数学杂志  2018, Vol. 38 Issue (5): 912-920   PDF    
扩展功能
加入收藏夹
复制引文信息
加入引用管理器
Email Alert
RSS
本文作者相关文章
韦才敏
林先伟
范衠
分数布朗运动下带交易费用和红利的两值期权定价
韦才敏1,2, 林先伟1,2, 范衠2    
1. 汕头大学数学系, 广东 汕头 515063;
2. 汕头大学数字信号与图像处理技术重点实验室, 广东 汕头 515063
摘要:本文研究了在分数布朗运动环境下带交易费用和红利的两值期权定价问题.在标的资产服从几何分数布朗运动的情况下,利用分数Itô公式和无风险套利原理建立了分数布朗运动环境下带交易费用和红利的两值期权的定价模型.再通过用偏微分方程的方法进行求解此定价模型,得到了在分数布朗运动下带交易费用和红利的两值期权定价公式.所得结果推广了已有结论.
关键词两值期权    期权定价    无风险套利原则    交易成本    分数布朗运动    
BINARY OPTION PRICING WITH TRANSACTION COSTS AND DIVIDENDS IN A FRACTIONAL BROWNIAN MOTION ENVIRONMENT
WEI Cai-min1,2, LIN Xian-wei1,2, FAN Zhun2    
1. Department of Mathematics, Shantou University, Shantou 515063, China;
2. Guangdong Provincial Key Lab of Digital Signals and Image Processing, Shantou University, Shantou 515063, China
Abstract: This paper deals with the problem of pricing Binary option with transaction costs and dividends under the fractional Brownian motion. Suppose that the stock price follows geometric fractional Brownian motion, by using fractional Itô formula and no-arbitrage principle, we establish the binary option pricing model with transaction costs and dividends. The method of partial differential equation are used to solve this model, and then we get the pricing formula of the binary option with transaction costs and dividends in a fractional Brownian motion environment, which extends the previous conclusions.
Key words: binary option     option pricing     no-arbitrage principle     transaction costs     fractional Brownian motion    
1 引言

期权定价的研究是数理金融的核心内容之一.自从1973年著名的Black-Scholes模型和结果发表后, 期权定价理论和应用得到了迅速发展[1].但是经典Black-Scholes模型过于理想化, 与实际存在很大程度的摩擦.而分数布朗运动具有金融市场所需要的长时间相关性和自相似性等特征, 其已经成为描述标的资产价格过程的一个有力工具. 1989年Peter[2]提出了将资产价格的变化用分数布朗运动来刻画, 并论证了资产价格服从几何分数布朗运动则其收益率服从分形分布; Ducan[3]研究了关于分数布朗运动的随机积分理论; Elliot和Hoke[4]研究了在Hurst指数在$H\in(\frac 12, 1)$情况下的分数布朗运动, 他们通过Wick积的方法得到了Girsanov定理和分数Itô公式; Hu和Øksendal[5]通过Wick积分和分数白噪声分析进一步发展了分数布朗运动积分理论, 证明了Itô型分数Black-Scholes市场无套利且完备的; Christian和Bender[6]将其推广到任意Hurset指数.

交易成本对于期权定价来说是一个重要的因素, 许多学者对此做了很多研究. Leland[7]提出了将修正的波动率应用在解决有交易费用的Black-Scholes模型的对冲误差; Gaussion[8]证明了在分数布朗运动下任何正的成比例交易成本大小能够消除套利机会. Zhang和Pan[9]给出了分数布朗运动模型下带交易费用和红利的亚式期权定价公式. Wang[10]解决了在分数布朗运动下离散时间的含有交易费用期权定价问题.此外, 随着交易市场的不断成熟, 由标准的看涨期权和看跌期权衍生出的非标准化产品日益增多, 这些产品称之为新型期权.两值期权就是新型期权的一种.目前关于两值期权的研究有: Thavaneswaran等人[11]用模糊理论研究两值期权的定价问题; 袁国军[12]在半离散化CEV过程中得到两值期权价格的差分格式; Hofer和Leitner[13]研究了两值期权的相对定价问题; 吴云和何建敏[14]推导了两值期权的解析解并阐述了二叉树方法在两值期权定价中的应用; 孙天宇[15]研究了标准布朗运动下带交易费用和红利的两值期权定价问题.然而, 在标准布朗运动下研究两值期权定价, 并不符合实际的金融市场.

