In the television drama, "Person of Interest," massive amounts of surveillance data, complex algorithms, and some interesting special effects help us imagine just how far data analysis and predictive analytics might someday go in order to predict risk. While the show’s early 2011 premiere was purely fictitious, the explosive case of N.S.A. whistle-blower Edward Snowden served to blur the show’s basis between fiction and reality.

So, what is the reality of predictive analytics in our industry and how does it affect you, the analytical broker? Lets start with a definition of predictive analytics:

Predictive analytics is the result of combining large amounts of data, business knowledge, and analytical techniques to generate new insights and business value.

There are two areas I want to address in this blog:

1. Insurance Companies: How are insurance companies using (or going to use) predictive analytics and how does that effect your clients?
2. Clients: How can you use predictive analytics to help your clients?

Predictive Analytics: Insurance Companies

Predictive analytics’ most dramatic impact stems from the fact that insurance companies can now price each risk on a large number of parameters tailored specifically for each individual client. A recent study by Towers Watson shows continued growth in the usage of predictive analytics to price risk and make underwriting decisions. This growth has occurred across all lines of coverage. Additionally, most carriers indicate that they believe predictive modeling has improved their profitability and pricing "accuracy".

I believe it is that last observation, "accuracy" of pricing, that may require you to adjust the way you guide clients through the underwriting process. In the past, depending on how far you want to go back, the underwriting process involved a questionnaire, a potential site visit, and a talk with the account’s broker. This process represented the human brain’s attempt at "predictive modeling." We did so without the aid of technology, instead utilizing various decision-making techniques. This old process gave considerable consideration to the broker’s knowledge and view of the account. Said broker often felt like they influenced the underwriter’s perception of the account and, for the most part, this was true; the broker added considerable value to the process.

Today, in an attempt to improve the decision-making process, less emphasis is placed on human judgment and more is placed on the analytically based and data driven predictive model. The following explains what this means for you and your clients.

You need to ask the carriers you are working with to explain their models to you. They will, undoubtedly, resist or tell you that their modeling process is a "black box." When this happens, push hard to get as much detail as you can about how pricing and underwriting decisions are being made. Your knowledge of the model must then be combined with your knowledge of the client to form a strategic plan for improving the performance of your client’s data within the model’s context.

The most current models are quite sophisticated and proving themselves to be more accurate than the old human judgment-based models. We should not resist this new process of pricing and evaluating risk, but we must understand what the models tell us. What areas are crucial to the assessment? What client characteristics create the greatest risk profile? What can you do to bring the traits and attributes of your client in line with an improved model assessment?

The bottom line is that you must shift your focus onto understanding the predictive models that are determining your client’s pricing and underwriting successes or failures. This will not be easy, but I suggest finding carriers that are willing to help you develop this knowledge base.

Predictive Analytics: Clients

The potential applications of data mining and predictive analytics within a business setting are limited only by your imagination. Really, the question is: What are the appropriate areas where your consultation might be expected and beneficial? First, if your client has any self-insured risk, then the estimates for future losses and reserves (for past years) should be based on the latest analytical methods. This is something we can help you with here at SIGMA. Second, if your client has significant amounts of claims data, then said data may prove to be a gold mine of information for predicting claim severity or preventing future claims. Gathering that information is certainly something an actuarial firm can assist you with, but you may want to hire a general data mining expert to conduct both this type of research and other data mining and predictive modeling efforts for your clients.

How you assist your clients with predictive modeling will partially depend on their definition of risk and the extent to which they have implemented a robust Enterprise Risk Management (ERM) effort. If you recall from my blog last month, ERM provides the framework within which analytics can work. The definition of risk could be broad enough to include all business risk and opportunity, thus introducing a broader set of topics that may require the utilization of predictive modeling techniques.

Conclusion

The role that an analytical broker plays within a business is certain to expand in the decade ahead due to the broader adoption of an ISO 31000 type ERM framework by employers. If you prepare and staff for this expansion, you will differentiate yourself in the marketplace and be positioned to represent the largest and most profitable clients.

If you have not registered for access to our portal, you can do so by clicking here. Registration is quick, easy, and access to all of our resources is free.

We welcome your feedback by posting a comment, or contact Tim at TLC@SIGMAactuary.com.
© 2015 SIGMA Actuarial Consulting Group, Inc.