7
$\begingroup$

So called closed FX-Forwards are well known forward contracts where some amount of foreign currency is bought at a specified date in the future for a price fixed "today". Such contracts can be valuated using the well known cost-of-carry formula.

Recently, I learned about open FX-forward contracts. In this kind of contract the holder has the flexibility to make as many drawdowns as he wants during a specified period as long as the full amount is paid by maturity see e.g. this page.

What is market practice to value such open FX-forward contracts?

$\endgroup$
8
  • $\begingroup$ I feel it could be some sort of average between "closed" FX Fwd quotes. Say we enter into an "open" FX Fwd with two exchange dates: at the 3M & 6M mark. Then in the absence of arbitrage the buyer should be projected to be indifferent to exchanging money at 3M or 6M (as seen from today), since the FX Fwd quotes represent exactly this equilibrium. Hence it should be 50-50 which date he chooses. So the closed quote is the average of the two. Thoughts? $\endgroup$
    – Phil-ZXX
    Commented Jul 10, 2018 at 16:50
  • $\begingroup$ I wonder whether one could use some dominance argument similar to American options on non-div paying socks which have the same price as European options. $\endgroup$
    – Richi Wa
    Commented Jul 10, 2018 at 17:02
  • $\begingroup$ This could depend on the interest rate differential between the two currencies @Phil-ZXX $\endgroup$
    – Richi Wa
    Commented Jul 10, 2018 at 17:03
  • $\begingroup$ Thinking about it more, you probably need a volatility term-structure model since each "open" date (on which the buyer is allowed to exchange money at the predetermined fx "strike rate") essentially represents an option. So american/bermudan pricing can probably be applied. Perhaps longstaff schwartz or binomial trees are applicable? $\endgroup$
    – Phil-ZXX
    Commented Jul 10, 2018 at 19:15
  • $\begingroup$ @Phil-ZXX thank you for your comments. So far in the methods you mention the amount at each of the Bermudan dates does not matter ... maybe it should? I wonder whether there exists some market practice - a rule of thumb? $\endgroup$
    – Richi Wa
    Commented Jul 11, 2018 at 6:25

3 Answers 3

1
$\begingroup$

The flexible forward contract is very much like an American option: at each exercise date, you have the choice to receive the payoff $(S-K)$ or not. The difference with a regular option is that you must choose a date.

In effect, this is a classical optimal stochastic control problem and may be solved using exactly the same techniques as for an American (or a Bermudan) option: typically a finite difference method applied to the Black-Scholes PDE with the linear complementary constraint $$f(t_i, s_j) \geq s_j - K$$, where $t_i$ is an exercise date and $s_j$ the asset price in the FDM discretization grid.

In particular the price will depend on the asset volatility, and the contract will have a non-zero vega.

See https://www.worldscientific.com/doi/abs/10.1142/S2424786316500109, https://www.globalcapital.com/article/k6b8msb96708/american-currency-forwards.

$\endgroup$
1
$\begingroup$

To answer your answer: Suppose you are the holder of the open contract. You hedge it by executing a vanilla forward at 1.1679 for date 92. You now have an arbitrage, for if the fx forward for one of the dates 88 to 91 becomes higher than that for date 92, you can switch the hedge to that other date, This means that the true price of your open contract must be slightly greater than 1.1679. However, the switch in this case is unlikely, because it would only occur if euro rates exceed usd rates. It is an option on the rate differential. If you created an open FX forward on a currency pair where rates are very similar , the effect would be greater.

$\endgroup$
3
  • $\begingroup$ Thank you for this remark. By "slightly greater than 1.1679" you mean in order to reflect the option? If so then this would be the new content. The rest is contained in "my answer" below. Do you agree? If you bring up new aspects then I would be happy to read about them. Thanks! $\endgroup$
    – Richi Wa
    Commented Sep 11, 2018 at 11:00
  • $\begingroup$ Yes, the excess over 1.1679 is the value of this switch option. It would be complex to value it precisely. $\endgroup$
    – dm63
    Commented Sep 12, 2018 at 4:24
  • $\begingroup$ The intuition is somewhat right, but the problem is really an optimal control problem, and you need a numerical method to solve the non-linear PDE. The price may be vastly different from 1.1679 and will involve the underlying asset volatility. $\endgroup$
    – jherek
    Commented Jan 23, 2021 at 12:11
-1
$\begingroup$

I am trying to answer my own question to make discussion possible.

Say we have an open FWD with period $[T_1,T_2]$ in which we can settle it. The strike price $K$ is fixed today.

As a example for EUR USD we have a spot of 1.16 (USD per EUR) and let us assume that the strike price for the above forward is $K = 1.1677$ (we have much higher USD rates than EUR leading to this higher forward price).

Then, on any given day $t$ I can compare this $K$ to the forward prices of forwards that stettle on all the days in the interval $[T_1,T_2]$.

The fair price is $$ F_{T_i} = S_t \exp ( (r_d(T_i)-r_f (T_i))\cdot (T_i-t)/365 ) $$ and a rational agent will settle when the gains are highest thus at $$ T^* = \arg max_{t \in [T_1,T_2]} \{ F_{T_i}-K \}. $$ Thus the price $K$ has to equal this one $F_{T^*}$. If the ir-differential does not changes too much during $[T_1,T_2]$ then I assume that this $T^*$ will be the first of the last day of the period depending on the sign of the differential.

Continuing the example we can calculate the forward prices (crudely) for some $T_i$:

  • $T_i = 88$ then $F_{88} \approx 1.1675$ and the gain is $-0.00017$
  • $T_i = 90$ then $F_{90} \approx 1.1677$ gain is zero.
  • $T_i = 92$ then $F_{92} \approx 1.1679$ gain is approx $0.00017$

Thus if the period is $\{88, 89, 90, 91, 92\}$ the price of the open forward should be $K=1.1679$, which is the forward price for settlement at the last day of the period as the interest rate differential between USD and EUR (USD-EUR ir) is positive.

$\endgroup$
1
  • $\begingroup$ The problem to solve is an optimal control problem, the reasoning here leads to a lower bound, but underestimate the true price of the contract. $\endgroup$
    – jherek
    Commented Jan 23, 2021 at 12:22

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

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