From Persistency Curves to a Retention Budget: Voluntary Policy Discontinuation in a Synthetic Indonesian Life Insurance Portfolio

Voluntary discontinuation of life insurance policies (lapse, surrender, non-renewal) erodes the value of an in-force book. This preprint studies it end to end on a fully synthetic portfolio of 40,920 policies issued between 2022 and 2025, built from seven linked tables with known planted effects and documented leakage traps. RQ1 describes where policies are lost (Kaplan-Meier persistency, cause-specific hazards, Cox model). RQ2 predicts voluntary exit in months 13 to 24 from signals available 30 days before the first anniversary, comparing an Explainable Boosting Machine with logistic regression and gradient boosting, with calibration and robustness checks. RQ3 turns calibrated scores into a profit-maximising contact rule, compares targeting strategies, quantifies sensitivity and sizes the randomised pilot needed to measure the save probability. All data are synthetic: no data from IFG Life or any other insurer were used, and no result is a finding about a real company. The archive contains the paper, the data with a data dictionary and ground-truth file, the code and the results, and regenerates every number in the paper with one script.