On the Analysis of the Interaction Effects in Casualty Insurance

[[abstract]]現今汽車保險之保險費率釐訂大多採取從人因素及從車因素為原則，以確保保費之合理。但是從分析結果看來，發現變數間存在交互效果，可能破壞費率之公平原則。本研究之目的是為了瞭解，目前市場上的汽車任意第三人責任保險之風險分類是否具有適當的分類和公平性。所以利用對數線性模式之獨立模型及條件獨立模型探討變數間關係，並且利用Commom Odds Ratio和Cochran Mantel Haenszel Test去了解各層之間反應變數與解釋變數是否有關係；同時，本研究利用三維列聯表對變數做比較，結果顯示：(1)在不同分層下，變數之間有相關性，如在性別分層下，三年內出險次數和保險種類有相關性；(2)在控制不同變數下，變數之間存在交互效果，如在控制性別變數下，三年內出險次數和保險種類有交互作用。 以上之敘述，我們可知目前的任意汽車第三人責任保險之風險分類可能有失適當性，在費率的釐訂上可能導致公平性不足。[[abstract]]To ensure the equity of the premium the charge of automobile insurance is currently based on two principles: personal factors and car factors In our analysis we find that some interaction effects between variables exist It may lack consistency to the principle of equity and justice Thus we investigate whether the risk classification of commercial automobile liability insurance in the market employs an appropriate classification We also use Loglinear Model and CMH Test to examine the association between the response variable and the explanatory variables We intend to use three-way contingency table for comparison The results indicate that they have correlated relationships and some interaction effects between variables in different stratifications In summary we suggest that the current risk classification of commercial automobile liability insurance are not equitable in charging premium and are not classified adequatel