Data is often the key ingredient to an insurtech’s success. From the use of drone data to automatically pay insurance claims after a hurricane, to AI that can predict the likelihood of insurance fraud – big data has the potential to revolutionize the insurance industry. But before an insurtech can harness its data models to underwrite insurance more effectively, it oftentimes must first must obtain regulatory approval to do so.

For many lines of insurance, the rates an insurtech may charge its policyholders must be filed with and approved by the insurance regulator of the state in which the insurtech wants to offer its policy. Regulators will want to understand and need to approve the data models used to charge different policyholders different rates. Many of our clients find this rate filing process to be opaque and worry that obtaining prior regulatory approval will significantly hinder their speed to market.

In order to address these industry concerns, the National Association of Insurance Commissioners’ Casualty Actuarial and Statistical Task Force is drafting a white paper on best practices for the regulatory review of the predictive data models used by insurance companies to underwrite insurance and set rates. The task force exposed portions of the draft white paper for public comment at the NAIC’s 2019 Summer Meeting in New York and is currently working on revising the draft based on the feedback they have received to date. Two areas of concern identified by the task force, which are certainly key issues for our clients, are to what extent the insurer’s data models should be kept confidential, and whether an insurer’s data models must establish a causal relationship rather than just a correlation between the data and the risk insured.

In addition, the NAIC’s big data working group is developing a training program for regulators on data analytics and is looking to hire data experts to provide technical support to regulators in its review of predictive data models. These efforts should help streamline and add transparency to the rate filing process, allowing insurtechs to leverage their data in innovative ways to mitigate risk and provide their customers with cheaper and more effective policies. In the meantime, below are four recommendations for how to navigate the rate approval process as an insurtech.

First, we often recommend partnering with an established insurance or reinsurance company at the outset of the rate filing process. Insurance companies have the experience, expertise and long-standing relationships with the regulators to help smooth the regulatory approval process. Startups can benefit from leveraging these resources to get their data models approved in a cost-effective and timely manner.

Second, many of our clients will begin by filing a plain-vanilla policy that is closely based on a policy that has been previously filed with and approved by the regulators. This allows the policy to be approved as quickly as possible. Once the insurtech has its policy approved and is expanding the customer base, it can file amendments to the policy to offer customers different rates based on the data models. This two-step process enables a speedy entry to market while providing insurtechs with the flexibility to offer innovative products going forward.

Third, insurtechs should take all steps necessary to protect their intellectual property. If an insurtech is partnering with an insurance company, the partnership agreement should, to the greatest extent possible, make it clear that the data models developed by the insurtech and the data generated by the partnership are the property of the insurtech, not the insurance company. Insurtechs should also work to keep their data models as confidential as possible when making rate filings with regulators to protect any trade secrets or other intellectual property.

Fourth, insurtechs should know their data models inside and out and be prepared to justify their rates to regulators. As discussed above, regulators will often require insurtechs to provide a causal relationship between the data factors and the risk insured, not just a statistical correlation. Insurtechs should be prepared to provide regulators with an explanation of how the data they are relying on impacts risk. In addition, insurance laws prohibit ratemaking on the basis of certain protected classes such as race, gender, sexual orientation, etc., and insurtechs must make sure that none of the data they are relying on could be construed as a proxy for race, gender or any other protected class.

Regulators recognize the opportunity to harness big data to offer more consumer-friendly products to policyholders and are willing to work with insurtechs to understand and approve the data models they use to offer these products. Following the recommendations above and working proactively with regulators should make the data model regulatory approval process as smooth as possible.