A competing risks dynamic hazard approach to investigate the insolvency outcomes of property-casualty insurers

Abstract

This study applies a competing risks approach and an event time dynamic estimation framework to identify the characteristics underlying different insolvency resolutions incurred to U.S. property-casualty insurers during 1998–2010. The estimated hazard model relates the time-varying probability of a specific insolvency outcome to insurers’ characteristics and macroeconomic conditions. The study finds that (i) the hazards for different insolvency outcomes are neither equal nor proportionate; (ii) the model for generic insolvency events and the models for outcome-specific insolvency events feature different significant factors, which are not due to random variations; and (iii) the outcome-specific insolvency models exhibit better forecast performance than the generic insolvency model within a five-year forecast horizon. The results of the study provide regulators with early warnings of financial distress, aid them in prioritising troubled insurers and identifying the areas most likely to reveal material problems, and inform about the interventions that should be taken under specific circumstances.