Commodity matters

More than a year into the pandemic, investors are probably used to wild swings in the economy and financial markets. But, unlike past cycles, things moved faster and to a greater extent over the past 12 months. The key reason is that, to contain the virus spread, activity was shut down by government decree. At the same time, massive doses of monetary and fiscal stimulus were injected globally. Now that vaccine rollouts are ramping up and reopening is happening while the policy backstop remains in place, we’re seeing equally large bouncebacks in economic data and cyclical assets.

The risk-aversion phase of the pandemic resulted in a sharp increase in gold prices, which was boosted by a lower opportunity cost of holding the precious metal, as nominal and real bond yields fell. The price of copper picked up even more, reflecting rising demand for technological gadgets as consumers were stuck at home. While these dynamics have partly reversed for gold as yields have risen from record-low levels, copper prices have continued to surge – pushing the copper/gold price ratio even higher (figure 1). What’s likely to happen going forward, given the macro outlook?

Fair-value estimates: gold in line, copper overshooting 

We have refreshed our models for the price of gold and built comparable ones for the price of copper (figures 2 and 3). In essence, we estimate fair value by running statistical tests on a range of potential drivers to establish how well they perform – mostly as standalone relationships but also using large-scale econometric frameworks where available and appropriate (such as the one provided by the World Gold Council, to capture a range of demand and supply dynamics including those related to industrial uses). We then average out the price of these commodities implied by our models, with the weights representing how well they fit the data, along with their minimum and maximum to provide a sense of different possible outcomes depending on specific assumptions.

Our estimates suggest that the price of gold – at about 1,750 USD/oz – is fairly valued relative to fundamentals such as real yields, the US dollar and the share of negative-yielding debt to capture the opportunity cost of holding competing assets, plus gold ETF holdings to capture factors such as risk, uncertainty and short-term flows (figure 4). We cross-check these results against a wider battery of models, including inputs such as physical demand/supply. These other drivers do improve the robustness of the statistical relationship when combined with real yields, but not so much on their own and are difficult to project into the future. Our estimates also suggest that copper prices are overshooting fundamentals such as copper supply and various proxies for industrial production, along with exchange rate fluctuations (figure 5).

The drivers of gold: what’s changed? A real yield story

Let’s look at the main drivers of gold prices and how they’ve changed recently. The most important shift is that real yields, which are still in negative territory, have risen from the pandemic lows. This is by far the most important explanatory variable for gold and, in different specifications, features in all our models. The increase in real yields from the lows reflects expectations of stronger growth as economies reopen and, for a period of time, activity runs at a very strong pace before normalising once supply bottlenecks ease and pent-up savings are spent. Rising real yields are likely to remain a headwind for the price of gold going forward, in our view (figure 6). However, while overshoots are possible, at this juncture their level doesn’t suggest any further downside to gold from a valuation perspective.

Other drivers paint a similar picture. For example, the share of negative-yielding debt has now declined (given rising bond yields) (figure 7). That said, these other metrics too don’t point to extra downside. Rather, they basically indicate that gold prices seem more or less fairly valued according to these fundamentals, nothing more but nothing less either. Other important drivers are just a moderate headwind, such as the fact that the US dollar is no longer weakening (but not strengthening much either). Shifts in physical demand/supply and fast growth in money supply are supportive, but they’ve either improved marginally (physical factors) or don’t fit the data that well (money supply), while ETF holdings – which are key from a short-term perspective – have continued to decline recently (figure 8).

The drivers of copper: what’s changed? A tale of demand overshoot

The main drivers of copper are more ‘physical’ in nature. Our models are mainly based on proxies for global demand – ranging from industrial production calibrated using our GDP numbers (figure 9) to combinations of the purchasing managers’ indices for manufacturing in China (by far the largest source of demand) and the US (figure 10) – and refined copper supply. The US dollar is included in some of our models. But financial variables appear to matter less for copper than for gold (and exchange rates haven’t moved that much anyway recently). Similarly, we do look at physical demand/supply for gold (see Counterpoint Weekly 14–20 December 2020, ‘Good as gold’), but it appears to matter less than for copper, and in any case it has changed marginally lately.

One key takeaway from our econometric exercises is that copper prices appear to have risen over and beyond the model predictions, especially considering that China manufacturing is expanding less briskly. One explanation could be that a lot of the anticipated bullish news has already been priced in (consensus forecasts for global growth have risen). So copper won’t likely continue to trend higher, even if it may be slow to fall. As reopening progresses and vaccine rollouts boost mobility, the likely normalisation in services will be one of the key drivers of activity. This should imply a rotation from buying metals-intensive, working-from-home electronics, kitchen appliances and gym equipment to spending money on oil-heavy travel, favouring energy commodities at the expense of metals in the short term (figure 11).