Big Data Sharing and Predictive Modeling: 5 Things for Regulators to Consider

Big Data technologies are no doubt recalibrating the insurance industry. This expanded form of modeling by insurers is incredibly dynamic having an almost limitless range of new variable assumptions and branched connections that will significantly improve and revolutionize how insurers measure risk. Yes, this means that some will pay less and others more but with an added chance to get these costs offset via direct sales that eliminates the broker and underwriting expenses. Regardless, it will arguably be more equitable because everyone gets charged a cost closer to their actual risk. But, this equity dispersion comes at a cost in the form of individualized risk profiling that if too prolific, could lead to excessive risk pool segmentation, adverse risk selection and new individualized risk assessments that could price out new segments of policyholders.

(Related: 3 Ways to Use Data Analytics to Drive Growth)

I believe the U.S. market is ready to realize this potential from these expansive technologies and networks but a regulatory pathway needs to be established to safeguard adherence to our core insurance principles and an individual’s right of privacy.

This new and exciting frontier of seemingly limitless data, cross-integrations, and fluid networks brings a new approach in measuring and identifying risk which will ultimately transform our industry. Perhaps the most significant advancement is in the development of the policyholder behavioral risk profiles. This new risk measure brings a new degree of precision but since behavior is often inconsistent and ever changing, its success depends upon a commitment to its continuous refinement.

The power of these developing data technologies adds real value to sales, pricing, underwriting, claims and service administration which all translates to expense reduction, fraud reduction, improved precision/accuracy, streamlined administration and time savings. But, the current excitement dominating discussions is in drastic need of sober regulation to move it along.

It is understood that all data targeted must be associative, authorized, secured and disclosed. Most importantly, it is expected that every piece of public or non-public data targeted and collected must credibly correlate to the risk transferred. Using data that is credible (strong probability or likelihood) and that maintains quantifiably demonstrable experience correlating to the risk is an insurance gold standard and something that cannot be compromised.

Here are five issues regulators must tackle:

1. “Disruptors” creating wide scale disintermediation and cherry picking risk or selecting against traditional insurance measures and methods.