Μέθοδοι αποθεματοποίησης γενικών ασφαλειών : σύγκριση των micro και των macro προσεγγίσεων
In order to meet future liabilities, general insurance companies must set up reserves, in accordance with EU's directive Solvency II. Therefore, it is very important to predict future cash flows reliably, using appropriate actuarial loss reserving methods. In general, most methods are based on aggregated run-off triangles and present several important disadvantages. However, new method called "micro-level" method has been recently proposed, which take advantage of the detailed historical data set of the insurance companies, instead of relying exclusively on aggregated data. Thus, by fitting these detailed data to appropriately selected statistical models, the new method could provide the actuary with more detailed and valuable reserve calculations. In this thesis, the "micro-level" loss reserving method is presented and compared with the conventional aggregate run-off triangles. After justifying the need for more precise reserve calculations, conventional chain ladder techniques, the Bornhuetter-Ferguson method and generalized linear reserving methods are presented. Moreover, the recently proposed "microlevel" method is presented. Finally, the advantages of the new method are demonstrated through computer simulations using the well known statistics package R.