Could someone please confirm whether I have understood the below paragraph from the core reading properly (chapter 15 - triangulation methods)? Sometimes we use the ultimate claims derived from an aggregate method (for example, chain ladder) to estimate the average claim amount for the ACPC method by dividing the ultimate by the estimated number of claims for the origin period. The averages for more developed origin years are then trended and used for more recent origin periods. Does this mean we'll calculate the ultimate claims per origin period (AY, UWY, RY, etc.) using one of the chain ladder methods, then divide that ultimate by the total number of reported claims from that origin period and use that figure (adjusted for inflationary differences) for all development periods corresponding to that origin year? For example, for accident year 2017 we have recorded 150 claims in total. Our aggregate ultimate claim amount by the chain-ladder is £15,000. Therefore, our average cost per claim over all claims relating to the 2017 accident year is £100. Our analysis shows a trend of 7% between 2017 and 2018 so we apply this trend to get an average cost per claim of £107 for the 2018 accident year. I can't understand what the point of this would be. Is it in the scenario where our most recent origin periods have very volatile data and hence don't give reasonable estimates under the standard ACPC method?
Hi Your example looks fine. For years that have not fully developed your average cost per claim is based on only complete claims, not on all claims since other claims are yet to come or develop entirely. So you go to a fully developed year, get the average cost per claim, adjust it for trends etc and get an estimate at full development of the average cost per claim for those years that are still developing. My understanding is there are a lot of ACPC methods and the one taught in earlier subjects is but one of many. The notes are just discussing a different approach.
