基于远期合约的外汇风险管理：

基于远期合约的外汇风险管理：
多阶段随机规划应用探究
Exchange Rate Risk Management Using Forwards:
An Application of Multi-Stage Stochastic Programming
宋智铖
指导教师姓名：牛霖琳助理教授
专业名称：金融学
摘要
在国际贸易活动中，进出口企业通常需要根据贸易合约，在未来特定时期结算确定数额的现金账款。

这意味着外贸企业在实际结算日前，将面临外汇波动带来的汇兑风险。

因此，如何通过外汇远期合约进行外汇资产管理，以期在规避风险的同时，提高资产组合的期望收益，具有重要的现实意义。

本文利用外汇远期合约作为风险管理工具，在多阶段随机规划的框架下估计时变外汇资产配置比率。

具体而言，首先，最优化目标函数在传统的均值-方差模式的基础上设定为均值-广义半方差形式(Mean-GSV)，在最大化期望收益的同时最小化汇率波动可能带来的亏损。

其次，应用基于协整的VECM-BEKK模型对外汇的短期收益率和波动率进行预测，并在此基础上通过二叉树模型模拟随机规划方法所需要的随机情景。

最后，通过遗传算法求解最优化模型，找出在设定的随机情景中实现目标函数最大化的资产配置策略。

本文实证部分，针对美元-英镑07年02月至09年08月31个月的汇率数据进行外汇风险管理案例分析，通过滚动操作估计资产配置比率，计算外汇资产组合的实际收益率，并将结果与传统外汇风险管理方法得到的资产组合收益率进行比较。

虽然随机规划方法、VECM-BEKK模型以及Mean-GSV函数在已有的风险管理文献中都分别略有提及，并得到不同程度的应用与认可。

但本文首次将VECMBEKK模型和Mean-GSV函数纳入随机规划框架，使得模型求解得到的最佳动态资产配置比率既能根据汇率波动率变化作出调整，又能在最小化风险的同时最大化期望收益。

实证结果表明，在多阶段随机规划框架下，VECM-BEKK模型预测与Mean-GSV目标函数估计得到的动态对冲资产组合既能有效规避汇率波动的引起的风险，又能实现比传统方法更高的期望收益。

关键词：外汇风险；随机规划；外汇资产配置
Abstract
In the international trade activities, import and export enterprises often have to squarea certain amount of foreign currency in a pre-specified settlement date, which means the foreign trade companies would bear the exchange rate risk before the actual settlement.From this perspective, it is of practical significance to improve management on foreign exchange exposure management using forward contracts, so as to avoid risk and enhance the return rate of the hedged portfolio concurrently. In this paper, forward contracts are adopted to construct hedged portfolio under multi-stage stochastic programming (MSSP) framework. We estimate dynamic optimal hedge ratios with our model. Specifically, first of all, a mean-generalized semi-variance (Mean-GSV) optimal objective function, as an extension of mean-variance objective
form, is introduced to minimize the downside risk and maximize the expected return at the same time. Secondly, an integration based vector error correction (VEC) and BEKK-GARCH model are adopted to predict both the exchange rate mean and volatility. The famous CRR binomial tree model is included further to simulate the probability space of exchange rate market performance, which we prefer the term of scenariotree. And the random parameters produced by this scenario tree are required by MSSP method. Finally, genetic arithmetic (GA) is applied to approach the optimal numeric strategy, which is in fact an optimal dynamic hedge ratio sequence. In the empirical part, dollar versus sterling exchange rate market during the periods from Feb.2007 to Aug.2009 comesinto investigation. And a rollover framework is applied to estimate the optimal hedge ratio; real return for the hedged portfolio are calculated. Prevalent methodologies are included also for hedging effect comparison.
While stochastic programming, VECM-BEKK model and Mean-GSV objective function are all mentioned separately by literatures in the field of foreign exchange exposure management, this paper is the first to incorporate the VECM-BEKK model and Mean-GSV objective function under the MSSP framework. This model is formulated to solve the dynamic hedge ratio, which can not only adapt to the change of exchange rate volatility but also realize risk minimization and return maximization simultaneously. The empirical result shows that VECM-BEKK model and Mean-GSV function
benefit from working together under the MSSP framework. It really leads to lower downside risk and higher expected return rate.
Key Words: Foreign Exchange Exposure; Stochastic Programming; Foreign Exchange Management。