This page is free and needs no sign-up. It reproduces material from The Life Insurance Underwriter by Julian R. Sterling so you can print it or keep it open while you work.
To print it, use your browser's print button. Figures are as stated in the book; check the official source before relying on any of them.
Tool 5: The accelerated program review check-list
An accelerated program replaces exam evidence with data and a predictive model, and a review asks whether the replacement still holds. The check-list below is general practice, and Chapters 10, 11 and 13 give the rules and methods. Take the figures from your own program; the Northgate figures are a worked line.
- Referral rate. How many applications are sent to full underwriting, and why? A rate that drifts up or down says the data or the model is changing. In the case, 41 percent of 23,500 applications were referred.
- Decisions and placement. How many accelerated decisions became policies in force? In the case, 13,865 decisions led to 12,478 policies issued and placed, which is 90 percent.
- Class mix. Does the mix of issued classes match the program's design? Northgate's mix is Preferred Plus 36 percent, Preferred 34 percent, Standard Plus 22 percent and Standard 8 percent, with a weighted mortality multiple of 74.9 percent of standard.
- Random post-issue audit. Take a random sample, collect full evidence after issue and compare the issued class with the class full underwriting would give. In the case, 1,000 policies were audited: 820 confirmed (82.0 percent), 155 in a better class than warranted (15.5 percent) and 25 in a worse class (2.5 percent).
- Slippage. Compare the audited and issued mortality multiples. In the case, 74.9 percent as issued and 77.8 percent as audited, a mortality slippage of 3.9 percent.
- Economics. Weigh evidence saved against the cost of slippage. In the case, the saving is $3,119,625 and the cost of slippage is $3,890,960, so the slippage costs more by $771,335. Faster decisions also raise placement, which the case does not value.
- Actual to expected. Compare actual deaths with expected deaths on your table. In the case, 11 actual against 9.1 expected is 121 percent by count, and by amount $4,960,000 against $3,822,866 is 130 percent.
- Credibility. State how much weight the result can carry. In the case, 11 deaths give a credibility factor of 0.10 and a blended estimate of 102 percent, so the study cannot prove the program sound or unsound, and the audit is the stronger signal.
- Contestable claims. Read every early death with the claims director. In the case, all 11 deaths fell inside the two-year contestable period, material misrepresentation was found in 3, and those policies were rescinded with premiums refunded. Without them, the count ratio is 8 over 9.1, or 88 percent.
- Governance file. Keep one place where the model version, the data sources, the audit design, the committee decisions and the compliance officer's review are recorded, with dates. The rules on governing data and models are in Chapter 11, and each state writes its own, so work with compliance and counsel on the current text.
Close the review with a short list of decisions for the committee, each with an owner and a date. In the case, the committee reviews the study on 15 October 2026, and the plan applies its changes to applications signed from 1 January 2027, with the next study at the 30 September 2027 cut-off.