Author: Song, Yunah(Korea Insurance Research Institute)

This study examines how deteriorating credit conditions affect insurance lapse, using de-identified administrative data from the Korea Credit Information Services (KCIS). I construct a panel of approximately 22.6 million contract-months by tracking 733,779 insurance contracts held by 152,362 policyholders at monthly frequency for up to 37 months. On this panel, I estimate a discrete-time complementary log-log survival model combined with a Mundlak (1978) correlated random-effects approach to separate the within- and between-individual effects of the credit variables. Non-bank loans and the delinquent amount significantly raise the lapse hazard, and the between-individual effect substantially exceeds the within-individual effect. This indicates that credit conditions influence lapse more through structural differences in financial vulnerability across policyholders than through transitory within-individual fluctuations. It suggests that, alongside ex-post liquidity support, suitability management at the point of sale and product design aligned with policyholders' premium-payment capacity are needed. Heterogeneity analysis shows that the lapse-inducing effect of the delinquent amount disappears among policyholders aged 60 and over, for whom the opportunity cost of lapse is high; that the lapse response to credit deterioration is weakest in the TM·Online channel; and that savings-type insurance responds to increases in borrowing regardless of its source.