TOP
Over the past several years, the concept of Behavioral Economics has turned mainstream, even as it challenges the very foundation of traditional economic thinking by introducing the concept of dual system thinking (“fast” and “slow”). Behavioral Economics is defined as “the study of psychology as it relates to the economic decision-making processes of individuals and institutions.”1
In a parallel fashion, Life insurers have taken a fresh look at their established traditional underwriting practices as they consider and implement streamlined accelerated underwriting (AU) options, removing the need for paramedical exams and/or fluid collection for qualified applicants.
Given the many benefits of AU – including decreasing time from application to issue and improving the experience of qualified customers – there are many favorable impacts. However, given the more streamlined nature of the process, it also has an increased potential for material misrepresentation, anti-selection and fraud. As the list of carriers embracing the benefits of AU continues to grow, so does the need for a methodical and targeted approach to designing related applications and personal health information (PHI) scripts.
By implementing a toolkit using Behavioral Economics (BE) approaches, carriers can offset some of the challenges of AU and promote improved disclosure of medical conditions in the application process.
Gen Re has conducted several BE‑focused studies on application disclosure. We’ve found that widely used, open-ended question formats are less effective than other designs (see earlier article).
Based on carriers’ interest and experience in AU, our most recent BE experiment was inspired by “real life” examples – current applications used by Individual Life carriers in the marketplace today, including many who are currently offering AU.2
Our latest BE experiment resulted in three key takeaways:
From Gen Re’s earlier studies, we learned that the scale grid pattern is the most effective in encouraging greater medical condition disclosure, and this result was further highlighted in our latest experiment. Questions with either 3- or 5‑point scale grid designs prompted the highest level of condition disclosure, with over 80% of respondents reporting any condition, compared to results as low as 55% for other treatment groups tested. While both scale grid designs were deemed effective, the 3‑point scale had a slightly stronger impact, providing an alternative for those who are concerned about the relative size of the question on the application, especially when using a cell phone or other mobile device to complete. The average time spent on the scale grid design questionnaires was higher than other treatment groups, which may be due to the increased focus needed while completing.
When implementing BE methods, it’s clear that small changes can make a big difference. We tested two treatment groups, both using “Yes/No” checkboxes followed by several conditions available to select from. However, one treatment group stated up front that the individual must provide details for any “Yes” answer. For the other group, applicants were not made aware that details would need to be provided until all questions were answered. The results clearly showed that the simple change of holding off to ask about details until after they had selected conditions resulted in a higher level of condition disclosure.
A question design showing smaller groups of conditions followed by “Yes/No” checkboxes was not as effective at increasing condition disclosure as compared to other options, including a more expanded list of singular conditions with “Yes/No” checkboxes. This may be due to lower engagement and thoroughness on the part of the respondents, as the average time to complete was the lowest across all groups tested.
Gen Re actively employed another BE approach, adding a cognitive task after the individuals had completed the application. We presented an exercise with eight scrambled words (only four represented actual words) and asked respondents to unscramble the list. We took this approach to test the hypothesis that questions requiring more work to complete would place a greater cognitive load on the individual completing the application.
In our study, over half of respondents were able to correctly “unscramble” the four words, no matter the question design. However, the 5‑point scale grid pattern mentioned above received the highest percentage (60%) answering correctly, suggesting that the increased time to complete this design does not impact the level of engagement.
There are other applications where BE approaches can be helpful in the AU space. This includes:
