Predictive Modeling’s New Mantra

With the power of predictive modeling and its resulting efficiencies, life insurers can simplify underwriting, reducing both time and complexity.


Many discussions about disruption and innovation in life insurance have focused on the distribution channel and improving the customer experience. Others have directed attention to adding efficiencies in the underwriting process to shorten issuance times. The challenges, particularly against the headwinds of COVID-19, have been a priority of life insurance carriers and reinsurers. Predictive modeling has been their primary tool, with an emphasis on its integration with other elements of the underwriting process.

What is predictive modeling?

Predictive modeling is the statistical process of using historical data and machine learning techniques to analyze patterns and predict possible future outcomes or probabilities. It is widely used in many industries. In the insurance industry, its leading uses are to increase automation levels, improve the accuracy of pricing models and fine-tune marketing initiatives.

In life insurance, predictive modeling is used for a variety of applications to benefit the industry and the consumer alike. For example, predictive models build underwriting programs that accelerate the application process and improve the overall customer experience. In the past, an application would take weeks and sometimes even months to complete, as bodily fluid tests were usually required. Thanks to predictive modeling for accelerated underwriting programs, the process can be shortened to just a few days. Often, fluid tests can be waived if the carrier can obtain favorable data from applicants, such as their prescription drug history. It’s a win-win for the insurer and the consumer.

See also: Predictive Analytics: Now You See It....

Pandemic effects

The pandemic has heightened the need to improve efficiencies in life insurance. A tight labor market has led to shortages of medical personnel, and higher volumes of patient visits have created stress in the system, meaning that the test results underwriters need may take longer to obtain. This leads to frustrations for insurance companies, insurance agents and ultimately the applicant, as the time to issue a policy has increased.

One positive outcome of COVID for many life insurance companies has been the push to digitalize as traditional face-to-face selling and paper-based applications are no longer the industry norm. The acceleration of digitalization has allowed more historical data to become available for predictive modeling.

As an increasing number of life insurance companies implement data digitalization, predictive modeling should become more widely used. As more advanced machine learning techniques are developed, the predictive models should continue to become more refined.

Accelerated underwriting programs mean that predictive models will mostly handle the straightforward and repetitive cases, and human underwriters can focus their review on more complex cases. A concrete example is the removal of fluid-test requirements for policies with lower face value. These innovations will lead to faster adoption of predictive modeling.

Using artificial intelligence with integrity

As predictive modeling becomes increasingly prevalent in the life insurance industry, a legal and regulatory framework is emerging to govern its use. For example, the National Association of Insurance Commissioners (NAIC) has adopted five key tenets in its guidance on the use of artificial intelligence (AI). Known as “FACTS,” the acronym stands for fair (and ethical), accountable, compliant, transparent and secure (safe and robust). It is important that the industry embraces the NAIC's AI principles, taking social responsibility into account when building predictive models. What is chosen to use in the model is as important as how the model is used. 

Quite often, underwriters will develop a hypothesis that is then tested against available data. As the industry continues to accumulate data related to the near- and long-term impacts of COVID, this approach will allow for more accurate forecasting of how to underwrite and price life insurance. And this work is not being done in a silo; it intersects with other initiatives that the life insurance industry is undertaking to improve the customer experience and the efficiency and effectiveness of the underwriting process.

See also: ITL FOCUS: Life Insurance

The ultimate goal is to integrate the underwriting process with evolving predictive analytics capabilities. To accomplish this objective involves correlating details from a variety of evidence sources, including prescription history, criminal history and digital electronic health record (EHR) databases, with information collected on the policy application to provide underwriting recommendations in real time.

Some view the increasing use of data analytics by insurance companies with concern. However, the thoughtful and responsible use of predictive modeling benefits a variety of stakeholders. With the power of predictive modeling and its resulting efficiencies, life insurance companies can simplify the underwriting process, reducing both time and complexity. This is significant as it enables broader reach, allowing insurance companies to target underserved demographic segments such as the middle market, which is important to the industry and society at large. Ultimately, the longstanding life insurance protection gap should narrow.  

Life insurance companies need to embrace these technology-driven solutions to stay relevant and better serve their customers ― which will ensure their ability to not just survive but thrive.

David Zhu

Profile picture for user DavidZhu

David Zhu

David Zhu, FSA, FCIA, Ph.D., is vice president, head of Americas Data Analytics at SCOR Global Life Americas. He leads the design and creation of predictive models and artificial intelligence capabilities.

A fellow of the Society of Actuaries, Zhu’s expertise focuses on topics related to new statistical techniques for designing future generation retirement and insurance solutions that address asset allocation and policyholder behavior.

He holds a Ph.D. in operations research from Massachusetts Institute of Technology. 

Read More