I'm working on the following problem and would appreciate some input because I'm stuck.

Consider a fund that works as follows. The fund starts with $S_0$ worth of assets following a Geometric Brownian Motion. At each time of the schedule $\{t_i\}_{1 \leq i \leq N}$, party $B$ gets a payment worth $\alpha S_{t_i}$. This goes on for $N$ dates, then a party $A$ fills the fund back to $S_0$ if $S_{t_N} < S_0$. If $S_{t_N} > S_0$ the difference is paid to party $C$.

Now I need to value the claims of all three parties. At maturity the claim of $C$ is worth $\max(S_{t_N}-S_0, 0)$, while the claim of $A$ is worth $\min(S_{t_N} - S_0, 0)$. Now let $Q$ be the risk-neutral measure. I thought the initial value of these claims to be $e^{-rt_N}E^Q(\max(S_{t_N}-S_0, 0))$ and $e^{-rt_N}E^Q(\min(S_{t_N}-S_0, 0))$ respectively. I approximated this expectation by simulating under $Q$ (so simulate $S_{t_i}$ under $Q$ with drift term $r$, where $r$ is the risk free rate). However at each period I reset as follows: $$S_{t_i^+} = (1-\alpha)S_{t_i^-}$$ where: $$S_{t_i^-}=S_{t_{i-1}^+}e^{(r-\frac{\sigma^2}{2})(t_i-t_{i-1})+\sigma W_{t_i-t_{i-1}}}$$ 1) Can I still use this pricing method? Doesn't taking a cut of $S_{t_i}$ every period ruin it?

2) The second question I have is what is the claim of $B$ and how to price it? My intuition: After $N$ periods $B$ has received $\sum_i \alpha S_{t_i}$. But how to price this?

3) And lastly: there is supposed to be a parity relationship between the values of the three claims. The fund transforms $S_0$ and the contribution of $A$ into payments to $B$ and $C$ and $S_0$. So I thought, somehow the price of the claim of $A$ should be equal to the prices of $B$ and $C$. Is this correct? If so why? If not what should I do?

  • $\begingroup$ Regarding the "additivity of values" question: Who puts up the amount $S_0$ in the first place and who receives $S_0$ when the arrangement ends? If this is another party D, then you need to include them, i.e.$−A+B+C+D=0$. The value of $D$ clearly is negative as he gives up the return on his capital $S_0$ for N years. This amount plus the contributions from A are being used to pay the cash flows to B and C. Of course some of A,B,C,D could be the same person. $\endgroup$ – noob2 Oct 6 '17 at 14:40

To be rigorous, I have modified your original question to define:

$$ S_{t_i^+}=(1-\alpha)S_{t_i^-}$$

where $t_i$ is a payment date for $B$. At $t_i$, $B$ receives the amount $\alpha S_{t_i^-}$. This way we distinguish $S_{t_i}$, $S_{t_i^+}$ and $S_{t_i^-}$ where:

$$ S_{t_i} = S_0e^{(r-\frac{\sigma^2}{2})t_i+\sigma W_{t_i}} $$

1) To me there is no particular issue. Letting $0<t_0<t$, note that:

$$ \begin{align} S_t&=S_0e^{(r-\frac{\sigma^2}{2})t+\sigma W_t} \\[9pt] &=\left(S_0e^{(r-\frac{\sigma^2}{2})t_0+\sigma W_{t_0}}\right)e^{(r-\frac{\sigma^2}{2})(t-t_0)+\sigma (W_t-W_{t_0})} \\[6pt] &=S_{t_0}e^{(r-\frac{\sigma^2}{2})(t-t_0)+\sigma W_{t-t_0}} \end{align}$$


$$ \begin{align} S_{t_0^-} & = S_0e^{(r-\frac{\sigma^2}{2})t_0+\sigma W_{t_0}} \tag{1} \\[6pt] S_{t_0^+} & = (1-\alpha)S_{t_0^-} \tag{2} \end{align}$$

We have:

$$ S_{t_0^+}e^{(r-\frac{\sigma^2}{2})(t-t_0)+\sigma W_{t-t_0}} = (1-\alpha)S_{t_0^-}e^{(r-\frac{\sigma^2}{2})(t-t_0)+\sigma W_{t-t_0}} = (1-\alpha)S_t$$