本文在上述研究的基础上, 将标准布朗运动下带交易费用和红利的两值期权定价问题推广到Hurst指数为$H\in(\frac 12, 1)$的分数布朗运动更一般的情况.假设标的资产服从几何分数布朗运动, 通过无风险套利原则和分数Itô公式建立了分数布朗运动环境下两值期权的定价模型.利用偏微分方程的相关知识求解此模型, 得到了现金或无值看涨期权(CONC)和资产或无值看涨期权(AONC)的定价公式, 并由此推出了现金或无值看跌期权(CONP)和资产或无值看跌期权(AONP)的定价公式.

2 基本模型

两值期权(binary option)是合同条款变化产生的新型期权, 具有不连续收益的特点[16].一般分为两种类型:

(1) 现金或无值看涨期权(cash-or-nothing call) (简写为CONC):在到期日, 若股票价格低于执行价格, 则期权价值为零; 若大于执行价格, 则按规定支付现金1元.

(2) 资产或无值看涨期权(asset-or-nothing call) (简写为AONC):在到期日, 若股票价格低于执行价格, 则期权价值为零; 若大于执行价格, 则按规定支付股价.

引理1  (分数Itô公式[9])假设$V=V(S_t, t)$是一个关于$S_t$$t$的二元函数, 随机过程$S_t$满足以下随机微分方程

$ dS_t=\mu_t S_t dt+\sigma_tS_t dB_{H}(t), $

从而得到

$ dV=(\frac{\partial V}{\partial t}+\mu_t S_t \frac {\partial V}{\partial S_t}+H \sigma_t^2 t^{2H-1} S_t^2\frac {\partial^2V}{\partial S_t^2})dt+\sigma_t S_t\frac {\partial V}{\partial S_t}dB_{H}(t), $

其中$\mu_t$, $\sigma_t$分别表示漂移项和扩散项, $ B_{H}(t)=\left\{{B_H(t), t \geq 0}\right\}$是一个带有Hurst指数$H (0<H<1)$的分数布朗运动.

定义1[10] 假设$(\Omega, \digamma, P)$是一个完备的概率空间, Hurst指数为$H (0<H<1)$的分数布朗运动$B_{H}(t)=\left\{{B_H(t), t\geq 0}\right\}$是一个连续的高斯过程, 且满足

(1) $B_{H}(0)=E(B_{H}(t))=0$;

(2) $E_{p}(B_{H}(t)B_{H}(S))=\frac 12(|t|^{2H}+|S|^{2H}-|t-S|^{2H}).$

$H> \frac 12$时, $B_{H}(t)$具有长期相关性, 即对$r(n)=E(B_H(1)(B_{H}(n+1)-B_{H}(n))>0$, 并且$\sum\limits_{i=0}^n r(n)=\infty.$ Hu和Øksendal[5]$H> \frac 12$时通过Wick积和分数白噪声理论证明了Itô型分数Black-Scholes市场无套利且完备的.

本文仅讨论$H>\frac 12$的情况.现设$(\Omega, \digamma, \digamma_t, P)$是一个具有$\sigma$-流的完备的概率空间. $F_t$是由$B_{H}(t)$生成的$\sigma$-流, $P$表示风险中性测度.现在考虑Itô型分数Black-Scholes市场有两种资产, 设无风险资产(如债券)价格$M_t$满足下面方程$ dM_t=r_tM_tdt, $其中$r_t$表示无风险利率.

风险资产(如股票)价格$S_t$满足下列随机微分方程$ dS_t=(\mu_t-q_t)S_t dt+\sigma_tS_t dB_{H}(t), $其中$\mu_t, q_t, \sigma_t$分别表示预期收益率, 红利率和波动率, $ B_{H}(t)=\left\{{B_H(t), t \geq 0}\right\}$是一个带有Hurst指数$H (0<H<1)$的分数布朗运动.

