricing processes. In that same Towers Watson survey, 97% of personal lines auto carriers stated that they already use predictive models, with an additional 3% planning to use them. Homeowners insurance providers reported similar numbers.
Adoption of predictive models has been slower in commercial and specialty lines for many reasons, including the nature and complexity of the business and the amount and accuracy of data available. Specialty lines continue to lag behind in surveys, with close to 30% of carriers stating they do not use, or have no plan to use, predictive models; and only 44% of carriers actively using predictive models at this time.
What’s this all mean? That there’s a huge opportunity for specialty carriers to use predictive models to gain a competitive advantage.
Unfortunately, it’s not that simple. Along with technical challenges, organizations must address two other key areas to successfully make the transition: organizational change and operational integration.
Organizationally, companies struggle making the cultural shift from a traditional approach to underwriting and pricing to a data-driven, predictive modeling approach. To make this cultural shift successful, all levels of the organization must support the change and understand the reasons behind it. This starts with the company’s C-Suite executives setting the vision and providing guidance on how the successful implementation of predictive models will help the organization achieve that vision.
Slalom partnered with a company in the specialty insurance space that was struggling with the adoption of its new predictive models. The company’s underwriters felt that their many years of experience and expertise were being replaced by the predictive models—which wasn’t the case. This misinterpretation was partly due to inconsistent messaging within the different lines of the business, and partly due to minimal involvement from the underwriting teams in the development of the tools. As we have continued to partner with the company, it has modified its change management approach to drive better adoption, including consistent messaging from leadership and greater transparency throughout the development process.
Organizational adoption also requires user education on how the predictive models work; the benefits that they are intended to provide; and how the users can help improve the models over time. Providing this transparency helps the end users see the models as an additional tool to assist them with underwriting and pricing work, rather than a replacement for their knowledge and experience.
As part of the adoption process with the specialty insurance client mentioned earlier, we helped develop reporting capabilities that give underwriters insights into how the models are performing. This has greatly improved adoption as subsequent predictive capabilities have been deployed.
Along with organizational challenges, companies are also faced with the operational challenges of how to integrate predictive models into their existing processes and technologies. Companies must determine whether or not predictive models will be fully integrated and automated into their underwriting and pricing processes, or if they will take a more manual approach. Once a company has determined which way it wants to go, it can look at its existing business processes and determine how it needs to change to incorporate the modeling results.
Along with redesigning business processes and solutions to incorporate predictive models, organizations also need to establish technology support structures to ensure that the models are available when needed and that the data—both model inputs and outputs—is captured and reportable. Additional processes also need to be established to monitor performance to ensure that the models are performing as designed, and if something is amiss, that those models can be swiftly recalibrated.
We helped a client build the technical infrastructure that gives end users on-demand access to its predictive models to make quicker underwriting and pricing decisions. Addressing the company's operational and organizational challenges has allowed the client to provide holistic insights into adoption, and identify areas where additional focus is needed. This approach has allowed our client to become a leader in the predictive analytics space and reap the benefits of improved underwriting results, pricing consistency, and underwriting efficiency.
The bottom line
The data is clear: specialty insurance carriers have a tremendous opportunity to capitalize on the benefits of predictive models. But to do so successfully, they need to surmount organizational and operational hurdles—and find the right partner to help.
Matt Edwards is a practice area lead in Slalom’s Insurance Practice. With 15+ years of industry experience, he partners with our local markets and national delivery teams to develop and implement solutions that address our insurance clients' most strategic business and technology challenges.