Predicting the Damage of Hull Car Insurance, Using Time Series Methods (Case Study: Parsian Insurance Company)

In Iranian insurance market, car insurance has the highest volume in terms of issuance and loss. So, the need for financial decisions and predicting the behavior of this market is imperative for insurers. Time series are powerful statistical methods that use statistical methods based on random processes to modelise and predict the behavior of such variables depend on time. In this paper, damage of hull car insurance (Parsian Insurance Company from March 1385 to July 1389) is considered as a variable, then by using two models time series, i.e. Box - Jenkins and Winters, the behavior of this variable is modelized. Since observations has seasonal trends, SARIMA and Winters multiplicative models are used. MAPE error indicator tools show that Winter’s multiplicative model has more accurate fitting. Finally, by using Winters model, damage of hull car insurance for a period of six months is predicted.