上海交通大学硕士学位论文 1权证发行商发行权证风险对冲研究 摘 要 权证发行商发行权证风险对冲研究 摘 要 本文考察权证发行商发行权证的风险，及对冲风险的方法。 首先，为了更好的描述金融资产收益分布，引入随机波动率模型，介绍模型的估计方法。然后把随机波动率模型引入到期权定价理论中，简要回顾各种随机波动率模型下的期权定价方法和特点。 其次考虑发行备兑权证及其动态对冲， 动态对冲的方法就是保持 delta 中性，通过对 delta 中性组合的风险收益分析，发现间断交易所带来的误差服从卡方分布；通过一个日历价差组合的例子，指出“卖出波动率”并不是很明晰的概念，此组合做多 gamma， 做空 vega。 由于使用不同于实际波动率的波动率计算的 delta买卖标的资产对冲会导致复制误差，所以考察了不同的波动率（隐含波动率、实际波动率）计算 delta 对冲对最后收益的影响。由此引入本文的假设模型，将来波动率在两个极端值（最大值和最小值）之间，为规避复制误差带来的风险，权证发行商应采用最大值计算 delta 买卖标的资产对冲。在随机波动率环境下，发行商可以根据风险承受能力， 使用给定置信水平下的最大波动率计算 delta 买卖标的资产对冲，使收益 VaR 在给定置信水平下最小。最后，通过对上证指数的算例分析，同自回归条件异方差模型的预测波动率计算 delta 买卖标的资产对冲的结果比较，此策略避免了较大的损失。对最近香港权证市场的历史、制度、估值进行分析，并指出权证发行商在不同市场状况下的盈利模式。 最后考虑奇异期权及其静态对冲。由于静态对冲相对于动态对冲的诸多优点，所以考察奇异期权在随机波动率下的静态对冲。本文重点考虑障碍期权，首先考虑欧式 0-1 期权的风险收益特征，考虑美式 0-1 期权的风险收益特征，指出其更像是对于时间的权力，而不是价格的权力。考虑障碍期权各个风险参数的变化，指出其 delta 的不连续性，价格对于不同的波动率表现出凹性等不同于普通期权的特征。通过引入认购和认沽期权的对称关系（Put-Call Symmetry） ，可以将障碍期权可以分解为普通期权和美式 0-1 期权的组合。 通过对港交所一个交易上海交通大学硕士学位论文 2比较活跃的牛证 6476 的分析，结果表明分解后的组合是有效的，静态对冲策略完全可行。同时由于路径依赖期权对于波动率非常敏感，所以必须建立模型描述隐含波动率的微笑和偏斜，由于 SABR 模型能够很好地刻画了微笑的动态过程，并且符合市场中的隐含波动率曲线，提供计算市场风险的办法，包括波动率的一阶和二阶风险，所以选择建立 SABR 模型描述资产价格分布。可得出在 SABR模型下的修正后的普通期权组合。通过对 2007 年 11 月 21 日以恒生指数为标的的香港权证/期权市场的分析，得出微笑和偏斜依赖于到期期限，而且到期期限对于微笑和偏斜的影响是非常大的，同时在 SABR 下给出了考虑标的资产收益 分布存在偏斜的障碍期权的静态对冲方法。 关键词：关键词：随机波动率，SABR 模型，动态对冲，障碍期权，静态对冲 上海交通大学硕士学位论文 3RESEARCH ON WARRANT ISSUERS HEDGING STRATEGY ABSTRACT The paper investigates warrant issuers risk due to warrants issuing and his hedging strategy. First, in order to better model the financial asset return distribution, introduce the stochastic volatility model, give model estimation method. Then introduce the stochastic volatility model into the option pricing theory, give a brief review of various stochastic volatility model of option pricing methods and characteristics. Secondly, consider covered warrants and dynamic hedging. Dynamic hedging is maintaining delta neutral, through analysis of return and risk of the delta neutral portfolio, we find that the hedging error caused by discrete hedging follows chi distribution, and by a calendar spread example, pointing out that “selling volatility“ is not a very clear concept, for this portfolio long gamma and short vega. As the volatility to calculate delta for delta hedging is different from actual volatility, it will lead to replication errors, investigates the affection of return distribution using the different volatilities (implied volatilities, actual volatilities) for delta hedging. This paper introduced our model, assuming the volatility lies between two extreme values (maximum and minimum), to avoid the risk of replication error, warrants issuer should use the maximum volatility to calculate delta for delta hedging. In stochastic volatility environment, publishers can choose different volatility according to his risk tolerance, use the maximum volatility for delta hedging at a given confidence level, to achieve the smallest VaR at the given confidence level. Finally, through analysis of examples with SH index, comparing to result of delta hedging with forecast volatility using GARCH model, the strategy avoid a greater loss. For the domestic brokers, this strategy is also very easy to reach, they can estimate GARCH (1,1) model using historical data, find maximum volatility according to given confidence level. Analyze the recent Hong Kong warrants market, and point out different profit mode under different market conditions. Finally consider exotic options and static hedging. Since static hedging have many advantages relative to dynamic hedging, so we investigate static hedge for 上海交通大学硕士学位论文 4exotic options with stochastic volatilities. This paper focus on barrier options, first consider the risk and profit characteristics of 0-1 options, then consider the risk and profit characteristics of american 0-1 options, pointing out that it is more like option of the time, rather than option of price. Consider the behavior of risk parameter for barrier options, pointing out that the delta isnt continuity, many features that are different from the characteristics of vanilla options. By introducing Put-Call symmetry, barrier options can be decomposed into portfolio of vanilla options and american 0-1 options. Through analysis of warrant 6476, an active trading callable bull contract in HKEx, results show that the decomposition is valid, static hedging strategy is workable. Since the path-dependent options are very sensitive to volatility, we must build model to describe the skew and smile. As SABR model is a good model to capture the dynamic process of skew and smile, and in line with implied volatility curve in the market, so it provide an easy way to compute market risks, including first-order and second-order risks of volatility. Therefore, we use SABR model to model the asset return distribution, the revised composition of the vanilla options can be drawn under the SABR model. Through analysis of HSI index warrants / options at November 21, 2007, results sho