考虑标的资产支付红利, 红利率为$q_t$, 到期日为时间$T$, 敲定价格为$K$, 作以下假设

(1) 假设标的资产(股票)价格$S_t$服从几何分数布朗运动

$ dS_t=(\mu_t-q_t)S_t dt+\sigma_tS_t dB_{H}(t), $ (2.1)

这里$\mu_t, q_t, \sigma_t$分别表示预期收益率, 红利率和波动率, 它们都是关于$t$的已知函数, $ B_{H}(t)=\left\{{B_H(t), t \geq 0}\right\}$是一个带有Hurst指数$H\in (\frac 12, 1)$的分数布朗运动;

(2) 没有税收, 允许卖空;

(3) 没有无风险套利机会;

(4) 投资组合的预期收益率等于无风险利率;

(5) 投资组合被每个$\delta t$修正, 其中$\delta t$是有限的, 固定的, 小的时间间隔;

(6) 有成比例交易成本.设$k$表示每单位股价的双向交易成本.假设以价格$S_t$买入$(\nu_t >0)$或卖出$(\nu_t <0)$ $\nu_t$份股票, 那么买入或卖出的交易成本为$\frac k2 |\nu_t|S_t$, 其中$k$为常数.

$V=V(S_t, t)$表示CONC (或AONC)在时刻$t$的价格.构造一个投资组合:一份CONC (或AONC)多头, $\Delta_t$份标的资产空头.在时刻$t$投资组合的价值为$\Pi_t=V-\Delta_t S_t$.应用分数Itô公式, 在$[t, t+\delta t]$时间段内,

$ \begin{eqnarray*}\delta \Pi_t&=&\delta V_t-\Delta_t\delta S_t-q_t\Delta S_t \delta_t-\frac k2|\nu_t|S_{t+\delta t} \\&=&(\frac{\partial V}{\partial t}+\mu_t S_t \frac {\partial V}{\partial S_t}+H \sigma_t^2 t^{2H-1} S_t^2\frac {\partial^2V}{\partial S_t^2}-\Delta_t S_t \mu_t -\Delta_t S_t q_t)\delta t \\&&+(\sigma_t S_t\frac {\partial V}{\partial S_t}-\Delta_t \sigma_t S_t) \delta B_{H}(t)-\frac k2|\nu_t|S_{t+\delta t}, \end{eqnarray*} $

其中$\nu_t=\Delta_{t+\delta t}-\Delta_t$.

为了使投资组合$\Pi_t$$[t, t+\delta t]$无风险, 取$\Delta_t=\frac {\partial V}{\partial S_t}$, 从而可得

$ \delta \Pi_t=(\frac {\partial V}{\partial t}+Ht^{2H-1}\sigma_t^2 S_t^2 \frac {\partial^2V}{\partial S_t^2}-q_tS_t \frac {\partial V}{\partial S_t} )\delta t-\frac k2|\nu_t|S_{t+\delta t}, $

其中

$ \nu_t=\Delta_{t+\delta t}-\Delta_t=\frac {\partial V}{\partial S_{t+\delta t}}-\frac {\partial V}{\partial S_t}=\frac {\partial ^2V}{\partial S_t^2}\sigma_t S_t\delta B_H(t)+o(\delta t). $

由于股票价格$S_t$满足(2.1)式, 则得

$ S_{t+\delta t}=S_t+\delta S_t=S_t+\sigma_t S_t\delta B_H(t)+o(\delta t). $

因此, 可得交易费用如下

$ \begin{array}{lll}E[\frac k2|\nu_t|S_{t+\delta t}]&=\frac k2 |\frac {\partial^2V}{\partial S_t^2}|\sigma_t S_tE(|\delta B_{H}(t)|S_{t+\delta t})+o(\delta t) \\&=\frac k2 \sqrt{\frac 2 \pi} \sigma_t S_t^2|\frac {\partial^2V}{\partial S_t^2}|(\delta t)^H+o(\delta t). \end{array} $

忽略高阶项$o(\delta t)$, 有

$ E[\frac k2|\nu_t|S_{t+\delta t}]=\frac k2 \sqrt{\frac 2 \pi} \sigma_t S_t^2|\frac {\partial^2V}{\partial S_t^2}|(\delta t)^H. $

