Prudential Financial Applies Predictive Analytics to Life Insurance

The property-and-casualty insurance industry has long dominated in data-driven insights and analytics, but big data is only starting to become a game changer in life insurance. Prudential Financial’s actuary and predictive analytics expert, Christine Hofbeck, explains how it could transform the way she does business.

Predictive analytics can result in lower auto premiums, better ideas on which customers to insure, and the ideal product design. The property-and-casualty insurance industry has been using this technology for the past several years, but life insurance has taken longer to get on board. Despite this sector’s late entry, predictive analytics could fundamentally change the way that actuaries do business.

Christine Hofbeck of Prudential Financial has worked as an actuary and pricing and predictive analytics expert for several years. She explains that it’s taken life insurance much longer than property-and-casualty for two reasons. The first pertains to the longer time horizon required to see outcomes. The life cycle of a property-and-casualty policy can be between six and twelve months, which is much shorter than the sometimes thirty-year life cycle of a life insurance policy.

The second is the fact that property-and-casualty  insurance also entails more claims in a significantly shorter time span than life insurance does. This means that life insurance has less data to draw from, which makes it exceedingly more difficult to model events that happen infrequently.

In initial areas of exploration, insurers find the most success by starting small and gaining quick wins, while introducing the ideas of predictive modeling and testing archaic systems’ capabilities. “Life insurers need to learn to walk before they can run,” Hofbeck says, adding that she’s starting to see advancements in operational efficiencies, pricing accuracy, risk selection, and the customer experience.

Hofbeck contends that, in this industry, the early adopters will have the greatest advantages. It gives them the ability to push undesirable business to competitors through price differentiation. She points to a property-and-casualty example of using credit scores to segment risks: it turns out that those with better credit generally experience fewer losses than those with lower credit, and they are, therefore, a better risk to cover.

Now if this applied to a life insurance policy, the first people to learn about this would capture that client and give him or her a discount before other competitors. Hofbeck likes to use a contemporary analogy to illustrate how pricing and predictive analytics can help the industry find profitable business and push out unprofitable business: “Find your arugula,” she says.

Hofbeck sees this model affecting business the most in four areas: behavioral, operational, pricing, and risk selection. Predictive capacities allow companies to understand policyholder behavior, determine what drives the customer, and predict lapse or churn (when a customer is going to quit a policy). Operationally, life insurance could apply the behavioral insights to inform product design and distribution, optimize marketing segmentation, and optimize plan types for marketing strategies while refining pricing and adding lifetime value and retention models.

But these benefits don’t come without challenges. One obstacle has to do with the fact that life insurance’s margins are razor-thin and the competition is fierce. “Everyone wants to know what everyone else is doing,” Hofbeck says. “That makes companies wary of trying something completely new, like changing the typical pricing variables.”

On top of this, systems are often too outdated to handle the wide variety and amount of data that predictive analytics can provide. Without the budget needed for the upfront rework, it’s often impossible to house the data received from these abilities.

Pricing can often be more regulated than property-and-casualty policies, Hofbeck says. An example of this relates to price elasticity. When companies charge higher premiums to customers that are willing to pay more, insurers raise the price until or just before the customer walks. While this is often allowed in the property-and-casualty world, it is generally not allowed in the life insurance sector, where premiums must be based on risk characteristics.

Because of these challenges, leadership needs to exhibit a great deal of charisma and creativity. Because the outcomes can’t be certain until years into the future, leaders need to be able to influence and convince business partners to take that leap of faith without the power struggle. “I view predictive analytics as a capability that supports and informs business strategy by providing data-driven insights—not one that dictates it,” Hofbeck says. Communication is critical at all stages of the modeling, and Hofbeck urges actuaries to think ahead because life insurers have limited customer data.

The future of data modeling rests on data sources. When a potential customer wants to sign up for a policy today, they may have to go through a medical exam, which can sometimes be quite invasive. If the insurer can instead consider readily available information—such as driving records or prescription histories—for a faster and better customer experience, that may even be less expensive for the insurer.

The problem with this is where to draw the line. Should life insurance companies be able to use social media to track drug usage of customers? Should genetic data be implemented in coverage decisions? With all of the benefits that predictive pricing and analytics provide for the industry, Hofbeck says, “It’s not just a question of whether we can, but also whether we should.”