In this second part of our updated gold price framework we take a deep dive into the true energy exposure of gold mining companies. We find that gold miners are not just exposed to significant direct energy costs such as fuels and power; their indirect energy exposure is even larger. Our bottom-up analysis shows that ~50% of production costs of the average gold miner are closely linked to energy prices. This is in line with the findings of part I of our gold price framework which showed that a 1% change in longer-dated energy prices impacts gold prices by about 0.5%.
In the first part of this report we reviewed the gold pricing model we introduced last year and developed it further. We highly recommend reading it here to get a better understanding of the findings presented in this report. Using econometric tools, we showed that changes in energy prices - more specifically longer-dated oil prices - are a major driver for changes in the USD/gold price (with changes in real interest rates being the other main driver). In the second part of this report we take a more qualitative and comprehensive approach by examining the true energy requirements of gold mining.
The statistical analysis of gold prices in the first part was complicated by the fact that price data is often hard to get, incomplete or sometimes non-existent. However, the same problems are compounded when it comes to actual data on energy consumption by the gold industry. Gold producers are not required to disclose their energy consumption. And even if they were, it is far from clear what they would have to report and whether they would even have the data.
Data on direct energy costs in the form of fuels and power is relatively easy to find. It becomes obvious that gold mining is energy intensive when looking at the direct energy exposure, that is the fuel and power consumption of gold producers. Even comparably simple open pit mining consumes a lot of fuel for trucks and excavators and underground mining consumes electricity for cooling in addition to that. The processing of the gold ore is also highly energy intensive. Most large gold mining companies report these direct energy costs in one way or another. Typically, these reported costs are somewhere around 15-25% of all-in operating costs at current energy prices but have been higher in the past when energy prices were higher than today.
But gold mining is also energy intensive beyond the diesel and electricity that is consumed to mine and process the gold and cool the underground mines. Gold mining requires a lot of energy intensive resources and materials such as steel, chemicals, cement and tires and also machinery (trucks, excavators, mills) which consume energy in the manufacturing process. Ultimately, even wages partially reflect energy costs as changes in energy prices affect the living costs such as housing and food of workers. We call these the indirect energy costs. We presume that the companies don't even know themselves what the true energy exposure of these costs is. However, for this report, we meticulously dissected the expense side of the income statements of the largest gold producers in the world in order to estimate how much these indirect energy costs drive the expense side of gold mining.
Our bottom-up analysis shows that ~50% of production costs of the average gold miner are closely linked to energy prices. This is in line with the findings of part I of our gold price framework which showed that a 1% change in longer dated energy prices impacts gold prices by about 0.5%.
How does this bottom-up analysis tie in with our gold pricing framework? We used our gold price model to analyze how much of the major moves in gold prices since 2001 was driven by changes in longer dated energy prices. Our model predicts that USD610/ozt of the total gold price increase from 2001-2008 was driven by rising longer dated energy prices. In contrast, using the results of our bottom-up analysis that energy costs account for about half of gold production costs, the move in energy prices would explain USd570/ozt. We get to similarly close outcomes for other major upside moves as well as downside moves (see Figure 1).
Neither our top down gold price model nor the bottom-up analysis presented in this report should be misconstrued as a trading tool. Instead, it is a very useful framework for determining where are in the gold price cycle. Energy prices are clearly one of the most important drivers for gold prices. The bottoming of longer dated energy prices therefore has reinforced the floor for gold prices set by real-interest rates. We believe that the low in longer-dated energy prices is behind us and will likely have to move sharply higher over the long run (will dive deeper into this in the third and last part of this report). This, combined with the view that real interest rates have little upside but a lot of downside from here, plus the persistent risk of new and untested unconventional easing measures by central banks, leave the outlook for gold prices skewed sharply to the upside. In other words, there is a strong asymmetry for the gold price outlook.
The first section of this report looks at the direct energy exposure of gold miners. In the second part, we analyze the indirect costs, using the World Gold Council's All-in sustaining cash costs (AISC) framework. In the last section we discuss the results.
The Direct Energy Exposure Of Gold Miners
The CDP (formerly known as Carbon Disclosure Project) is a UK company that collects self-reported data from over 3000 large companies. Among other data, companies are asked to report their detailed energy consumption. The mining companies typically report their direct energy consumption for electricity (from the grid) and fuels, whereby the latter is often broken down into the different type of fuels consumed (some of it also to produce off-grid electricity). We analyzed data for the 25 largest pure gold producers in the world (gold output being 50% or more of annual output by value). Of the largest 25 publicly traded gold producers in the world, accounting for about 36% of total annual mine supply, 18 reported to the CPD for 2016. Polyus Gold (OTC:PGILF) was the only large gold producer not reporting to the CDP. However, the company reports its energy consumption independently in the 2016 annual report. The result can be found in table 1.
The data allows to calculate roughly what the energy intensity of the individual companies by dividing the total direct energy consumption in one year by the amount of gold mined. On average, the top gold miners need about 8.3 Terajoules (TJ) per ounce of gold1. However, there are some caveats;
The companies report energy consumption in the form of fuels and electricity. The electricity consumption data is broken down into electricity that is purchased, electricity that is produced from renewable sources and electricity that is produced from non-renewable sources. The latter thus comes from power production based on the fuels the company consumes, which it reports separately. In addition, all energy data is reported in Terajoules. Fuels such as petroleum products and coal have an energy content called "calorific value." It's the amount of energy that is in a barrel of oil or a tonne of coal. However, when fuel is used to produce electricity, a lot of energy is lost in the process. Hence, a gold mining company that is producing its own electricity will consume a lot more energy in the form of fuel to produce the amount of energy in electric form it reports to the CDP. In other words, two companies could use the same amount of energy per ounce of gold mined, but because one produces its own power and the other purchases it from a 3rd party, the latter company would seemingly consume less energy. Hence, the data is not suitable to compare energy per ounce mined between companies, but rather to look at an average energy consumption across the industry.
Hence, the data itself has to be handled and analyzed with care. However, we found that once we translated the energy consumption into costs, the data matched largely with how much the companies reported their energy costs indeed are, both to the CDP (only the very largest companies reported those estimates) as well as in their annual reports.
