Loss Reserving Methods for an Icelandic Workers' Compensation Portfolio: A Comparison of Chain Ladder, Bornhuetter-Ferguson and Benktander Method

This report addresses the challenge of estimating outstanding claims reserves for a workers’ compensation portfolio at Sjóvá-Almennar tryggingar hf, an Icelandic insurance company. The analysis covers accident years 2010 to 2025. Monetary figures have been modified to protect confidentiality. Multiple reserving methods are applied to both paid and incurred development triangles, the Chain Ladder, the Bornhuetter-Ferguson, the Benktander, and the Average Cost method. To assess the sensitivity of the results, the portfolio is evaluated under two inflation frameworks, the Wage Index and the Consumer Price Index, and under three outlier scenarios, including the full dataset and exclusions at the 99.5th and 95th percentiles. The results show that incurred models can produce negative Incurred But Not Reported (IBNR) estimates due to a systematic case reserve redundancy, making them unsuitable for balance-sheet provisioning. On a paid basis, the Wage Index adjustment provides a more realistic inflation structure for workers’ compensation than the Consumer Price Index. The Benktander method applied to the paid Wage Index triangle is selected as the best estimate, producing an outstanding reserve of 1,283 M. This result is cross validated by the Average Cost Benktander estimate of 1,300 M, a difference of only 1.3%. The convergence of two independent approaches, one based on aggregate payments and the other on frequency and severity components, provides a high degree of confidence in the reliability of this estimate.