因此

$ E[\delta \Pi_t]=(\frac {\partial V}{\partial t}+Ht^{2H-1}\sigma_t^2 S_t^2 \frac {\partial^2V}{\partial S_t^2}-q_tS_t \frac {\partial V}{\partial S_t} )\delta t-\frac k2 \sqrt{\frac 2 \pi} \sigma_t S_t^2|\frac {\partial^2V}{\partial S_t^2}|(\delta t)^H. $ (2.2)

由假设(4), 得到

$ E[\delta \Pi_t]=r_t\Pi_t \delta_t. $ (2.3)

将(2.2)式代入(2.3)式, 得到

$ \frac {\partial V}{\partial t}+Ht^{2H-1}\sigma_t^2S_t^2 \frac {\partial^2V}{\partial S_t^2}+(r_t-q_t)S_t\frac {\partial V}{\partial S_t}-r_tV-\frac k2 \sqrt{\frac 2 \pi} \sigma_t S_t^2|\frac {\partial^2V}{\partial S_t^2}|(\delta t)^{H-1}=0. $

${\rm Le}(H)=\frac {k}{\sigma_t (\delta t)^{1-H}}\sqrt{\frac 2\pi}$, 则有

$ \frac {\partial V}{\partial t}+Ht^{2H-1}\sigma_t^2S_t^2 \frac {\partial^2V}{\partial S_t^2}+(r_t-q_t)S_t\frac {\partial V}{\partial S_t}-r_tV-\frac {\sigma_t^2}{2} S_t^2|\frac {\partial^2V}{\partial S_t^2}|{\rm Le}(H)=0. $ (2.4)

将(2.4)重写如下

$ \frac {\partial V}{\partial t}+\frac 12 \hat{\sigma_t}^2 S_t^2 \frac {\partial^2V}{\partial S_t^2}+(r_t-q_t)S_t\frac {\partial V}{\partial S_t}-r_tV=0, $ (2.5)

其中

$ {\hat{\sigma_t}}^2=\sigma_t^2(2Ht^{2H-1}-{\rm Le}(H)\mbox{sign}(\Gamma)). $ (2.6)

注1

$ {\rm Le}(H)=\frac {k}{\sigma_t (\delta t)^{1-H}}\sqrt{\frac 2\pi} $

称为分数次Leland数.

注2  对于做空的单个欧式两值期权, 也可以得到(2.5)式, 若修正波动率如下

$ {\hat{\sigma_t}}^2=\sigma_t^2(2Ht^{2H-1}+{\rm Le}(H)\mbox{sign}(\Gamma)). $ (2.7)

注3  对于做多的单个欧式两值期权, 到期日的收益为$(S_T-K)^+$$(K-S_T)^+$.由于它们是凸函数, 所以$\Gamma>0$.然而, 对于做空的单个欧式两值期权, 到期日的收益为$-(S_T-K)^+$$-(K-S_T)^+$.它们是凹函数, 所以$\Gamma<0$.因此, 对于单个的欧式两值期权, (2.6)和(2.7)式能够做如下表示

$ {\hat{\sigma_t}}^2=\sigma_t^2(2Ht^{2H-1}-{\rm Le}(H)). $

从而得到在分数布朗运动下带交易费用和红利的两值期权定价模型如下

$ \left\{ \begin{array}{ll} \dfrac {\partial V}{\partial t}+\dfrac 12 \hat{\sigma_t}^2 S_t^2 \dfrac {\partial^2V}{\partial S_t^2}+(r_t-q_t)S_t\dfrac {\partial V}{\partial S_t}-r_tV=0, \\ V|_{t=T}=\left\{\begin{array}{ll} H^*(S_t-K) \;\;\;\;\;\;\; \mbox{(CONC)}, \\ S_tH^*(S_t-K)\;\;\;\; \mbox{(AONC)}, \end{array} \right.\end{array}\right. $ (2.8)

这里$H^*(\xi)$是Heviside函数.如果$\xi \geq 0$, 那么$H^*(\xi)=1$.否则, $H^*(\xi)=0$.

