Toward a molybdenum price model

If you’re a pure-play molybdenum name, a lot rides on the price of mol. To a first approximation, the quality of your ore deposit sets production costs, with an important secondary variable being the price of energy. In order to assess the feasibility of a given project, one has to model selling prices per pound going forward. For example, in General Moly‘s bankable feasibility study for the Mt. Hope deposit, the forward price assumptions are as follows:

The above chart was drawn up in April 2008, prior to the financial crisis. The ’08-’09 price collapse basically dispatched the early-year predictions into the realm of wishful thinking, but hope springs eternal.

The molybdenum price chart for the past fifteen years makes it clear that molybdenum (like many commodities) is prone to extraordinary price volatility:

It’s thus interesting to try to evaluate what the price of molybdenum might look like going forward. As a first attempt, I took the above time-series and sampled it at 91 equally spaced intervals (corresponding to a two-month sampling cadence). I then differenced the resulting samples to get a set of 2-month price moves. Sampling from this distribution of 2-month returns, and enforcing a hard floor of 3 dollars per pound, yields a set of 15-year forward trajectories. Here’s a plot of ten representative samples:

For the most part, these trajectories look awfully good. Based on 1,000 trials, the average finishing price of mol fifteen years from now is 41.04 ± 25.26 USD/lb. The reason for the high average price, of course, is that while mol has come way down from the glory days of 2004-2008, it is still up by more than a factor of three from its abysmal mid-1990s lows. Sampling the price trajectory differentials over the last 15 years thus bakes in a big average gain.

For an improvement in realism, we can multiply the sampled differences randomly by -1 or 1. This removes some of the upside bias, but because we retain the assumed 3 USD/lb price floor, we experience the opposite of gamblers ruin in a fraction of the trials, which leads to a average net gain after 15 years. The average finishing price (again after 1,000 trials) is 30.62 ± 23.37 USD/lb. A random sampling of ten representative trajectories looks like:

These trajectories look somewhat more believable, and in particular, they are better able to capture the occurrence of long periods when the price is down in the dumps with seemingly little hope of recovery. What they don’t show, however, is a tendency to return back to “baseline” after a sharp upward spike. I think a realistic model needs to include this tendency, which comes about either from the bursting of asset bubbles, or from a situation in which high prices lead to a surge in output, leading to a crash in prices. More on this in an upcoming post…