|
|
 |
| |
|
|
|
Composite Distributions and their Associated Risk Measures for Auto-mobile Insurance Claims Data |
|
|
|
PP: 203-217 |
|
|
doi:10.18576/jsap/150204
|
|
|
|
Author(s) |
|
|
|
Williams Kumi,
Henry Otoo,
Charles Kwofie,
Sampson Takyi Appiah,
|
|
|
|
Abstract |
|
|
| Insurance losses comprises of small and large claims rendering single distributions incapable of holistically capturing the different sizes together accurately. Risk measures associated with insurance losses are crucial for determining reserve levels and for assessing solvency. Hence, in estimating risk measures, the right probabilistic distributions has to be carefully fitted in order not to underestimate or over estimate associated parameters. In view of this, this paper employs a two component composite distribution to describe automobile insurances losses from Ghana using 11,879 data points. This research fitted 240 composite distributions and results of the top ten are selected and presented based on some goodness of fit criteria. Threshold values and mixing weights for each composite distribution are also estimated and presented. Value at Risk and Tail value at Risk are then estimated and presented for the top ten composite distributions at 95% and 99% security levels. |
|
|
|
|
|
|
|
