Thoughts on the valuation of gold equities

Reading through financial textbooks, portfolio manager commentary and sell-side analyst reports, you’ll come across various metrics how to value a company’s equity.

For example, the most common are price multiples such as P/E, P/CF and P/B.

Though you may see the same metrics used occasionally when discussing mining companies, you are more likely come across net asset value (NAV).

In a nutshell, a company’s NAV is calculated by summing up the discounted cash flows (DCF) produced from its mine(s), adding cash and subtracting any debt.

There are a number of reasons why NAV, instead of price multiples, should be used when valuing mining companies, such as NAV takes into consideration a company’s debt position, as well as the cash flows over an entire mine’s life (and hence the size of a mine’s reserves/resources along with the metallurgical recoveries), and the initial and sustaining capital required to buld and operate a mine .

NAV is by no means perfect. One of its drawbacks is the sensitivity to the, and how to determine the appropriate, discount rate (analysts use a lower discount rate for gold mines than base metal mines. One reason is because the most popular method to calculate the discount rate is the Capital Asset Pricing Model (CAPM) that states the cost of equity equals the risk-free rate plus beta times the market risk premium. Since gold has a negative beta (i.e., negative correlation to the market in general), it therefore has a theoretical cost of equity that is less than the risk-free rate. I am not defending this method, just pointing it out). Other drawbacks of NAV include the commodity price assumption (should we use a static price? the futures curve?), it only accounts for trailing reserves (which tend to be understated in a rising metal price environment as they are booked at lower prices) unless the analyst makes assumptions regarding the size of a deposit, and penalizes long-life assets (particularly when using higher discount rates).

Despite these drawbacks, analysis by NAV is still more theoretically sound than price multiples.

When preparing a DCF model for an established producer, the analyst assumes a mine will open at some specific time and continue to produce without cessation until the ore is exhausted. The price of the underlying commodity is assumed to be static, follow the futures curve or converge on some long-term forecast.

Upon completion, the analyst will then find that the market price of shares of a senior base metal producer is generally in-line with NAV, yet equities of senior gold producers trade well above NAV (depending on the analyst’s assumptions, it can be up to 2.5 x).

This is not a local or recent phenomenon. Discussions with experienced portfolio managers and sell-side analysts, along with references in the literature, indicate that precious metal equities have historically traded at a multiple to NAV, and that this has occurred globally.

Some of the reasons that are proposed on why this occurs include:

• Exploration potential
• Potential M&A activity
• Cyclical premium expansion
• Resource scarcity
• Leverage to commodity price
• Operational efficiencies

Though some of these may make intuitive sense, the same reasons could equally apply to shares of base metals producers.

Instead, the answer has to do to how metal prices change over time (i.e., the path the commodity price follows). In essence, a DCF model using a mean-reverting (random walk) commodity price will create a NAV lower than a model using a non mean-reverting commodity price.

For example, Slade (2001) found that a mine where the commodity is valued as a random walk is almost twice the mean-reversion value (i.e., 2 x NAV).

“When a random variable is modeled as a random walk, in the course of 20 years it is bound to drift into areas that are well outside of its historic range, irrespective of its initial condition. When it is modeled as mean reverting, in contrast, it has no such tendency. Furthermore, flexible operation limits downside risk but not upside return and therefore enhances the value of uncertainty.” (Slade, 2001)

In other words, when the commodity price is modelled as a random walk, some years the mine will ramp up production to take advantage of higher prices, whereas during times of low prices the mine has the option to either slow production or temporarily halt production. These options are worth something to the mine operator and should be factored into the NAV.

Referred to as real options, a whole body of literature exists on the subject.

Real option analysis has been used in the valuation of exploration stage companies. For example, modelling gold as a random walk and copper as mean reverting, Davis and Samis (2006) found that the optimal exploration policy depends on the commodity. Specifically:

For gold, “during periods of low and moderate prices, focus on promising green-fields deposits and deposits with previous positive exploration results while deferring geologically unpromising green-fields prospects until prices improve. This is because given gold’s random walk price characteristic, there is always a chance that higher and higher gold prices could make up for poor geological potential.

“On the other hand, our model suggests that unpromising green-fields copper projects should not be kept in deferral mode during low prices, since future prices are capped by the trend of copper prices to a long-run mean. Green-fields copper projects should therefore be explored immediately if promising or permanently abandoned if unpromising, regardless of current copper price.”

The question then is: does gold follow a random walk? There are numerous statistical studies that have shown precious metals prices to be non-mean reverting (e.g., Dincerler et al., 2005; Borenstein and Farrell, 2006). I personally have not been able to show mean reversion using month end gold prices.

Why is the price of gold a random walk, yet not the price of base metals? It is because the factors influencing the supply-demand balance for gold are different than base metals: specifically, gold is stored whereas base metals are consumed.

Consider copper. Once it is mined and refined, it is consumed in a manufacturing process, like copper wiring, where it remains (it may eventually be recycled). Therefore, the price of copper is primarily influenced by the amount of mine supply and the demand from end users. If mine supply decreases or demand from end-users increases, the price tends to increase, and vice-versa. Over time, this causes the price to revert to the cost of production

On the other hand, gold is not consumed in the traditional sense (outside of minor industrial applications) in that mined and refined gold is sent to vaults or made into jewellery, a form in which it can be sold quickly (unlike copper wire in a house). Hence almost all of the approximately 150,000 tonnes of gold ever mined can be easily resold. Even if every gold mine in the world shut down tomorrow, a roughly 2,500 tonne annual source, there would still be approximately 150,000 tonnes of gold available for sale.

The reason gold is not consumed in the traditional sense is because gold is money - money that offers no yield (a negative yield after taking into account storage and insurance) and which is continually valued against fiat currencies that offer a yield.

As a result, investment demand is the prime driver of the gold price. As Barsky and Summers (1999) stated: “…it is the demand for the existing stock, as opposed to the new flow, that must be modeled. The willingness to hold the stock of gold depends on the rate of return available on alternative assets.” And the prime driver of investment demand is real interest rates, which are non-mean reverting.

So to sum up, the reason senior gold equities trade at a premium to NAV is due to the non mean-reverting price performance of gold, where the supply-demand factors that affect base metals (e.g., mine supply, global economic growth) are different for gold (e.g., real interest rates).