In addition, the gold mining companies we analyzed tended to use a similar energy mix over time. Hence, even as the data does not allow for comparing the energy consumption of different companies with each other, it does allow one to analyze the energy consumption per ounce mined of a company over time. At this point we only have comprehensive data for a few producers (luckily the largest) and only going back to 2011. (Earlier data is available but, in many cases, it is inconsistent with the current reporting framework). What the data shows is that, despite the continual efforts of gold producers to reduce their energy consumption, the amount of energy consumed to produce an ounce of gold has been trending up (see Figure 2).
The data reveals that average energy spent per ounce of gold produced has been going up from 2011 to 2013, declined in 2014 and has since rebounded and reached a new high. In our view this reflects mainly two things:
1. Rising energy intensity is consistent with declining ore grades in the mining industry (copper, zinc, nickel, as well as oil production where non-traditional forms of oil production such as oil sands or shale oil are much more energy intensive per barrel of oil output than conventional oil fields). This also makes intuitive sense as the mining industry should be exploiting cheaper sources before moving on to more expensive ones and given the importance of energy for operating costs. There is naturally a link between cheap/expensive and low/high energy intensity. Hence, we expect that over the long run, the energy intensity of gold mining will increase.
2. Rising gold prices incentivizes energy intensive mining and vice versa. Very energy intensive ore bodies that were uneconomical to mine in the past became suddenly economical when gold rose to USD1670/oz on average by 2012. In 2013, prices began to correct sharply from their highs which in turn made mining these ores uneconomical again which probably led to increased high grading. As a result, the average energy consumption per ounce of gold mined declined slightly. However, as prices have since resumed their long term upward trend, we expect gold miners to increase their energy consumption again in order to maximize output.
As time goes by and we get more data, it will be interesting to see how energy intensity evolves over the long run. On one side, we expect ore grades to continue to decline; on the other hand, gold miners will continue to drive efficiency.
The cost of direct energy exposure
Most companies report their direct energy costs directly to the CDP. On average, it's around 20% of total operating costs for 2016, the most recent reported year. However, in previous years, when energy prices were higher, some companies reported direct energy spending of 30% or even more of total operating costs. We then tried to estimate energy costs based on the reported energy consumption data. Because of the very diverse energy mix of each company as well as huge discrepancies in local commodity prices (e.g. a ton of coal delivered to a mine in Australia has a very different price then a ton of coal in China), any calculation will inevitably fall short of the real costs a miner has. As a proxy, we used the average cost of a barrel of Brent oil and multiplied it with the energy consumption of each company. In most cases, the results are in the same ballpark as what the companies have reported themselves and on average both were around 20% in 2016 (see Table 3).
Importantly, these are the largest gold producers in the world. They are not the fringe producers that operate at the margin. What really matters for the price of gold are the highest cost producers. If gold prices fall below the production costs of the producers at the margin, these will reduce production and gold output will decline. Hence, if world gold output is to continue to grow at about 1.5-2%, which it has done for decades, prices need to be high enough for that last once to come out of the ground. Unfortunately, small producers tend not to report their energy consumption. We could not obtain any data for China for example, which, as a country, is the largest gold producer in the world at this point. It reasonable to assume that many of these companies have higher energy exposure than the largest traded gold producers we analyzed in this report.
The importance of oil for direct energy costs
In our gold price model, we presented in part I of this report, we found long-dated oil forward prices deliver the best results in a multiple regression analysis, meaning that the t-value in the regression indicated high significance and it improved the fit of the model better than any other energy price variable. We have already outlined above why longer dated energy prices matter for gold rather than spot prices. But why do oil prices work so well in the model even as gold miners use a variety of energy sources? In our original framework piece, we argued that oil simply tends to be a good proxy for overall energy prices. This was consistent with previous work we had done.
For example, China is the most important player for steel and aluminum. The manufacturing of these materials requires vast amounts of energy which in China is mostly coming from coal-fired power plants. However, we have found that coal prices in China move with oil prices. The reason is that while China produces a lot of coal, it still must be moved from the producing to the consuming areas, which is done by rail, and those trains run on diesel. The price of imported coal is further impacted by seaborne shipping costs, which is highly corelated to the price of oil. Hence our findings that longer-dated oil prices can explain a large part of changes in gold prices didn't come as a surprise to us, even as we didn't exactly know how much of the energy costs from gold miners were directly linked to oil vs all other energy forms. In addition, we suspected that for smaller suppliers, which are likely among the marginal producers (meaning the producers with the highest costs and thus the most relevant for replacement costs for gold), oil would be the dominant form of energy consumed. The large capital outlays prevent smaller operations in remote areas from being connected to the grid and thus they are more likely to rely on oil even for electricity production. This is consistent with other mining operations such as copper, zinc or nickel.
But when we analyzed the CDP data, we were surprised to find that even among the very large companies, oil is by far the largest source of direct energy consumption. Oil and oil products (mostly diesel and fuel oil) accounts for about 50% of total direct energy consumption for the top gold producers. Grid electricity is only 33%; natural gas and propane account2 for 12% and coal for 9%. On top of that, part of the grid electricity also originates from oil-fired power plants, especially in remote areas. Hence the share of oil in mix is likely even higher. On net, oil is by far the largest source of direct energy use in the gold mining industry and it will remain so for the foreseeable future. The full breakdown in the mix of direct energy consumption can be found in figure 3.
The Indirect Energy Exposure Of Gold Miners
While the CDP database provides comprehensive data for the direct energy exposure of gold miners, the indirect exposure is much harder to find. There is no third-party database that would deliver any answers. Hence the only source of information are the financial statements and company presentations of gold producers. However, in their financial statements, gold mining companies typically report a myriad of different cost types and they often differ between companies. Most of them don't reveal anything in terms of energy exposure. Fortunately, the largest producers do or at least have done so in the past.
