Large claims in non-life insurance

Abstract: It is of outmost importance for an insurance company to apply a fair pricing policy. If the price is too high, valuable customers are lost to other insurance companies while if it’s too low – it nets a negative profit. To achieve a good pricing policy, information regarding claim size history for a given type of customer is required. A problem arises as large extremal events occur and affects the claim size data. These extremal events take shape in individually large claim sizes that by themselves can alter the distribution for what certain groups of individuals are expected to cost. A remedy for this is to apply what is called a large claim limit. Any claim exceeding this limit is thought of as being outside the scope of what is captured by the original distribution of the claim size. These exceeding claims are treated separately and have their cost distributed across all insurance takers, rather than just the group they belong to. So, where exactly do you draw this limit? Do you treat the entire claim size this way (exclusion) or just the bit that is exceeding the threshold (truncation)? These questions are treated and answered in this master’s thesis for Trygg-Hansa. For each product code, a limit was achieved in addition to which method for exceeding data that was best to use.