Policy cancellation remains a significant risk to life insurer solvency in digitally mediated markets. This study aimed to model and forecast lapse-driven solvency erosion using behavioral, institutional, and macroeconomic predictors structured into a unified econometric cascade. The study analysed 1559,661 policyholder records across 880 firm-quarter observations from 11 Chinese life insurers (2013–2023). Behavioural metrics (entropy, latency, notification fatigue) were derived from weekly user logs. Panel GMM, SVAR, Cox models, and regime-switching threshold regressions were implemented in Stata SE 18.0. Models were evaluated via log-likelihood, AIC/BIC, Wald tests, impulse response functions, and forecast error variance decomposition. Entropy (HR = 1.44), latency (HR = 1.27), and notification fatigue (HR = 1.52) significantly predicted lapse hazard. Lapse rates rose from 4.98 % to 8.47 % across CRI tertiles. Interaction terms (NFI × ACR, HR = 1.62) intensified risk. In GMM, CRI had a marginal solvency effect of 0.124; reserve mismatch and lapse rate had an adverse impact (–0.112, –0.087). SVAR attributed 42.1 % of solvency variance to CRI shocks; IRF peaked at quarter 4 (IRF = 0.056, p = 0.0034). A CRI threshold of 0.56 yielded a post-threshold reversal (β = –0.064, p = 0.0043). Predictive AUC = 0.772 with 84.3% TPR and 42-day median lead time. Behavioral metrics embedded in digital platforms enable early detection of solvency risk and provide intervention windows.
扫码关注我们
求助内容:
应助结果提醒方式:
