wilmott.com

Hi all,Quick question about window FX forwards, where the buyer has the right to settle the forward at any time within a specified window of time.The impression I get from speaking to colleagues is that the general approach used to value these is to value to the worst possible result for the bank, given the current forward rates i.e. there is no explicit modelling of optimal exercise time. Is this right? Are there any other approximations used?Thanks in advance,Cheers,Donal

I would agree with your understanding, pricing is generally done on a worst case basis. I think the valuation of this fairly common product is an under studied problem and someone could really make a name for themself by shedding more light on them.

What's the particular modelling issue with these?

So if I understand well, you only have to choose the date (in a given window) when the forward contract is settled ? The forward is settled at the fair forward rate according the interest rate differential at the moment of the settlement right ?I once ago opened a topic on american forward start american option in the technical forum at it reminds me a bit the idea.Just correct me if I'm wrong in my understanding of the features of the contract. I will think a bit about this interesting problem

An example would be that you enter into a one year forward at rate X. You can settle this at any point in the last 3 months at rate X.

Ok so the option is in the choosing of the expiration of the contract (between T1 and T2).If I'm right I would try to model this as an option named W.I would specify a stochastic process for the current exchange rate s, X is your forward, rf is the foreign rate and rd is the domestic one. We could use a Garman & Kohlhagen formalism (BS one if you prefer). I think that it can be valued as an american option with the free boundary constrain W(s,t)>Payoff(s,t) where the payoff is the classical forward price s*exp(-rf*t)-X*exp(-rd*t) with t between T1 and T2.The hard point also is that there is also a forward start feature of this option contract. I will think later about that. A hard problem but an interesting one EDIT: If I'm right since the strike is fixed at T1, the value of the option at time zero is simply the value at time T1 discounted by exp(-rd*T1).So it should really be like an american like option.

These very simple contracts are not "traded" which is why the valuation is very simple.The need arises from an exporter/importer, who is expecting to be paid, or will pay, for goods needed at some time in the future, but that date is not certain. For instance, a construction company is engaged in a project to build a warehouse. The project requires, say, securing the site (construction of fencing, security gates, whatever), demolition of current buildings, clearance, digging of lower floors/foundations, construction of the warehouse, etc.Let's say that all of the materials can be sourced locally except for the air conditioning plant for the completed warehouse, and the supplier for that is in South Africa. The supplier wants paying in ZAR, and the construction company is based in France, so needs to buy ZAR using EUR. The purchase of the aircon is agreed, the price is set, it is expected to get to the point in the project that the aircon needs installing in 6 months time.The construction company could agree an outright FX contract for the specific date required. But what happens if the project proceeds smoother than expected... do you just have your workers sit idle for a couple of weeks before the FX settles? You could negotiate an 'early delivery' but you have to pay penalties for that. A similar issue occurs if the project is delayed.The simplest to to agree a windowed forward, and drawdown when it is needed. This means that the drawdown on the windowed forward has nothing to do with the current market factors, so there can be no 'logic' to the valuation. [This is rather like the inclusion of 'investor irrationality' when pricing mortgage-backed securities.]Hope that helps.

I'd agree that factors other than purely financial considerations can influence the drawdown time on a window forward. But the whole point of the transaction in the first instance is to transfer price risk, so with all due respect I think characterizing the pricing of these things as simple and not logical is way off the mark.

The main problem is that 'buyer' of this contract is the corporate and, while I agree that their purpose is to transfer price risk, it is purely for the purpose of the corporate knowing how many of their domestic currency they can expect to record in their accounting system. Their sole purpose is to mitigate price variability at the time that the need arises, not when the financial market conditions are optimal.I agree that this means that their valuation is extremely difficult, however as the corporate may be forced (by the original business condition coming due) to exercise the option (drawdown the contract) at a time when market conditions may dictate that holding on to the contract may be financialy more advantageous, that is not feasible (especially if the contract is the subject of regulation, such as foreign exchange controls).

From my limited knowledge,It is just a basket option (long and short FRAs in the 2 ccys) or a call option on the spread (FRA(CCY1) - FRACCY(2)).purely an interest rate product as you only need an option on the forward pointsEarly exercise suboptimal - better to sell the option than draw down.

Hi Donal,As i understand from description, CP has right to exercise forward on any day within specified time window.For this i think best solution is slightly modified version of LSMC( Longstaff Swartz Monte Carlo) because in this MC pricing you need to compare continuation value with intrensic value( which is not only forward looking or backward ,you need to use combination of both of them).LSMC will allow to capture optimal exercise time point in each and every path & simultaneously it will capute path dependency.Regards