复杂衍生品定价的模型风险度量——以中信泰富杠杆式外汇合约为例

Measuring Model Risk in Pricing Complex Derivatives: Case Study of the Leveraged Foreign Exchange Trading Contract of Citic-Pacific

随着场外衍生品的日益复杂化,定价的模型风险越来越受到重视。标的资产随机过程的设定是衍生品定价的关键环节,但随机过程选择与估计样本选择尚无公认的标准,这成为衍生品定价模型风险的主要来源。本文以中信泰富杠杆式外汇合约为例,根据已有文献选择了几何布朗运动、OU过程、Schwartz单因素模型来描述汇率走势,运用不同区间和频率的样本估计这3个随机过程的参数,进一步采用蒙特卡洛模拟法得到此合约的价格分布,使用业界普遍使用的"最坏情况"指标来度量买卖双方承担的模型风险。实证结果表明,随机过程设定导致的模型风险显著存在,模型风险会随着随机过程与估计样本的变化而变化,衍生品买卖双方的模型风险呈现非对称性特征,买方承担的模型风险显著高于卖方,这反映了模型风险管理的重要意义及非专业买方参与场外复杂衍生品交易时的困境。

With lhe increasing complexity of financial derivatives, more and more attention has been paid to model risk in pricing. Specifying the stochastic process of underlying asset is a crucial step in pricing complex derivatives but the true stochastic model and appropriate sample are unknown. These uncertainties are the main sources of model risk. This paper takes the leveraged foreign exchange contract signed by Citic-Pacific as an example and measures its model risk deriving from specification of stochastic process with Worst Situation Case＇of buy side and sell side. Samples with different range and frequency are adopted to estimate Geometric Brownian Motion process, OU process, and Schwartz one factor mod- el. The results show that model risk does exist. The choice of stochastic model and sample will significantly affect the magnitude of model risk. The model risk exhibits asymmetric pattern among the two sides, namely, buy side undertakeslarger model risk than the one of sell side. It has implications on model risk management and the dilemma that non-so- phisticate investors participate in complex derivative trading.