3 两值期权定价公式
3.1 现金或无值期权定价公式

定理1  假设股票价格满足(2.1)式, 在时刻$t$带有交易费用和红利的现金或无值看涨期权的定价公式为

$ V_{CC}(S_t, t)=N(\frac {\ln \frac {S_t}{K}+\displaystyle\int_{t}^{T}{(r_\theta-q_\theta-\frac 12 \hat{\sigma_\theta}^2)d\theta}}{\sqrt{\displaystyle\int_{t}^{T}{\sigma_\theta^2(2H \theta^{2H-1}-Le(H))d\theta}}})\exp(-\displaystyle\int_{t}^{T} r_\theta d\theta). $

  由方程组(2.8)可以得到现金或无值看涨期权定价模型如下

$ \left\{ {\begin{array}{*{20}{l}} {\dfrac {\partial V}{\partial t}+\dfrac 12 \hat{\sigma_t}^2 S_t^2 \dfrac {\partial^2V}{\partial S_t^2}+(r_t-q_t)S_t\dfrac {\partial V}{\partial S_t}-r_tV=0}, \\ {V|_{t=T}=H^*(S_t-K)}. \end{array}} \right. $ (3.1)

$\xi=\ln \frac {S_t}K$, 有

$ H^*(S_T-K)=H^*(\frac {S_T}{K}-1)=H^*(e^{\xi}-1)=H^*(\xi). $

因此, 转化为Cauchy问题

$ \left\{ {\begin{array}{*{20}{l}} \dfrac {\partial V}{\partial t}+\dfrac 12 \hat{\sigma_t}^2 \dfrac {\partial^2V}{\partial \xi^2}+(r_t-q_t-\dfrac 12 \hat{\sigma_t}^2)\dfrac {\partial V}{\partial \xi}-r_tV=0, \\ {V(\xi, T)=H^*(\xi)}. \end{array}} \right. $ (3.2)

为求解Cauchy问题, 作函数变换$W=Ve^{\beta(t)}, \eta=\xi+\alpha(t), \tau=\gamma(t)$, 可得

$ \begin{array}{ll} \dfrac {\partial V}{\partial \xi}=e^{-\beta(t)} \dfrac {\partial W}{\partial \eta}, \dfrac {\partial^2V}{\partial \xi^2}=e^{-\beta(t)}\dfrac {\partial^2W}{\partial \eta^2}, \\ \dfrac {\partial V}{\partial t}=e^{-\beta (t)}(\frac {\partial W}{\partial \tau}\gamma^\prime(t)- \beta^\prime(t)W+\frac {\partial W}{\partial \eta}\alpha^\prime(t)). \end{array} $ (3.3)

将式(3.3)代入上述方程(3.2), 得到

$ \gamma^\prime(t)\frac {\partial W}{\partial \tau}+\frac 12\hat{\sigma_t}^2\frac {\partial^2W}{\partial \eta^2}+[(r_t-q_t-\frac 12\hat{\sigma_t}^2)+\alpha^\prime(t)]\frac {\partial W}{\partial \eta}-(r_t+\beta^\prime(t))W=0. $ (3.4)

在方程(3.4)中令$r_t-q_t-\frac 12\hat{\sigma_t}^2+\alpha^\prime(t)=0, r_t+\beta^\prime(t)=0, \gamma^\prime(\tau)+\frac 12 \hat{\sigma_t}^2=0, $并结合终值条件$\alpha(T)=\beta(T)=\gamma(T)=0$, 有

$ \begin{array}{*{20}{l}} {\alpha (t) = \int_t^T {({r_\theta } - {q_\theta } - \frac{1}{2}{{\widehat {{\sigma _\theta }}}^2})d\theta , \beta (t)} = \int_t^T {{r_\theta }d\theta } , }\\ {\gamma (t) = \tau = \frac{1}{2}\int_t^T {\sigma _\theta ^2(2H{\theta ^{2H - 1}} - {\rm{Le}}(H))d\theta } .} \end{array} $

因此, 方程组(3.2)转化为如下行形式

$ \left\{ {\begin{array}{*{20}{l}} {\dfrac{\partial W}{\partial \tau}=\dfrac{\partial^2W}{\partial \eta^2}}, \\ {W(\eta, 0)=H^*(\eta)}. \end{array}} \right. $ (3.5)

