中文摘要：

由于可转换债券隐含期权具有的强路径依赖结构和它所承受的多种风险，蒙特卡罗模拟方法对它的定价具备一定优越性，其中，最小方差蒙特卡罗方法（LSM）又因为其简单性而受到重视。然而，面对近年来其他定价方法不断改进的挑战，可转换债券的LSM定价方法也有改进的必要。本文首先分析和评述了Rasmussen等针对美式期权LSM定价方法的改进，受到它们的启发，尝试对可转换债券LSM定价方法进行改良。实证结论显示，将Rasmussen式控制变量结合到可转换债券的LSM定价方法中，可以有效地减少其模拟方差。

英文摘要：

Because convertible bonds are subject to multi-risks and because of its implied option＇s strong path-dependence and structure, Monte-Carlo method has a stronger applicability for its pricing, and LSM is highly recommended for its convenience in implementation. However, because some superiorities of Monte Carlo method-challenged by some recent improvements of other methods, have been weakened, further improvement becomes also necessary for Monte-Carlo method for convertible bond pricing. The paper first analyzes and comments the improvements of LSM pricing method for American option proposed by Rasmussen, then suggests combining the improvements and importance sampling with LSM pricing method for convertible bond. The empirical study shows that combining Rasmussen-control variant with LSM approach could efficiently reduce the simulation variance in convertible bond pricing.