Claim Reserving: Performance Testing and the Control Cycle
By Yi Jing, Joseph R. Lebens, Stephen P. Lowe
Fundamentally, estimates of claim liabilities are forecasts subject to estimation errors. The actuary responsible for making the forecast must select and apply one or more actuarial projection methods, interpret the results, and apply judgment. Performance testing of an actuarial projection method can provide empirical evidence as to the inherent level of estimation error associated with its forecasts. Performance testing of alternative methods provides formal assurance that the actuary is using the best methods for the given circumstance, and also provides insight into the appropriate weight to give to the indications produced by each method. Performance testing is an integral part of the actuarial control cycle associated with the loss reserving process. It provides the necessary feedback loop to the actuary, assuring that he or she is not overconfident about his or her forecasts. This paper describes how to construct sound performance tests, consistent with statistical crossvalidation, within the reserving control cycle. It illustrates the application of the techniques via a case study, including some interesting empirical results.
KEYWORDS: Chain-ladder, claim reserving, control cycle, cross-validation, estimation error, forecast error, loss reserves, overconfidence, parameter risk, process risk, predictive skill, reserve accuracy, reserving process