Cost measures reported by gold producers
In a first step we look at how gold producers typically report costs. Encouraged by the World Gold Council (WGC) a few years ago, gold producers began to report what is called "All-in sustaining cash costs (AISC)". According to the WGC, AISC is a "non-GAAP measure which are intended to provide further transparency into the costs associated with producing gold. The "all-in sustaining costs" is an extension of existing "cash cost" metrics and incorporate costs related to sustaining production. The "all-in costs" includes additional costs which reflect the varying costs of producing gold over the lifecycle of a mine." And further: "It is up to individual companies to determine how they report to the market and to decide whether their stakeholders will find these new metrics of value in understanding their businesses."
The WGC also published detailed guidance for calculating AISC. There are 17 cost items that together add up to AISC. Items (a-k) add up to operating cash costs and adding (m-r) leads to AISC. It is not mandatory for gold producers to publish each item individually. In fact, most producers publish their own individual breakdowns, which tend to be less granular than the WGCs guide-lines. Though the way gold miners break down AISC differs from company to company, it typically consists of: operating cash costs (a-k), sustaining CAPEX (q+r), general and administrative expenses (m), exploration and evaluation (o+p) and there is usually also an "other" (n) category (see Figure 4)3.
In addition to AISC, the WGC also encourages producers to report All-in costs, which in addition to AISC contains so-called "non-sustaining" costs as well. There are 7 more cost items in the WGCs guidelines (t-z) that, combined with AISC, lead to All-in costs. As most of the largest producers report AISC at this point, and quite a few report All-in costs in addition to that, we can work along this framework to evaluate the extent to which both direct and indirect production costs for gold miners are driven by energy prices.
Operating Cash Costs
Operating cash costs is the largest cost block of AISC (items a-k )4. Broadly speaking, operating cash costs are the costs associated with the production of gold. It includes things like salaries and refining and transportation and usually excludes costs such as office administration and overhead costs as well as interest, depreciation and amortization.
While none of the companies reported items a-k individually, some of the companies break down operating cash costs further. Unfortunately, each company has a very specific way to report the breakdown of its operating costs which are mostly incompatible with each other. For example, one company reports contractor costs separately while the other reports part of it as labor and part of it as site maintenance costs. In order to assess operating costs in a consistent manner across the gold mining industry we grouped them into the following categories: Labor, Fuel & Power, Consumables, Maintenance and Other (see Figure 5).
The direct energy consumption reported to the CDP earlier in this report fits mostly into this fuel and power category. However, not all the direct energy consumption of a company will show up here. Particularly "Sustaining CAPEX" reflects a lot of direct energy input as it includes production phase stripping costs at open pit mines and underground development. Some direct energy consumption is likely reported under exploration and evaluation and to a smaller extent G&A. Regardless, we can safely conclude that 100% of what companies report as fuel and power costs is driven by energy prices.
This still leaves the question what the (indirect) energy expenditures within the other categories of operating cash costs are. As explained before, the exact figures of indirect energy exposure can't be found anywhere. However, using data provided by the gold miners themselves in their annual reporting and other presentations, we were able to dissect these categories further and derived an estimate for the average energy exposure of each category. In the next sections, we analyze the remaining four cost blocks (consumables, labor, maintenance and other) of operating cash costs with the goal to identify how much of these costs are driven by energy prices. Later in this report we will then do the same for (non-operating) all-in sustaining cash costs.
Operating Cash Costs: Consumables
Consumables account for 20% of operating costs. Consumables encompasses a wide range of goods used in the mining process: Grinding media, explosives and other chemicals such as nitric and sulfuric acids and sodium hydroxide, cement, cyanide, lime, tires and lubes. The production of a lot of these materials is heavily intertwined. For example, lime is used not just in the gold leaching process but also in the production of ammonia nitrate, which is used as explosives to drill and blast limestone, which in turn is the source material for lime. Lime is also an essential ingredient in steel manufacturing, the source for grinding material, which is used to grind the limestone. As we will show, all these materials consume a lot of energy in their respective manufacturing processes, but they also consume a lot of each other, which means the true energy content is likely higher than just the energy consumed in the individual processes.
Grinding Media are basically steel balls or alloys that are used in the grinding process of the ore rock. The manufacturing of the steel balls is very energy intensive. According to the World Steel Association, about 20-40% of the costs in the steel making industry are from energy. According to the International Steel Statistics Bureau (ISSB), a ton of steel consumes about 18GJ in the manufacturing process, equivalent to about three barrels of oil. Steel is currently trading at around USD600/tonne and converted to barrels of oil, 18GJ is about three barrels worth USD240 or 40% of the cost of steel at current prices. Normally steel mills don't run on oil but a cheaper form of energy5, but this confirms the higher energy intensity of steel manufacturing. But energy is not just consumed in the manufacturing process itself. Costs for raw materials (including transportation) such as iron ore, scrap steel, ferroalloys, lime and oxygen - all highly energy intensive - whereby iron ore accounts for the bulk of it. Iron ore production is very energy and labor intensive by itself. According to estimates from Goldman Sachs (GS) equity research, the costs of iron ore to steel producers can be broken down into 30% labor costs, 15% fuel and power, 25% transportation costs, 15% other costs and 10% port fees. The latter reflects energy and labor costs at the port.
Iron ore mining is often conducted in countries with low wages. We explore how energy affects wages in more detail in the next section but for the time being we assume that roughly 40% of the labor costs in iron ore mining are tied to energy prices. On net, we estimate that iron ore costs for steel producers reflect roughly 50-55% energy costs. And energy is not just present in the manufacturing of steel but also in the transportation of the grinding media to the mine. On net we estimate that the costs of grinding media at the mine is driven 60-70% by energy. According to IHS Markit, the most common Explosive Agent used in the mining industry today is ammonium nitrate (AN), mostly in the form of ammonium nitrate fuel oil (ANFO). ANFO consists of 94% AN and 6% fuel oil. Ammonium nitrate 34-0-0 is upgraded NH3 and Nitric thus about 85% of the manufacturing costs reflect energy according to our analysis. The ammonium nitrate used in mining is slightly different from fertilizer in that is less dense (it contains about 20% air) but otherwise they are very similar. Given that the 6% fuel oil is pure energy, we conclude that the energy costs of ANFO are slightly over 85%.
