EMPIRICAL-EVIDENCE FOR ALTERNATIVE LOSS DEVELOPMENT HORIZONS AND THE MEASUREMENT OF RESERVE ERROR

This article examines the sufficiency of alternative loss development horizons in measuring reserve errors and the precision of reserve error measurements taken over the loss development horizon. We find that no single development horizon is adequate in all circumstances. Short development horizons are adequate for tests of significance involving the aggregate error in a sample of insurers. Substantially longer developments must be used when precise measurements of error, taken on individual insurers, are required. This study also provides evidence of statistically significant reserving errors during the years 1977 through 1987.