方程组(3.5)的解可以用Possion公式如下表示

$ \begin{array}{ll} \\W(\eta, \tau)&=\dfrac {1}{\sqrt{2 \pi \tau}} \displaystyle\int_{-\infty}^{\infty}\exp[- \frac {(\eta-y)^2}{4 \tau}]H^*(y)dy \\&=\dfrac{1}{\sqrt{2 \pi \tau}}\displaystyle\int_0^{\infty} \exp[- \frac {(\eta-y)^2}{4 \tau}]dy \\&=N(\frac {\eta}{\sqrt{2 \tau}}). \end{array} $

经过变量代换, 有

$ V_{CC}(S_t, t)=N(\frac {\ln \frac {S_t}{K}+\displaystyle\int_{t}^{T}{(r_\theta-q_\theta-\frac 12 \hat{\sigma_\theta}^2)d\theta}}{\sqrt{\displaystyle\int_{t}^{T}{\sigma_\theta^2(2H \theta^{2H-1}-Le(H))d\theta}}})\exp(-\displaystyle\int_{t}^{T} r_\theta d\theta). $

推论1  假设股票价格满足(2.1)式, 在时刻$t$带有交易费用和红利的现金或无值看跌期权的定价公式为

$ V_{CP}(S_t, t)=N(-\frac {\ln \frac {S_t}{K}+\displaystyle\int_{t}^{T}{(r_\theta-q_\theta-\frac 12 \hat{\sigma_\theta}^2)d\theta}}{\sqrt{\displaystyle\int_{t}^{T}{\sigma_\theta^2(2H \theta^{2H-1}-Le(H))d\theta}}})\exp(-\displaystyle\int_{t}^{T} r_\theta d\theta). $
3.2 资产或无值期权定价公式

定理2  假设股票价格满足(2.1)式, 在时刻$t$带有交易费用和红利的资产或无值看涨期权的定价公式为

$ V_{AC}(S_t, t)=S_tN(\frac {\ln \frac {S_t}{K}+\displaystyle\int_{t}^{T}{(r_\theta-q_\theta+\frac 12 \hat{\sigma_\theta}^2)d\theta}}{\sqrt{\displaystyle\int_{t}^{T}{\sigma_\theta^2(2H \theta^{2H-1}-{\rm Le}(H))d\theta}}})\exp(-\displaystyle\int_{t}^{T} q_\theta d\theta). $

  由方程组(2.8)可以得到资产或无值看涨期权定价模型如下

$ \left\{ {\begin{array}{*{20}{l}} {\dfrac {\partial V}{\partial t}+\dfrac 12 \hat{\sigma_t}^2 S_t^2 \dfrac {\partial^2V}{\partial S_t^2}+(r_t-q_t)S_t\dfrac {\partial V}{\partial S_t}-r_tV=0}, \\ {V|_{t=T}=S_tH^*(S_t-K)}. \end{array}} \right. $ (3.6)

$V_{AC}(S_t, t)=S_tU(S_t, t)$, 则$ U(S_T, T)=\frac 1S V_{AC}(S_T, T)=V_{CC}(S_T, T). $很容易得到$U(S_t, t)$满足下列方程

$ \left\{ {\begin{array}{*{20}{l}} {\dfrac {\partial U}{\partial t}+\dfrac 12 \hat{\sigma_t}^2 S_t^2\dfrac {\partial^2U}{\partial S_t^2}+(r_t-q_t+ \hat{\sigma_t}^2)S_t\dfrac {\partial U}{\partial S_t}-q_tU=0}, \\ {U|_{t=T}=H^*(S_t-K)}. \end{array}} \right. $

$\xi=\ln \frac {S_t}K$, 因此转化为Cauchy问题

$ \left\{ {\begin{array}{*{20}{l}} \dfrac {\partial U}{\partial t}+\dfrac 12 \hat{\sigma_t}^2 \dfrac {\partial^2U}{\partial \xi^2}+(r_t-q_t+\dfrac 12 \hat{\sigma_t}^2)\dfrac {\partial U}{\partial \xi}-q_tU=0, \\ U(\xi, T)=H^*(\xi). \end{array}} \right. $ (3.7)