According to the US Energy Information Administration (EIA), the Cement industry is the most energy intensive of all manufacturing industries. Data from the department of Innovation, Science and Economic Development Canada (ISED), direct energy costs account for over 41% of the manufacturing costs in the Canadian cement industry. Materials and supplies accounted for another 40% of costs. Limestone is the single largest component of industrial cement. The mining of limestone itself is again very energy intensive and requires explosives (more to that below). The remaining 18% of costs are labor costs. However, it is reasonable to assume that in low income countries such as China, labor costs are a lot lower and energy and materials account for almost all of the costs. Further, the transportation of the material to the mines is a significant part of the costs for cement for miners and are not part of the ISED breakdown. Other sources show that transportation costs for cement are over 10% of total costs and for remote mining operations that number is likely a lot higher. On net, we estimate that cement costs reflect 65-85% energy costs.
Cyanide is used in the gold mining industry to dissolve and separate gold from ore in a process called cyanide leaching. Cyanide leaching has largely replaced extraction with liquid mercury and has become the global standard method. Most cyanide leaching uses a cyanide sodium solution. To manufacture cyanide sodium, hydrogen cyanide is treated with sodium hydroxide. Hydrogen cyanide itself is usually manufactured from methane and ammonia in a process called Andrussow oxidation in which the methane and ammonia react at very high temperatures in the presence of oxygen. Methane is natural gas and ammonia is made from natural gas. According to Chemlink PTY, an Australian chemical consultant, 1 tonne of cyanide sodium requires about 34MMbtu of natural gas, 0.65 tonnes of ammonia, 0.95 tonnes of caustic soda and 0.67MW of electricity. According to our calculations, at current prices the energy ingredients make up over half of the sales price of cyanide sodium and thus well over half of the production costs. Again, part of the costs for gold miners is also the transportation of cyanide sodium to the mine. We estimate energy reflects accounts for about 75% of the costs for cyanide for gold miners.
Lime is extensively used in the leaching process. Once the gold ore is grinded, cyanide sodium is added and enough lime is used to keep the pH of the solution at about 11.0 and suppress the formation of toxic hydrogen cyanide gas. Lime manufacturing, like cement production, requires a lot of energy. The feedstock for lime production is rock containing calcium carbonate. Most lime production uses limestone as feedstock, but chalk and marble also contain calcium carbonate. Limestone is extracted from quarries or mines by drilling and blasting, extracted by bulldozers and power shovels and finally trucked to a lime kiln. According to the US Environmental Protection Agency (EPA), rotary kilns account for about 90 percent of all lime production in the United States. A rotary kiln is a "long, cylindrical, slightly inclined, refractory-lined furnace, through which the limestone and hot combustion gases pass countercurrently." There are two different processes to turn limestone into lime, both require very high temperatures (900-1800℃), which explains the high energy footprint. The EPA states that "Lime production is extremely energy intensive. Assuming perfect efficiency, producing a ton of lime from pure calcium carbonate requires 2.77 million Btu. In practice, the process is considerably less efficient." The 1996 Annual Survey of Manufactures showed that 31.4% of the costs of lime manufacturing is coming from fuels. As energy costs have increased significantly since 1997 (oil was trading at USD17/bbl in 1997 vs USD75/bbl today), energy costs likely account for a larger share of costs today. This is consistent with data from Statcan from 2012 showing that 35% of the costs for Canadian lime producers are from Energy, Water and Vehicle Fuel and a more recent report from the Canadian lime institute showing direct energy cost account for about 40% of the costs for lime making.
However, this is just the costs for fuel and power associated with the lime manufacturing in the lime kiln. As with other industries, materials and supplies also account for a large part of the total costs. According to the same 1997 US census bureau data, 68% of the end costs are for materials, ingredients, containers and supplies where the cost for limestone and other rocks and glass is the largest cost factor, accounting for about 1/3. Limestone is a lot cheaper than lime (roughly USD10/tonne vs. >USD100/tonne for lime) but according to the USGS, one tonne of lime requires roughly 2.5-4 tonne of limestone. The EPA lists the 3 main ingredients for limestone production as electricity, fuels and water. USGS data from 1997 breaks down the production costs of limestone. Direct energy accounts for about 12% of costs, but including the energy content of explosives, rubber, plastics as well as machinery and to some extent wages show that energy accounted for roughly 40% of the costs of limestone production. Again, energy costs have gone up significantly since 1997, specifically fuel costs which means the energy costs are likely >50% of the costs of producing limestone today. On top of that, the cost of lime for gold producers consists also transportation costs from the lime manufacturer to the mine. On net, we estimate that energy accounts for about 65-70% energy costs.
Tires are an important cost factor for gold miners. Excavators and mining trucks require special tires - so called off the road tires (OTR) or earthmover tires that can be as tall as a house and weigh over 6 tonnes. OTR tires consist of 76% carbon based material, most of it rubber. There are two types of rubber, synthetic and natural rubber (latex) coming from trees. Natural rubber consists of up to 95% of cispolyisoprene, which can also be created synthetically from petroleum. Between 60-70% of all rubber produced is synthetic; the rest is natural rubber. Most of the natural rubber is consumed in tires manufacturing and tires typically are made from a blend of both. Other materials use for tire making are carbon, zinc oxide, textiles and stearic acid. According to a 2007 technical paper from Otraco, a subdivision of the Australian Downer group, a company that provides engineering and infrastructure management services, rubber contributes 60% of the weight of an OTR tire, steel wire 20%, carbon 10% and the remaining 10% is from other materials such as processing chemicals. From an energy perspective, most of what goes into a tire can be considered energy in some form. And unsurprisingly, tire-derived fuel has about the same heat value per pound as heavy fuel oil. However, even though a tire itself might be considered pure energy, the costs of making tires for the mining industry are not entirely driven by energy prices.
