Life insurers are cautious about integrating medical and technological advances into their underwriting procedures in order to avoid unintended consequences to decision accuracy. Still, some recent advances are intriguing, and companies are starting to incorporate these innovations into their processes. Four trends stand out.
All underwriting shops work toward delivering more accurate results more quickly, less intrusively and less expensively. Anything that makes the underwriting process less burdensome for both clients and producers without impairing mortality expectations would be considered a win.
In many cases, underwriters (or vendors appealing to underwriters) suggest replacing one test for another – for example, replacing MD examinations with a combination of paramed exams, prescription drug checks and tele-interviews. Some companies are looking to the NT-ProBNP protein test to replace EKG/treadmill exams, and cognitive tests could provide greater insight into an older-age applicant’s mental acuity (which is positively correlated to mortality).
There are trade-offs. Paramed exams and tele-interviews are convenient to the applicant, but neither provides the level of medical insight and critical face-to-face review possible with a physician’s exam. The NT-ProBNP test may match the overall protective value of the EKG/treadmill but could produce a different distribution of accepted and declined risks, with pricing implications. Lastly, any inconsistency in how a cognitive test is given (or evaluated) can produce inconsistent readings of the applicant’s mental state.
Straight-Through Processing (STP)
This term can have different meanings, even within a single company. The goal of STP is to use technology to increase processing efficiencies. Some people may think scanning an image of a paper application is STP. While that may accelerate transmission time, image capture provides no ability to manipulate the information collected. Additionally, the risk of incomplete or illegible applications still exists.
The purest definition of STP is: A process whereby data captured electronically and completely through the application is used throughout the entire life of the policy, from underwriting through issue, administration and ultimately claims. This definition implies that the company uses some form of an underwriting engine (another term with multiple definitions) to at least assist in managing clean-sheet applications. Ideally, all of this data feeds other systems, such as pricing models and actual-to-expected studies.
In truth, few companies have true STP capabilities. STP is all-encompassing from a systems perspective and therefore represents a big commitment of finances and other resources. Even the best capitalized life insurers can see their capability to share data limited by legacy systems and business unit silos. Not only is data not shared among profit centers (e.g., individual life and group life and annuities), functional areas are cut off as well. Underwriters may not appreciate the data’s value to their colleagues in pricing and valuation.
Increasing Lab Role
Traditionally, major labs have played a vital role as partner in the underwriting process, providing specific positive test result levels for various fluid panels, especially blood. This information helps company underwriters to determine the final rating based on proprietary test criteria.
Labs are increasingly expanding their services. They have developed multivariate algorithms that compare submitted applicants’ panel results with the Social Security Death Master File. The labs are also working one-on-one with life insurers and reinsurers to correlate the labs’ mortality to each individual company’s actual experience. Using an insured population instead of applicant populations is intended to provide more predictable results and credibility.
Some chief underwriters and medical directors appreciate the fine-tuning that labs may provide as well as the quicker and less invasive process. Others question whether the decisions meet the company’s stated expectations and raise concerns about the “black box” nature of the labs’ decision-making processes. There is also uncertain buy-in from pricing and product development experts. However, labs will continue to seek ways to deepen their partnership with direct writers (and reinsurers) to improve their risk assessment.
The popularity of predictive modeling in the property-casualty business has created interest in the life insurance sector. Predictive modeling is an estimation tool that uses unaffiliated but highly correlated data points to determine more accurate results. The models use complex algorithms to determine appropriate correlation to predicted outcomes.
Certain concepts mentioned above may be considered components of predictive modeling (e.g., prescription drug information). However, vendors are seeking ways to incorporate greater consumer behavior-oriented information to determine an applicant’s mortality risk.
Predictive modeling’s high degree of validity in the property-casualty industry has increased its allure in the life sector, but the concept’s future in the life business remains unclear. Perhaps the key challenge is developing an effective predictive model that can factor in the long inforce time horizon of a life insurance policy versus the six-month/one-year term of a typical P/C policy.
Preserving Protective Value
As a reinsurer we increasingly are asked to approve changes to clients’ underwriting guidelines. While we are happy to review such requests, our primary concern is that any requested change does not adversely affect the protective value of more traditional underwriting standards.
For companies considering requesting such changes, below are a few factors we consider when weighing any such changes:
Provide Supporting Data. Before a client initiates discussion, we strongly advise you to prepare data that supports using such technologies. For example, run the proposed system as a beta in parallel to your existing underwriting procedures and compare the results. Outcomes do not necessarily need to be identical, but outliers should be highlighted and sufficiently explained.
Limit New Procedures to Specific Classes or Products. New tests can be used as “gatekeepers” to mitigate potential losses due to erroneous application. For example, the NT-ProBNP and cognitive tests appear to be more effective for older-age applicants.
Early Inclusion Can Facilitate the Process. Keeping your reinsurers informed can not only aid in the ultimate adoption of any new tests but also allow them to provide valuable feedback during the development and testing process. We are happy to help you run protective value studies, both against your own block of business and our reinsured population, and suggest ways in which more accurate results can be achieved