我國已於2002 年通過「金融資產證券化條例」，在金融機構積極規劃不動產抵押債權證券化商品之過程中，推行房貸保險制度已刻不容緩。本文針對國內二大銀行之房貸資料，進行違約及提前清償行為模型參數之估計及其影響因素之探討與模擬分析。實證發現，市場利率、貸款成數、償還壓力及總體經濟變數均對提前清償行為有顯著影響，負權益機率、貸款成數、償還壓力及總體經濟變數也對違約行為有顯著影響。模擬保費結果得知，借款者面對不同房貸保險費率結構之選擇，因自利行為存在逆選擇的問題，且保險機構在資訊不對稱之下，二者間存在明顯利益衝突。因此保險機構在房貸保費定價上，應將借款者之自利行為納入評價因子考量。

This study examines different structures of mortgage insurance premiums and simulates the premiums from the viewpoints of borrowers and the insurance companies. We further explore the existence of profit conflict and the phenomenon of adverse selection under information asymmetry. We collect mortgage data from two large banks in Taiwan and construct the competing risk model (CRM) to explore prepayment and default behaviors. Results show that interest rates, loan-to-value ratio, payment-to-income ratio and the unemployment and divorce rates are significant to prepayment behaviors. The market housing price, loan-to-value ratio, pay-to-income ratio, and the unemployment and divorce rates are important factors affecting default behaviors. Moreover, the simulation results of mortgage insurance premium indicate that the self-selection behavior exists under the information asymmetry, leading to the conflict of the insurance companies and borrowers. To avoid this issue, insurance companies may take into account of the self-selection behavior of borrowers in pricing mortgage premium.