Some of the OTR tires used in mining are far from the standard mass-produced tires. They are precision made, requiring a lot of manual work. The largest tires cost close to USD50'000 and are almost hand made. Hence tire prices reflect a lot of other costs as well, particularly labor. According to the Otraco paper, 40% of the manufacturing costs comes from raw materials, 40% are conversion costs and 20% are service and administration costs. These ratios fluctuate over time, mostly because of fluctuations in the costs for raw materials. The numbers above reflect average numbers over a period of 17 years from 1990-2007. In 2007, when the paper was released, raw material prices had risen so much, particularly rubber, that raw material costs accounted for most of production costs. The paper also states that this does not include freight and handling costs which can add 15% or more to the total. We assume that the costs for raw materials are 90% energy, for freight and handling 75% energy. The 20% service and administration costs probably don't move much with energy prices. That leaves the 40% conversion costs. Conversion costs are manufacturer's production costs other than the costs of raw materials. They consist of direct labor and manufacturing overhead costs (including the energy needed to make the tires). Data from Michelin (OTCPK:MGDDF), the world's second-largest tire producer, shows that labor costs account for about 25% of sales revenues. As OTR tires are more labor intensive than regular tires, it is probably fair to assume that labor costs account for the bulk of conversion cost. We therefore assume that about 50% of the production costs are driven by energy prices. Including transportation, we estimate that the costs for tires for gold miners are driven to 60% by energy prices. The Otraco paper also shows that tire makers enjoyed much higher margins in the past, but those margins came under pressure going into the new century, in part due to new industry entrants in various emerging markets.
Tire prices initially did not reflect the massive price increases of rubber and petroleum from 2003-2006 because it simply led to an erosion in margins for the tire producers. Hence, the costs for tires for gold producers didn't rise as much as the costs for the energy that went in them in the past. However, the margins have now permanently contracted, and we would expect that tire producers will have to pass on future energy price increases in the selling price.
Lubes mainly consist of base oil plus some additives to give the lube the desired properties. For example, lubricants designed for internal combustion engines contain additives that reduce oxidation. The automotive sector is the largest source of demand for lubes. Industrial lubes are practically all made from petroleum and manufactured in refineries that produce motor gasoline and diesel fuels. Thus, lubes are 100% energy and reflect 100% energy costs in our calculations.
Other chemicals such as sulfuric acid and nitric acid are sometimes consumed in the processing of gold ore. Individually they make up a relatively small share of the total cost for consumables. Each chemical is different. Two that stand out as they seem to be used more extensively are nitric acid (NO3) and sulfuric acid (H2SO4). Nitric acid is manufactured from ammonia at high temperatures and thus the manufacturing costs are largely driven by natural gas prices. According to Johnson Matthey, a British specialty chemicals and sustainable technologies company, 85% of all nitric acid is used to make ammonium nitrate (AN) (we covered ammonium nitrate earlier under explosive agents as AN is the main ingredient for ANFO explosives). Sulfuric acid, on the other hand, is manufactured from sulfur (burned to become sulfur dioxide) and oxygen. Oxygen is produced through air separation in an oxygen plant and is usually highly energy intensive. Sulfur itself is mostly produced as a by-product from natural gas and petroleum production. In the hydrocarbon industry, sulfur is a contaminant that must be removed. While this is highly energy intensive process, it is difficult to put an energy tag on sulfur for several reasons. The sulfur would have to be removed from gas and oil even if there wasn't a market for it. Further, during the first step in the manufacturing process of sulfuric acid, the combustion of sulfur releases a lot of heat that can be recycled and used to melt more sulfur or power turbines for electricity.
What matters for this report is how much changes in energy prices impact the price of consumables used in gold mining. Without diving too much into the economics of sulfuric acid manufacturing, we assume that because the process to remove sulfur from crude oil and gas requires a lot of energy, energy price fluctuations do impact sulfur price at least to some extent. But the costs of H2SO4 are much less driven by energy prices than NO3. Besides NO3 and H2SO4 there are also other chemicals that are used in gold mining, some with very high energy costs like AN and some with less. For this report we assumed that the costs for other chemicals category (which account for only 4% of the consumables expenditure for the average gold miner) is 60% driven by energy costs.
We have used information of the largest gold producers to break down the cost for consumables into the individual items. The results can be found in table 2.
It becomes clear that consumables are driven primarily by energy prices. Over 70% of what gold miners spend to purchase and transport consumables to their mines are driven by energy. Consumables account for 20% of operating costs and thus energy accounts for 14%, almost as much as direct energy spending on fuels and electricity.
Operating Cash Costs: Labor
The largest block in operating costs is labor, accounting for 39% of total. Estimating the energy price impact on labor costs is the most difficult task. One can argue that labor, by definition, is energy: human energy. However, what matters for our gold pricing framework is whether labor costs for gold miners move with changes in energy prices. In our view, there is no straightforward calculation to estimate how much labor costs are driven by energy prices as it was for consumables, let alone direct energy costs. We find that the most sensible approach is to look at the share of energy, transportation and food in the respective CPI Baskets of gold producing countries6. However, we think it is important to distinguish between energy spending in developed markets and emerging markets.
Looking at developed markets, few people would argue that wages instantaneously react to changes in energy prices. For example, employers rarely compensate their employees if oil prices rise from USD50/bbl to USD100/bbl. Instead, workers will have to spend a higher share of their income for energy and transportation, and to some extent food as costs in the agricultural sector are strongly driven by energy prices. However, rising energy prices will - over time - inevitably lead to a general price inflation, which then in turn will spark demand for higher wage compensation. Hence energy prices eventually do impact wages in developed countries, but the impact is rather slow. Therefore, we adjusted the CPI basket weights for energy, transportation and food in developed markets to reflect that. More specifically, we adjusted the energy weighting by 50% and transportation and food by 75% to reflect that these costs are not just driven by the underlying raw materials and that there is a time lag between a change in energy prices and the impact on wage costs. As a result, we estimate that less than 8% of labor costs in gold industry in a developed country are impacted by changes in energy prices.
In developing countries however, we believe energy has a much higher and much more instantaneous impact on labor costs. Due to their lower incomes, workers in developing countries spend most of their salaries on basic goods such as housing, food and energy. Rising energy prices will therefore put much more upward pressure on labor costs than in developed countries. We find that, on average, energy, transportation and food accounts for about 48% of expenses according to the CPI baskets of the largest gold mining nations. This is more than 6 times as much as workers in developed markets. (This excludes meals abroad as well as alcohol.)
