The Product Development Section Presents Underwriting Issues & Innovation Seminar July 31-August 1, 2017 The Westin O Hare Chicago, IL Pricing Accelerated Underwriting Programs Moderator: Phil Murphy Presenters: Craig E. Hanford, FSA, MAAA Robin Holicky
Actuarial and Underwriting Perspectives: Pricing Accelerated Underwriting Programs CRAIG HANFORD SWISS RE ROBIN HOLICKY ALLIANZ LIFE August 1, 2017
Agenda Evaluating a Predictive Model Cost-Benefit Analysis How Are the Models Used? Evaluating the Performance of the Model 2
Evaluating a Predictive Model Actuarial Retrospective study Risk classification using a model Risk classification using traditional UW Vendor-provided mortality results: A/E mortality by risk score Underwriting Underwriting perspective Gaps in the process and analysis Capturing predictive information and applying it Competitive environment 3
Cost-Benefit Analysis Actuarial Comparison of underwriting expense savings against the additional expected costs from increased mortality resulting from predictive model Review and understand differences that emerge by age, face amount, and gender A good understanding of these details is recommended when setting maximum ages and amounts at which the predictive model will apply 4
Example: 10 year term product offered using a predictive model at face amounts of $100,000 Millions 2.0 1.5 1.0 Costs, Savings, and Net Benefit (NPV) displayed by applicant s Age (population = 100,000 applicants, product = term life with $100K face amount) 0.5 0.0-0.5 <25 25-34 35-44 45-54 55-64 65-74 75-84 85+ -1.0 Cost Benefit NPV 5
Cost-Benefit Analysis Underwriting Timing of the CBA is important to address the intent to go forward as well as the impacts during proof of concept Importance of determining and addressing the impact to the challenges at the different ages The better the understanding of the differences potentially reduces the surprises 6
How Are These Models Used? Actuarial Generally use some form of triaging Depends largely on what proportion of risks you want to run through predictive model accelerated process, and desired competitiveness of pricing Two examples Triage based on a single predictive model Non-triage, using predictive model to simply classify risks Underwriting By triaging experience in proof of concept will identify the specific potential gaps. Suggesting if some tweaking is needed in one or more of the triage requirements. Then if and when additional gaps are identified, going back to the drawing board for additional mods can be evaluated. 7
Single Predictive Model Triage Standard Risk Predictive Model P P M M P M + Predictive Model + Score Cut off = X Favorable Score Action Issue as Pref Risk Life Insurance Application + Full Application + External Data (MVR, MIB, Rx) + Smoker? Y,N Action Refer to Trad Underwriting Action Unfavora ble Action Issue at Risk Class determined by UW Action Issue, but cap at Standard Risk Class Substandard Risk Refer to Trad UW OR Decline
Issues and Considerations Single Predictive Model Triage Actuarial If you desire multiple preferred classes, how do you differentiate? Usually based on non-medical preferred criteria, which often have little variation between Preferred Best and Preferred Could use predictive model risk score to differentiate between preferred classes What do you do if prescription profile details are not available for the applicant? Underwriting Obvious impacts to limiting to a few rate classes, but testing the waters is an opportunity Incorporating more driven by impacts to mortality and threshold for change Gap in prescription checks and what does that suggest varies in many of the approaches. A/E Mortality (as a percentage of 2015 VBT) All Ages With Any Rx Detail 80.6% Eligibility Only 83.1% No Hits 93.0% 9
Predictive Model, Non-Triage Standard Risk Predictive Model P P M M P M + Predictive Model + Score Cut off = X Very Favorable Score Action Issue as Pref Risk Life Insurance Application + Full Application + External Data (MVR, MIB, Rx) + Smoker? Y,N Unfavorable Score Action Action Issue at Standard Rate Substandard Risk Decline OR Issue a limited prod.
Issues and Considerations Predictive Model, Non-Triage Actuarial Potentially most efficient structure quick, with significant reduction in underwriting expense Implications of this efficiency Might need to price with higher levels of expected mortality, or May need to be prepared to reject a higher proportion of applicants, or Could develop a specially priced/loaded product for non-qualifiers, or Could face increased uncertainty on the profitability of the block 11
Evaluating the Performance of the Model Actuarial Should have some approach to understand the quality of risks qualifying through the use of the predictive model and its overall program Waiting for credible mortality results to emerge is not recommended Two primary approaches are developing Random Hold-Outs Post-Issue APSs What could you do when you identify a concern? Are your expectations on mark? Under or over Waiting for credible mortality the risks Approaches are developing Random Hold-Outs Post-Issue Rx checks Post-Issue APSs will action be warranted, or back-end file detail importance Legal support Underwriting 12
Random Hold-Outs Actuarial Pull ( hold out ) a certain percentage of applicants Closely monitor the number and type of adverse risks identified Incredibly valuable in learning lessons quickly Most agents are very frustrated by random hold-outs A simple must Underwriting Experience frustrations by both underwriters and agent Managing agent expectations. Tools for preparedness, FAQs, communication role-playing, communication training. Which do you use as a warning sign? Inclusion of cases where evidence might be completed 13
Post-Issue APSs Actuarial For a certain percentage of insureds issued through the model/program, obtain an APS after issue and attempt to evaluate the types of adverse risks not caught by the program parameters. Will likely require some additional authorizations. Adds extra expenses and underwriter involvement into a process specifically designed to reduce each. What actions could you take if you found something really bad? Active rescissions would require having asked an application question specifically related to the identified condition on which you would want to rescind 14
Actions to Remedy Problem Areas Actuarial Very important to begin with a focus on actively managing the program and making adjustments What actions can you take to remedy such a concern? As an example: Under-reporting of weight with a program that has eliminated paramed exams. Ask the applicant to submit a selfie with the application? Critical nature and impact of proof of concept Modify the inclusion of certain age bands Downstream opportunities identified and the impact on your other business segments Smokers amnesia Sentinel effect Underwriting Limit use to agent assisted sales 15
Questions 16