Modeling Commodities Assets as Options

I built a real option model for a potential import facility we were looking into building. The logic being that it gave us the option, but not the obligation, to turn a foreign commodity into a domestically priced commodity of the same class at this specific location when it was profitable to do so (imported price domestic).

Take a nat gas fired power plant for example. If you own this asset you have the option, but not the obligation, to turn gas into electricity if it is profitable to do so. You can think of this as owning a real option to turn nat gas into electricity. Whether or not it is profitable depends on the spread between the price of nat gas and electricity (the spark spread).

Using Black-Scholes as your real option model:

Underlying: Average spread price between nat gas and electricity during the period you're interested in.

Strike: Whatever threshold you deem to be when it is profitable to turn nat gas into electricity. Maybe it is when electricity is 0.01 cents more than nat gas.

Term: Whatever time frame you want to look at the option over. Given seasonality and the fact that Black Scholes isn't very good at modeling really long dated options I would argue looking at it seasonally would be the best. So call it 3 months.

Risk-Free Rate: whatever you think works for a risk free rate

Implied Vol: Use the historical vol of your spread.

This would give you the value of the asset as a real option on a per unit of gas basis. Multiply that by how much gas you can actually process and it would give you a value of your asset as an option with 3 months to expiry. Do that for each season and add them all up for however long you think your assumptions will hold. Discount those future cash flows back to the present and you have the long term value of that asset.

When I built this model for work I referenced this paper a lot. Specifically the Margrabe Formula at the end of page 30.

https://carmona.princeton.edu/download/fe/sirev.pdf

Having said all that, another way that is easy with modern computing is to run a monte carlo simulation on it. Write a program that takes the assumptions from above and then introduces some randomness into the price action and see what the payout over a 3 month period is. This would allow you to play around with the profitable price to switch more since the math gets hairy on that when it isn't zero.

If your company leases assets then the logic is all still the same and I would argue even more valuable since you can lease for shorter amounts of time and theoretically could decide to lease or not each period based on your models valuation. The same logic can apply for something like pipe space, storage, etc.