Hence, on per ounce basis, workers in the 12 largest producing countries have about 40% of their spending directly tied to energy prices (see table 3). This calculation probably underestimates the true energy impact on the cost of living of the average worker in the gold mining industry. 30% of the output of the 12 largest producers occurs in developed markets. However, on a global basis it is only 22%. Because we weighted the energy impact on wage costs with the respective output data, our calculation probably understates the true energy impact on labor costs. Another reason why the energy impact on labor cost is likely higher than our 40% estimate is the fact that gold miners heavily rely on contractors. Gold producers use contractors for a wide area of activities. In some cases, gold miners simply employ workers through contractors instead of employing them directly. In other cases, gold producers task contractors with running entire operations. In that case, contractors provide mining equipment, operators, maintenance and supervisory personnel, and infrastructure and they conduct complete mining services, including drilling, blasting and haulage. Hence while in the former case, the energy costs of contractors are the same as for labor while in the latter case, the energy costs of contractors likely resemble the operating costs of gold miners themselves.
Unfortunately, gold producers don't all report contractor costs the same way. Some of them just lump it in with labor costs, some of them break it down into labor, maintenance and other costs and some of them report contractor costs separately. And from the way they report it is often hard (and often impossible) to decipher exactly which of the three it is. Most companies seem to report it either as labor costs or break it out separately. We therefore decided to lump the two together if they are reported separately to get a consistent average breakdown, which is how we get to our estimate that labor costs account for 40% of operating costs. This also means that we are likely understating the hidden energy costs within labor costs at 40%. Why? If the producers that break out contractor costs separately are any indication, then contractor costs are roughly 30-40% of total labor costs. If we assume that the energy cost of contractor work is close to the 54% average in operating costs, the energy impact on labor costs for gold miners would be closer to 45%.
Operating Cash Costs: Maintenance
The next cost block in operating costs is maintenance costs, accounting for 11% of the total. Maintenance costs are costs that arise from maintaining the working status of existing equipment and plant facilities that are not otherwise able to be capitalized. Thus, maintenance costs are mainly labor costs of nonproducing workers; for example, mechanics who fix and service equipment such as excavators, trucks conveyors and processing equipment. This may include costs for contractors doing this type of work. Maintenance costs also include spare parts and may include tires that aren't capitalized. For simplicity, we assume that maintenance costs have roughly the same energy exposure as labor costs. Operating Cash Costs: Other
Gold miners report all operating cost that don't fit into the four categories above as other costs. This includes royalties which make up a large chunk of other costs. For simplicity, we assume that the "other" category is not driven by changes in energy prices.
Conclusion: On net, we find that energy costs drive roughly 54% of operating cash costs. The complete breakdown can be found in table 4. This shows that direct fuel and electricity expenditures are only a fraction of the total costs. Fuel and power is still the largest energy cost block for gold miners, but there are significant hidden energy costs in other areas that add up.
Other Sustaining Cash Costs
All-in sustaining costs (AISC) is an extension of the existing operating cash cost metrics and incorporates costs related to sustaining current production. In addition to that. The largest share of AISC is operating costs (70%) which we already analyzed in detail above. The WGC methodology breaks down the remaining costs into7:
m) Corporate General & Administrative
n) Reclamation & remediation
o) Exploration and study costs
p) Capital exploration (sustaining)
q) Capitalized stripping & underground mine development (sustaining
r) Capital expenditure (sustaining).
However, while some companies do report each of these cost items individually, many gold producers only report a more aggregated version. Every company does it slightly differently, but generally most companies that don't report exactly m-r do report their non-operating costs along the four categories below: General and administrative, Exploration and evaluation, sustaining CAPEX and other. These are the same four categories which we have already introduced earlier and can be seen in figure 3. We then mapped the cost items of the AISC framework onto these four categories:
1. General & Administrative (m) - 5%
2. Exploration and evaluation (0 + p) - 4%
3. Sustaining capex (q + r) - 18%
4. Other (n) - 3%
The percentages indicate the share of AISC of the average gold mining company. As companies report differently, some of the items m-r will inevitably be in different categories than what we assume for some companies. However, we believe the way we mapped the four categories onto the AISC framework is the most sensible approach. We then tried to estimate the share of energy related costs within these categories. Below we outline our thinking process for each category.
For simplicity, we assume that the impact of energy prices on (1) G&A is zero. This undoubtedly understates the importance of energy for G&A as administrative work certainly requires electricity, and, as we have outlined earlier, labor costs are somewhat linked to energy prices. However, electricity costs are probably a negligibly small share of G&A, and while wages are probably much larger share (we assume the lion's share of G&A costs), we believe most of the workforce in G&A lives in developed countries where the link between wages and energy prices is much weaker, as explained earlier. Thus, for simplicity, we assume that that energy costs are mostly irrelevant.
According to the Aboriginal Affairs and Northern Development Canada (AANDC), (2) exploration and evaluation consists of the following steps:
2. Recording of Mineral Claims
a. Prospecting and geological mapping
b. Rock sampling
c. Geophysical surveys
d. Geochemical surveys
4. Advanced Exploration
a. Stripping and Trenching: This can include the use of heavy equipment and blasting
b. Drilling: According to the AADNC, drilling is the most expensive part of exploration and "on average accounts for 50% of total exploration spending"
5. Sampling and Assaying
6. Economic Evaluation
Stripping, trenching and drilling are where most of the costs in the exploration and evaluation process occur and those are also the activities that consume a lot of energy. Stripping and trenching can be compared to the normal operational cash costs of a mine less the processing. The drilling process can include loader trucks, diamond drills, rotary drills, percussion drills and drill boom jumbos. Energy is required to some extent for transportation of the equipment but mostly to power the drill. The energy to run the drills can either come from electricity or diesel. According to 2007 study from the US department of energy, on average drilling consumes 5% of the total energy consumed in mining in the US. Data from Infomine shows that energy in the form of fuel and lubes account for close to 60% of operating costs for a diesel powered rotary drill (crawler). Labor accounts for only 14% and wearing of parts for 18%. Overall, we estimate that roughly 30-35% of the costs for exploration and evaluation are driven by energy prices.
