How to model windfall profits taxes

Commentary

In planning a new mine, engineers can draw on a wealth of long-established procedures to prepare a finely detailed study to track construction and operating costs down to the last litre of fuel and metre of pipe.

But when it comes to drafting a financial model for that new mine, the old tools may no longer be good enough.

In the face of rising commodities prices, a number of governments have introduced new royalties and taxes that allow them to increase their share of the high profits likely to be enjoyed by mining firms.

Some of these new tax instruments have tax rates that vary with commodity price or project revenue levels. The contingent nature of these taxes change a project’s economics in a way that can’t be reliably evaluated by conventional financial modelling.

The proposed Mongolian windfall tax on gold and copper revenues is a much discussed example. Windfall taxes provide governments with a share of any unexpected profits when prices rise above a predetermined level. In such regimes, a mine typically pays traditional royalties for the metal it extracts.

In Mongolia, a copper-gold mine would pay additional windfall taxes when copper prices are above US$1.18 per lb. and gold prices are above US$500 per oz.

The impact of these taxes on project risk and value are often misunderstood during the development stage of a mining project because of critical limitations to the conventional analytical tools used to model them.

In particular, static spreadsheet cash flow models relying on single-point price forecasts often produce major errors in estimating and valuing the additional costs that windfall taxes impose on a project.

A small but growing number of developers and governments have recently begun turning to Monte Carlo simulations to achieve more accurate valuations of mining projects.

One example is AMEC America’s work with Ivanhoe Mines, where Monte Carlo simulation is used to better understand the impact of the proposed windfall tax on the economics of the Oyu Tolgoi project.

In a Monte Carlo simulation (named after the famed casino in Monaco) you begin, for example, with gold at US$400 per oz. and map out a large number of possible price paths over the life of the project.

Most of the price paths will include prices in the intermediate range, though some will hit prices as low as US$200 and as high as US$800 per oz. The result is a more realistic assessment of the possibility of paying windfall taxes over the life of the project.

Let’s take a simple example of the difference between the existing valuation process and Monte Carlo simulation.

Consider a static spreadsheet cash flow calculation for a mine anticipating a taxable income in one year of US$1 million. At a 40% tax rate, the estimated total tax liability is US$400,000.

Now consider an analysis that builds in a mix of three different price scenarios for that same year:

* a low-price scenario, creating a loss of US$8 million and a tax credit of US$2 million;

* a medium-price scenario, resulting in taxable income of US$1 million and a tax liability of US$400,000; and

* a high-price scenario, producing a taxable income of US$10 million and tax owing of US$4 million.

Assigning a probability of 33.3% to each price scenario produces an estimated tax liability of US$800,000, even though the expected taxable income is still US$1 million. The more realistic tax figure is double the estimate from the static cash flow model, even without worrying about windfall taxes.

In all three scenarios, the tax rate was a constant 40% and windfall tax didn’t enter the picture. Now imagine generating many price scenarios with Monte Carlo simulation with a range of input data including a full description of corporate income-tax calculations, projected years of operation and, of course, any extra payments due to breaching a windfall-tax price boundary. The result would be an even more realistic assessment of tax payments than those determined by a static cash flow model.

In order to test these principles in a real-life way, we investigated how the economic viability of an existing Canadian copper-gold project would be affected if it was actually located in Mongolia and exposed to the proposed windfall tax (a paper co-authored with Dr. David Laughton discussing this example is available upon request from the authors).

A comparison between a static spreadsheet cash flow model and a Monte Carlo simulation showed that:

* the static spreadsheet analysis underestimated cumulative windfall tax payments by US$236 million when compared with the Monte Carlo simulation because prices in the spreadsheet model are never high enough to trigger the windfall tax; and

* the static spreadsheet model correspondingly overestimated the present value of the project by US$115 million, in part because it underestimated the normal corporate income tax, and in part because it assigned no lost value arising from possible windfall tax payments.

The math for all this has been around since the 1960s and has proven itself. However, there was little opportunity to apply it as a practical project valuation tool until the advent of powerful and inexpensive computing power. Now, PCs can run complex simulations with Monte Carlo spreadsheet plug-ins costing as little as US$800.

And judging by the potential to better estimate tax exposure, it’s money well spent.

— Graham A. Davis is a professor in the Division of Economics and Business of the Colorado School of Mines. Michael R. Samis is Director of Financial Services (Mining and Metals) at AMEC.