Article

Accelerated underwriting in life insurance: what examiners need to know.

Aug. 19, 2021

Life insurance underwriting has historically been a manual and sometimes tedious process. The process has involved collecting evidence of insurability during the application process to assess the mortality risks of individual applicants. The two largest pieces of evidence have been paramedical exams, with collection of vital signs (build, blood pressure, etc.) and fluids for testing, and attending physician statements particularly for cases with large face amounts and/or older issue ages. There has been a shift in recent years to accelerated underwriting (AUW) programs, in which underwriting requirements are waived for certain applicants. These AUW programs are typically limited to certain issue ages and/or face amounts and enable only a portion of those eligible to qualify without traditional underwriting requirements. According to the SOA 2019 Survey, most companies currently limit the application of accelerated underwriting to term life insurance policies with a face amount between $100,000 and $1 million and applicants between the age of 18 and 60. The COVID-19 pandemic sped up the adoption of AUW in the industry as both consumers and insurers looked for options to purchase and write policies that relied more on technology and involved less in-person contact. The increasing use of AUW, poses new risks and considerations for regulators, examiners, and insurers.

What is accelerated underwriting?

The definition of AUW, per the VM-20 Underwriting Definitions Sub-group, is “a process to replace traditional underwriting and allow some applicants to have certain medical requirements (such as paramedical exams and fluid collection) waived. The process generally uses predicative models or machine learning algorithms to analyze data pertaining to the applicant, which includes both traditional and non-traditional underwriting data that comes from both the applicant and external sources.” A more palatable definition is that AUW is a process that is making it faster and easier for people with good health and good credit to obtain life insurance. With AUW, these applicants can typically get a term life insurance policy without a medical exam. Per the NAIC, “As consumers increasingly expect on-demand digital services, some insurers are able to use AUW to replace paramedical examinations with data from external sources along with new analytics and modeling techniques."

How is data science being utilized to drive increased use of accelerated underwriting?

Data science is one of the key pillars supporting AUW. Insurers are using a multitude of data to feed their algorithms to improve and “teach” their programs. Insurers employ predictive analytics or machine learning algorithms, which are tested and modified via back-testing, a process of running historical data through a model to test how actual outcomes compare to what would have been projected by the model. Underwriter feedback is also crucial in improving the use of AUW. Over time AUW programs use data science to become smarter and more efficient.

What are some of the key risks?

The NAIC is constantly monitoring ongoing trends within the insurance industry. AUW is one of the latest and greatest methods insurers are using to sell insurance products. Yet with new processes also come new risks. Currently, the NAIC does not have formal identified risks specific to AUW. However, there are a number of risks that may affect insurers:

• Insurers generally have significantly increased investment in and plans to use AUW at a larger scale for new business applications and pricing. There is a risk that process governance is not appropriate resulting in inadequate risk selection and mispricing.
• If class A (preferred) is priced appropriately according to AUW models and class B (standard) is priced appropriately according to AUW models but the AI puts a Class B in Class A categorization and pricing, there is a risk of misclassification and subsequent pricing.
• There is a potential for the models to have unintended impacts on protected class and unfair discrimination. 
• There is a risk that insurers relying on this less understood data may be systematically miscalculating an applicant’s expected mortality and putting their solvency at risk. 
• As companies begin to expand the maximum face values of benefits, the lack of traditional information used in pricing and reserving larger policies could lead insurers to expose themselves to pricing and reserving risks, among other exposures.

What procedures should examiners consider?

The data and models used in AUW pose new risks for the insurers. Initial assumptions about correlations with mortality need to be further tested. It is reasonable to believe a person’s behavior has a strong correlation with mortality risk. Relevant data for testing this assumption includes gym memberships, profession, marital status, family size, grocery shopping habits, wearable technology and credit scores. Although medical data may be better correlated with mortality, behavioral data may lead to questionable conclusions as correlation may be confused with causation. The following example presented at an Accelerated Underwriting Working Group, for example, describes a questionable conclusion: “a high-income individual is perceived as someone who has excellent medical care. However, a high-income individual may also have the resources for illegal drug use or other dangerous habits or hobbies. A healthy young couple, on the other hand, may not have the disposable income to join a gym, however, they may exercise on their own. In either case, the lack of a gym membership or lower income may not indicate an increased mortality risk.”

AUW involves a multitude of departments working in sync. Examiners should consider what sort of influence and governance the major departments, including Underwriting, Actuarial, Compliance, Reinsurance, and Data Analytics each have throughout the entire process. This increasing use of AUW presents new regulatory challenges. While differences in process have evolved, the concern regulators and examiners have is the same as with all underwriting - whether or not the process is fair, transparent and secure. Examiners should additionally consider some of the following:

• How does the insurer review the comparison of results between a human underwriter and AUW?
• How do the mortality results from AUW compare to the expected mortality?
• What sort of back-testing does the insurer perform to mitigate risks?
• What sort of policies, controls, and processes for the development, deployment, and maintenance of the artificial intelligence (AI) systems does the insurer have?
• How does the insurer ensure that the machine learning and predictive models are ethical and eliminate bias in accelerated underwriting?
• How does the insurer take steps to ensure data inputs are accurate and reliable?
• What does the insurer do to ensure that the external data sources, algorithms or predictive models are based on sound actuarial principles with a valid explanation or rationale for any claimed correlation and causal connection?
• What processes and controls are in place to protect consumer privacy and ensure consumer data is secure?
• What processes and controls are in place to catch and correct mistakes if found?

As insurers continue to innovate with AI and data driven underwriting and develop AUW, examiners will have to continue to dig deeper to fully understand all the risks related to these new processes. The process of gaining an understanding and assessing the effectiveness of AUW will likely require specialist expertise; both the traditional IT and Actuarial Specialists we are used to working with as well as new specialists such as Data Scientists. AUW is no longer just the future, but it is the present and examiners need to be aware to head off potential risks for the benefit of insurers and insureds.