Adaptive reserving

Imagine this scenario: you are in charge of the regular valuation of insurance liabilities for a large general insurance company. Perhaps you are even the Appointed Actuary. Now if you enjoy watching an army of minions turning the handle on reserve estimates all year round (or enjoy doing a large part of it yourself) then stop reading. However, if you are interested in alternatives, then let us provide you with one.

For any large insurer with many lines of business, regular valuation of liabilities, while necessary, is often both an expensive and time consuming process. Actuaries must examine each line and change the reserving models as required - it is not simply a case of updating an array of spreadsheets.

However, is this effort needed? Most classes will show little change from one valuation period to the next, particularly if valuations are relatively frequent - quarterly valuations are common across many insurers. Only a few will need modification. Is it possible to have an automatic process to update most of the classes leaving only those few that require some degree of remodelling?

Adaptive reserving - a type of robotic reserving - is one way that the reserving process can be automated. This uses adaptive filters to automate the updating of reserving models from one valuation period to the next. Using objective statistical methods, filters can judge new experience in light of previous experience and adjust the models to the degree warranted by the most recent data.