为求解Cauchy问题, 作函数变换$W=Ue^{\beta(t)}$, $\eta=\xi+\alpha(t)$, $\tau=\gamma(t)$, 可得

$ \begin{array}{ll} \dfrac {\partial U}{\partial \xi}=e^{-\beta(t)} \frac {\partial W}{\partial \eta}, \dfrac {\partial^2U}{\partial \xi^2}=e^{-\beta(t)}\frac {\partial^2W}{\partial \eta^2}, \\ \dfrac {\partial U}{\partial t}=e^{-\beta (t)}(\frac {\partial W}{\partial \tau}\gamma^\prime(t)- \beta^\prime(t)W+\frac {\partial W}{\partial \eta}\alpha^\prime(t)). \end{array} $ (3.8)

将式(3.8)代入(3.7)式, 得到

$ \gamma^\prime(t)\frac {\partial W}{\partial \tau}+\frac 12\hat{\sigma_t}^2\frac {\partial^2W}{\partial \eta^2}+[(r_t-q_t+\frac 12 \hat{\sigma_t}^2)+\alpha^\prime(t)]\frac {\partial W}{\partial \eta}-(q_t+\beta^\prime(t))W=0. $ (3.9)

在方程(3.9)中令$r_t-q_t+\frac 12\hat{\sigma_t}^2+\alpha^\prime(t)=0, q_t+\beta^\prime(t)=0, \gamma^\prime(t)+\frac 12\hat{\sigma_t}^2=0, $并结合终值条件$\alpha(T)=\beta(T)=\gamma(T)=0$, 则有

$ \begin{array}{*{20}{l}} {\beta (t) = \int_t^T {{q_\theta }d\theta , \alpha (t)} = \int_t^T {({r_\theta } - {q_\theta } + \frac{1}{2}{{\widehat {{\sigma _\theta }}}^2})d\theta } , }\\ {\gamma (t) = \tau = \frac{1}{2}\int_t^T {\sigma _\theta ^2(2H{\theta ^{2H - 1}} - {\rm{Le}}(H))d\theta } , } \end{array} $

从而得到

$ \left\{ {\begin{array}{*{20}{l}} {\dfrac{\partial W}{\partial \tau}=\dfrac{\partial^2W}{\partial \eta^2}}, \\ {W(\eta, 0)=H^*(\eta)}. \end{array}} \right. $ (3.10)

方程(3.10)的解可以用Possion公式如下表示$W(\eta, \tau)=N(\frac {\eta}{\sqrt{2 \tau}}).$经过变量代换, 得到

$ V_{AC}(S_t, t)=S_tN(\frac {\ln \frac {S_t}{K}+\displaystyle\int_{t}^{T}{(r_\theta-q_\theta+\frac 12 \hat{\sigma_\theta}^2)d\theta}}{\sqrt{\displaystyle\int_{t}^{T}{\sigma_\theta^2(2H \theta^{2H-1}-{\rm Le}(H))d\theta}}})\exp(-\displaystyle\int_{t}^{T} q_\theta d\theta). $

推论2  假设股票价格满足(2.1)式, 在时刻$t$带有交易费用和红利的资产或无值看跌期权的定价公式为

$ V_{AP}(S_t, t)=S_tN(-\frac {\ln \frac {S_t}{K}+\displaystyle\int_{t}^{T}{(r_\theta-q_\theta+\frac 12 \hat{\sigma_\theta}^2)d\theta}}{\sqrt{\displaystyle\int_{t}^{T}{\sigma_\theta^2(2H \theta^{2H-1}-{\rm Le}(H))d\theta}}})\exp(-\displaystyle\int_{t}^{T} q_\theta d\theta). $

注4  对于单个欧式两值期权, 如果做多头, 因为$\Gamma>0$, Leland数有可能大于$2Ht^{2H-1}$, 即当

$ k>\frac {2Ht^{2H-1}\sigma_t}{(\delta t)^{H-1}}\sqrt{\frac \pi2} $

时, 方程(2.8)变成正向抛物型方程的终值问题; 因为正向抛物型方程的终值问题是一个不适定问题.所以为了使定解问题(2.8)是适定问题, 必须要假定

$ k<\frac {2Ht^{2H-1}\sigma_t}{(\delta t)^{H-1}}\sqrt{\frac \pi2}, $

这表明交易费用比较小, 或者对冲风险的过程不能太频繁, 否则应用Leland模型求解有交易费用的期权价格是不正确的.