If we have a 2nd date $t_1>t_0$:

$$ \begin{align} S_{t_1^+}e^{(r-\frac{\sigma^2}{2})(t-t_1)+\sigma W_{t-t_1}} & = (1-\alpha)S_{t_1^-}e^{(r-\frac{\sigma^2}{2})(t-t_1)+\sigma W_{t-t_1}} \\[6pt] & = (1-\alpha)S_{t_0^+}e^{(r-\frac{\sigma^2}{2})(t_1-t_0)+\sigma W_{t_1-t_0}} e^{(r-\frac{\sigma^2}{2})(t-t_1)+\sigma W_{t-t_1}} \\[6pt] & = (1-\alpha)S_{t_0^+}e^{(r-\frac{\sigma^2}{2})(t-t_0)+\sigma W_{t-t_0}} \qquad \\[6pt] & = (1-\alpha)^2S_t \end{align}$$

So you can simulate your asset and then multiply by $(1-\alpha)^N$ posteriorly. From $\text{(1)}$ and $\text{(2)}$ you see that:

$$ S_{t_i^-} = (1-\alpha)^{i-1}S_0e^{(r-\frac{\sigma^2}{2})t_i+\sigma W_{t_i}} = (1-\alpha)^{i-1}S_{t_i}$$

2) The price of a claim is its discounted risk-neutral expectation, hence letting $\pi_B$ be the price of $B\text{'s}$ claim we have:

$$ \begin{align} \pi_B&=E^Q\left[\sum_{i=1}^Ne^{-rt_i}\alpha S_{t_i^-}\right] \\[6pt] & = \sum_{i=1}^N\alpha e^{-rt_i}E^Q\left[S_{t_i^-}\right] \\[6pt] & = \sum_{i=1}^N\alpha e^{-rt_i}(1-\alpha)^{i-1}E^Q\left[S_{t_i}\right] \\[6pt] & = \alpha S_0\sum_{i=1}^N e^{-rt_i}(1-\alpha)^{i-1}e^{rt_i} \\[6pt] & = \alpha S_0\sum_{i=0}^{N-1} (1-\alpha)^{i} \\[6pt] & = S_0(1-(1-\alpha)^N)\end{align}$$

3) First, note that:

$$ \min(S_{t_N}-S_0,0)=-\max(S_0-S_{t_N},0) $$

Hence party $A$ has sold a put with strike $S_0$ to the fund. Additionally, vis-à-vis party $C$ the position is equivalent to the fund having sold a call with strike $S_0$. As a result, combining the positions of $A$ and $C$, the fund is short a forward contract on the asset $S_t$.

Moreover, $B\text{'s}$ payments can be interpreted as dividends: the asset pays a proportional dividend with rate $\alpha$ at times $\{t_i\}_i$. After each payment the asset depreciates by $\alpha S_{t_i^-}$.

Assuming the asset is effectively dividend-paying, we conclude that to replicate all payments the fund only needs to hold one unit of the asset: it will pay $B$ with dividends received and at maturity $t_N$ it will sell it for a price equal to $S_0$ to either $A$ (if $S_{t_N} \leq S_0$) or $C$ (if $S_{t_N} > S_0$).


This is an addendum to the answer provided by Daneel Olivaw.

Let's assume that the payment at $t_i$, for $i=1, \ldots, N$, is $\alpha_i S_{t_i}$. We define the step function \begin{align*} M_t = \sum_{i=1}^N \alpha_i \pmb{1}_{\{t_i \le t\}}. \end{align*} Moreover, we assume that the fund value process $\{S_t, \, t \ge 0\}$ satisfies an SDE of the form \begin{align*} dS_t = S_{t-}\big(rdt-dM_t + \sigma dW_t \big). \end{align*} Then \begin{align*} S_t = S_0 e^{(r-\frac{1}{2} \sigma^2)t + \sigma W_t}\Pi_{t_i \le t} (1-\alpha_{i}). \end{align*} The valuation of the respective payoffs can now be followed as the answer of Daneel Olivaw.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.