Granular loss reserving methods for individual claim-by-claim data will be presented. Since there is a growing demand for prediction of total reserves for different types of claims or even multiple lines of business and there are possible structural breaks present in the historical data, time-varying frameworks embracing dependencies will be established.
The presented micro model for reserving provides model diagnostics and checking of the assumptions directly from data. We are in close collaboration with the Czech Insurers‘ Bureau. It allows us to use the database from its Guarantee Fund for car insurance. The data consist of claims developments from the beginning of 2000 and are continuously updated.