(3) Sustaining capex is the largest non-operating cost block within AISC. It contains capitalized stripping & underground mine development and capital expenditure. For simplicity, we assume that capital expenditure is not driven by energy prices. Stripping and underground mine development encompasses similar activities as normal operations of the mine less processing. It entails drilling, blasting and hauling among other activities. Unfortunately, none of the gold producers breaks down sustaining CAPEX costs. For this report we assume that half of the sustaining CAPEX costs are capital expenditure and the rest capitalized stripping & underground mine development. We, therefore, estimate that sustaining CAPEX is to 35% driven by energy costs.
Reclamation and remediation (we summarized it as (4) other costs) encompasses costs that occur from efforts that are aimed at minimizing the impact of a mine closure to the community (social closure) as well as the environment (environmental closure). Gold producers try to minimize the impact on communities by providing education and out placement services if workers can't continue working for the company at another location. For simplicity, in this report we assume social closure costs aren't impacted by energy prices.
Environmental closure however includes sloping, capping and covering, and revegetating waste rock facilities, leach pads and tailings impoundments. These activities require heavy machinery and labor similar to operating the mine itself but without the energy consumption for ore processing. There are no hard numbers how much reclamation and remediation can be attributed to environmental closure. For simplicity, we assume half and we estimate that 40% of the costs driven by energy.
Conclusion: On net, we find that energy costs drive roughly 48% of AISC. Non-operating AISC generally have less energy exposure than operating costs, which means that the energy component of total AISC is slightly lower. This means that - on average - close to half of what gold miners spend to maintain current gold output is linked to energy prices.
With the AISC metric, the WGC aimed at providing further transparency into the costs associated with producing gold. However, all-in sustaining cash costs still don't reflect the entire costs that occur to a gold producer. AISC only reflects the costs associated with current output and the expenses to maintain this output in the future. But on top of AISC, gold miners also spend money to extend production into the future. This includes community, permitting, reclamation and remediation costs not related to current operations as well as non-sustaining capital exploration, capitalised stripping & underground mine development, capital expenditure and exploration & study costs. A closer look at these costs categories shows that they are basically the same as those listed in the AISC framework under adjusted operating cash costs minus G&A. Hence, we assume the same energy exposure for the respective categories.
However, as these are non-sustaining costs, do they even matter? In theory, the companies could simply cease to invest in new projects aimed at increasing output while maintaining all other spending and gold production would be stable. And for an equity analyst trying to evaluate the stock, this might actually be the way to look at it. However, we argue that for the price of gold, these costs do matter. While in theory, global gold production could be maintained at current levels if all the non-sustaining expenses are scrapped, this would result in a slowdown in the growth of above ground stocks, which would mark a departure from how the gold industry worked for hundreds of years.
We can illustrate this with an example: Imagine above ground stocks have been growing at 2% p.a. for the past hundred years and global above ground stocks are currently 100,000 tonnes8. Next year mine output would have to be 2000 tonnes for above ground stocks to grow 2%. But the year after, mine output has to be 2040 tonnes in order for above ground stocks to grow 2%.
But why is this important? In order to explain this, we have to recall some of the concepts we discussed in our first gold price framework (Gold Price Framework Vol. 1: Price Model, October 8, 2015), namely why gold is money. An important function of money is its store of value function. Historically, above ground gold stocks kept pace with economic activity over very long time periods. This is part of the reason why prices in gold have remained remarkably stable over time. If above ground gold stocks would suddenly lag economic activity, it would result in broad price deflation (of goods and services priced in gold). Or in other words, gold prices would rise, which in turn would incentive new mining. Hence even as gold's role as money has been ignored by central banks, governments and large parts of the population (at least in the western world), gold mine output seemed to have continued to grow at the pace needed to maintain this relationship. And as global economic output continues to increase, the growth of the gold stock (money stock) should continue to grow at the corresponding pace and not decelerate meaningfully, which means mine output should also grow. Hence, while all-in costs might not be that important for equity valuation purposes, it surely is important for gold prices.
That said, the difference between all-in costs and AISC is rather small. Analyzing the 20 largest publicly traded gold miners, we find that of those that report both AISC and all-in costs, non-sustaining costs are just about 8% of total costs (see Figure 6). For our calculation we assume that these non-sustaining costs have the same energy costs associated as the non-operating AISC.
Conclusion: On net, we find that energy costs drive roughly 47% of all-in costs. This means that by taking non-sustaining cash costs into account, the picture hardly changes. Still about half of what gold miners spend to produce and maintain their current output and ensure that global gold production continues to accelerate at the same speed as it did historically, energy accounts for roughly half of these costs.
Bringing It All Together
On net, we find that energy accounts for roughly 50% of production costs for gold miners. However, there are a few important things to note. First, cost breakdowns change over time. For example, direct energy costs were higher than 20% a few years back when energy prices were much higher than today. Similarly, the indirect energy exposure also changes with energy prices. For example, we determined that cost for consumables are by roughly 70% driven by energy costs. But as energy prices rise, we would expect that number to increase too. What makes things more complex is that changes in energy prices also change behavior. Lower prices might encourage more wasteful mining for example and vice versa. Thus, the numbers are a rough guidance and our estimate of a 50% energy exposure for gold mining is probably more 40-60%, depending on where energy prices are relative to other things.
Second, while changes in energy costs impact gold mining costs by around 50% over the short run, this rises towards as much as 100% over the long run as rising (or occasionally declining) energy costs feed through every aspect of the economy and lead to a general price inflation9. For example, when energy prices double, prices for a new excavator don't double overnight. Some parts of the production costs for the excavator go up instantly, such as the steel, the synthetic parts and energy used in manufacturing process. But overhead costs probably don't change that fast. Thus, excavators become more expensive to build, but not 100% more expensive. And the manufacturer might not be able to pass on his increased costs to the miners straight away. Over time however, higher energy prices will trickle into every aspect of manufacturing. For example, as energy prices rise, building a factory becomes more expensive which over the long run will have to be reflected in sales prices. Administrative costs go up as electricity, IT, phone bills, and wages and ultimately the price of every paper clip reflects higher energy prices. However, this may take years or maybe even decades to materialize.
