Publication

Abstract : The process of estimating the accurate reserves for incurred but not reported (IBNR) claims is the important task performed by the actuaries in non-life insurance business. Chain ladder (CL) method is probably the most commonly used method in loss reserving. But this method is not derived from any fundamental theory about the way the claims occur. So that theoretical justification of this method is rather difficult. To overcome this, we introduced a modified method using a well defined non parametric model called Kernel Density Estimation (KDE) for estimating the IBNR claims reserves with the help of percentiles. In this paper, we developed a procedure for the modification of the CL predictors (outstanding claims estimates) for the future cumulative claims (lower triangle values) with the use of KDE and showed that our procedure provides better results compare to the existing CL method for finding the IBNR claims reserves in terms of the Mack’s standard error.