The Application of Composite Model in Fire Insurance Loss Data Modeling: A Case Study of Reinsurance Company Iqbal Hediananda Putra (a*), Siti Nurrohmah (b)
a) Faculty of Economics and Business, Universitas Indonesia
Depok, Indonesia
*iqbal.hediananda[at]sci.ui.ac.id
b) Faculty of Mathematics and Natural Sciences, Universitas Indonesia
Depok, Indonesia
Abstract
The loss model is important in insurance industry. The loss model should adequately represent underlying insurance risk. One of the commonly used loss models is the collective risk model. The collective risk model is based on frequency and severity distribution. The use of single distribution to model severity distribution often cannot represent both head and tail area of empirical data. One of solutions is by using different distribution on head and tail area referred as composite model. This paper aims to create a set of composite models- to estimate composite models parameters, and to compare the goodness of fit of composite models and single distribution-based models. The composite models are based on transformed beta family distribution and fire insurance loss data from one Reinsurance Company in Indonesia.
Keywords: insurance- loss model- composite model- transformed beta family distribution