What does that mean for gold prices? Gold prices only move in line with energy prices to the extent that gold producers are exposed to the instant cost change. For example, when longer dated energy prices double within a year and the average energy exposure is 50%, gold prices should go up by about 50%. But over the long run, gold prices will gradually reflect the entire price change. This is the reason why oil priced in gold has remained remarkably stable over very long time periods. In that context, it is important to understand that the reason energy prices - for example, oil prices - have gone up 6000% over the past 100 years is not due to oil being scarcer now than in was in 1918, it is because the unit we measure in (the USD) has depreciated 98%. Both gold and energy priced in USD must reflect that.
However, what we show in this report is that gold prices reflect changes in energy prices also when those changes are not monetary in origin. Gold prices do follow longer-dated energy prices when they rise because they reflect more scarcity in the future. And the same is true in reverse, gold prices follow longer-dated energy prices lower when they price more abundancy in the future, as for example happened from 2012 to 2015. How does this bottom-up analysis tie in with our gold pricing framework? In 2016 we published a report (Inverted Asymmetry - Gold Price Outlook, September 20, 2016) in which we used our original gold price framework to identify the price drivers of the gold cycles since 2001. The model indicated that the initial price increase from USD271/oz in 2001 to USD872/ozt in 2009 was largely driven by an increase in longer dated energy prices. We have repeated this analysis with our updated model and come to a very similar conclusion: The sharp increase in longer-dated energy prices drove about USD610/ozt of the move (see Figure 7).
From 2001 to 2008, longer dated oil prices (5-year forward Brent) went from USD20/bbl to USD103/bbl, a 5-fold increase (+420%). To simplify the calculation, we assume that half the price of gold in 2001 was tied energy, which would have been USD135/ozt. Hence, all else equal, a 420% increase in longer dated energy prices should increase long term replacement costs of gold by USD570/ozt. By way of comparison, as we have explained above, our model suggests that the change in longer dated energy prices increased gold prices by USD610/bbl.
Similarly, the decline in longer dated energy prices was responsible for about USD290/ozt of the USD510/ozt decline in the price of gold from an average USD1670/ozt in 2012 down to USD1060/ozt by the end of 2015, according to our gold price model. Over this time horizon, longer dated oil prices went from USD96/bbl to USD63/bbl, a decline of 34%. Again, assuming that energy accounted for about 50% (USD835/ozt) of the average price of gold in 2012, a 34% decline suggest that the decline in longer dated energy prices accounted for about USD290/ozt of the decline in the price of gold over this time frame.
Importantly, our bottom-up analysis should not be misconstrued as a trading tool to predict "energy price implied" gold prices. First, energy prices are only one of three main drivers for gold prices, the others being central bank policy (real-interest rates and asset purchases) and central bank net gold purchases. Second, because we estimated energy costs as a share of total costs, using the results of this bottom-up analysis would only allow us to estimate the impact of relative changes in energy prices if we knew exactly what the share of energy costs is at any point in time. Rather the purpose of this bottom-up analysis to illustrate that energy costs indeed account for a large share of gold mining costs as our top down model predicts. The simple calculation above suggests that the importance of energy price changes for changes of gold prices for both methods are fairly consistent.
As we have pointed out before, our top-down gold price model is not to be misconstrued as a trading tool either. Instead, it is a very useful tool for determining where are in the gold price cycle. Our gold price framework and the complementary bottom-up analysis in this report therefore are most useful in this context. Energy prices are clearly one of the most important drivers for gold prices. The bottoming of longer dated energy prices therefore has reinforced the floor for gold prices set by real-interest rates. We believe that the low in longer-dated energy prices is behind us and will likely have to move sharply higher over the long run (will dive deeper into this in the third and last part of this report). This, combined with the view that real-interest rates have little upside but a lot of downside from here, plus the persistent risk of new and untested unconventional easing measures by central banks leave the outlook for gold prices skewed sharply to the upside. In other words, there is a strong asymmetry for the gold price outlook.
1For simplicity, we ignored the output of other commodities which are produced as by-products. Taking the by-products into account would improve the ratio of energy consumed relative to the value of the minerals mined. However, despite this having a positive effect for the profitability of the respective miners, it doesn't change the fact that the current gold production volumes could not be achieved by deploying less energy than they currently do.
2Unfortunately, propane, LNG and natural gas are reported as a single net amount. The International Energy Agency (IEA) as well as the US Department of Energy (DOE) both count propane as liquid (propane supply fees into the oil balance not the natural gas balance), not the gas, and historically it has traded closer to oil prices.
3 We explain later in this report how we attribute items (m-r) to the four categories when we discuss other sustaining cash costs
4Operating cash costs are correctly referred to as "Adjusted Operating Costs" in the WGCs AISC framework. While simply the sum of (a-k), AOC are listed as an individual item, (l). Adjusted operating cash costs plus the remaining 6 times (m-r) add up to AISC. In this section we look at operating cash costs only, we will discuss the remaining items (m-r) in the next section.
5For example, electricity produced in a coal fired powerplant
6For developed economies we applied a 50% reduction to food prices as food prices tend to reflect more than just the raw materials (packaging, advertising, margins, etc.) and the cost for food away from home is to a larger extent driven by factors other than food, energy and labor prices (e.g. rent, advertising etc.). Gold mining in developed countries is very small to begin with and thus any overestimation of the link between energy prices and labor costs would have a marginal impact on the results at most.
7Letters according to WGC methodology
8According to the World Gold Council, above ground stocks are around 174,000 tonnes. We used the 100,000 tonnes figure in our example because it makes it easier to show mathematically why mine output must continuously accelerate for above ground stocks to increase at a constant growth rate.
9There is of course the second-order effect of technological improvements in the capital stock which can squeeze out efficiencies such that higher energy (or other input) costs can be 'absorbed' into better production processes. To paraphrase, "Rising energy prices are the mother of invention."