本文研究了在分数布朗运动下带交易费用和红利的两值期权定价问题, 将标准布朗运动下的带交易费用的两值期权定价问题推广到Hurst指数为$H\in(\frac 12, 1)$分数布朗运动的情况下.根据两值期权的分类分为现金或无值期权和资产或无值期权两种情况.通过利用无风险套利原则和分数Itô公式, 得到与之对应的数学模型.求解此模型得到了现金或无值看涨期权和资产或无值看涨期权的定价公式, 并在此基础上得到了现金或无值看跌期权和资产或无值看跌期权的定价公式.在这个领域还有很多问题去研究, 比如在Levy过程, 混合分数布朗运动下讨论此类问题.

参考文献
[1] Black F, Scholes M. The pricing of options and corporate liabilities[J]. J. Polit. Econ., 1973, 81(3): 637–654. DOI:10.1086/260062
[2] Peters E E. Fractal market analysis[J]. New York:Wiley, 1994.
[3] Duncan T E, Hu Y Z, Pasik-Duncan B. Stochastic calculus for fractional Brownain motion (I)-theory[C]. IEEE Confer. Dec. Contr., 2000, 38(2):212-216.
[4] Elliott R J, Hoek J V D. A general fractional white noise theory and applications to finance[J]. Math. Finance, 2003, 13(2): 301–330. DOI:10.1111/mafi.2003.13.issue-2
[5] Hu Y Z, ØksendalB. Fractional white noise calculus and application to finance[J]. Infinite Dimen. Anal., Quantum Prob. Rel. Top., 2003, 6(1): 1–32.
[6] Bender, Christian. An Itô formula for generalized functionals of a fractional Brownian motion with arbitrary Hurst parameter[J]. Stoch. Proc. Appl., 2003, 104(1): 81–106. DOI:10.1016/S0304-4149(02)00212-0
[7] Leland H E. Option pricing and replication with transactions costs[J]. J. Finance, 1985, 40(5): 1283–1301. DOI:10.1111/j.1540-6261.1985.tb02383.x
[8] Guasoni P. No arbitrage under transaction costs, with fractional Brownian motion and beyond[J]. Math. Finance, 2006, 16(3): 569–582. DOI:10.1111/mafi.2006.16.issue-3
[9] Zhang Y, Pan D, Zhou S W, Han M. Asian option pricing with transaction costs and dividends under the fractional Brownian motion model[J]. J. Appl. Math., 2014, 11: 1–8.
[10] Wang X T. Scaling and long-range dependence in option pricing Ⅰ:pricing european option with transaction costs under the fractional Black-Scholes model[J]. Phys. Stat. Mech. Appl., 2010, 389: 438–444. DOI:10.1016/j.physa.2009.09.041
[11] Thavaneswaran A, Appadoo S S, Frank J. Binary option pricing using fuzzy numbers[J]. Appl. Math. Lett., 2013, 26(1): 65–72. DOI:10.1016/j.aml.2012.03.034
[12] 袁国军. 基于半离散化的CEV过程下两值期权定价研究[J]. 系统工程学报, 2012, 27(1): 19–25. DOI:10.3969/j.issn.1000-5781.2012.01.003
[13] Hofer V, Leitner J. Relative pricing of binary options in live soccer betting markets[J]. J. Econ. Dyn. Contr., 2017, 76: 66–85. DOI:10.1016/j.jedc.2016.12.007
[14] 吴云, 何建敏. 两值期权的定价模型及其求解研究[J]. 管理工程学报, 2002, 16(4): 108–110. DOI:10.3969/j.issn.1004-6062.2002.04.026
[15] 孙天宇, 刘新平. 有交易成本且支付红利的两值期权定价模型[J]. 陕西师范大学学报, 2008, 36(6): 19–22.
[16] 姜礼尚. 期权定价的数学模型和方法[M]. 北京: 高等教育出版社, 2003.
[17] 姜礼尚, 徐承龙, 任学敏. 金融衍生产品定价的数学模型与案例分析[M]. 北京: 高等教育出版社, 2008.