Gold: A Modern Investment Framework For An Ancient Asset, Part II

Part II: A Fair-Value Approach to Modelling the Price of Gold

Gold has the unique and valuable characteristic of being a de facto form of ‘insurance’ for a portfolio of financial assets, and the national fiat currencies in which they are denominated. This is due to it not being a financial asset per se, as explained in Part I of this series. In Part II of this series, I present a way to model the price of gold based on fundamental macroeconomic variables. The variables with the strongest historical significance for determining the price of gold suggest that it is currently close to fair value. However, a case can be made that the explanatory variables themselves have a skewed future distribution at present, suggesting potential future upside for gold.

Gold’s Macroeconomic Drivers: Energy Prices and Real Interest Rates

As gold pays no yield or dividend there is normally some ‘opportunity cost’ of holding gold rather than cash or financial assets. Indeed, when looking at the historical data, one observes that the price of gold is inversely correlated to long-term real interest rates, as derived from the yields on long-dated inflation-linked government debt. To hold gold also means to forgo the opportunity to invest in productive, revenue-generating activities, and so we also find that the price of gold is positively correlated with that of the world’s single most important input to all production and revenue generation: energy.

The relationship is strongest with long-dated oil prices, which tend not to reflect the ‘noise’ generated by short-term supply and demand factors specific to the energy markets. Rather, long-term oil prices represent the longer-term cost of growing the economy’s productive capital stock, the benefits of which should accrue, over time, to the holders of financial and other economic assets. Hence here, too, we can understand that it is the opportunity cost of holding gold, rather than investing in productive economic assets, which helps to explain movements in price.

With reference to these two variables alone, one can model much of the medium-term movement in gold prices. My friend and former colleague, Stefan Wieler, developed a regression model along these lines some six years ago. He subsequently published a paper detailing the results.[1]

Among the key findings, the paper demonstrates what I mentioned in Part I, that supply-side factors are not statistically significant for gold, due presumably to gold’s highly stable, price-inelastic supply, which increases at a steady rate of about 1-1.5% per year regardless of the macroeconomic environment. Hence the focus must be on the demand side, where the paper shows that long term real interest rates and energy prices explain some 60% of the monthly variance in gold prices and 70% of year-over-year changes.

The overall fit, especially at shorter time horizons, is best with specific reference to gold demand data, including COMEX positions and gold ETF holdings. However, as these gold-specific factors correlate highly to changes in long-term real interest rates, they should not be considered statistically significant in of themselves; rather, they are timely, observable expressions of the impact of long-term real rates on actual gold demand. The same is not true of central bank gold purchases, however, as these are not highly sensitive to changes in real interest rates; rather, they are driven by strategic reserve policy decisions, and including these as a separate exogenous variable improves the overall explanatory power of the model.

[1] The paper introducing the model can be found at the Goldmoney website here

Fairly Valued

Putting the above model to work today, what can it tell us? The most recent published update of the model was in 2017, and it included a table which calculated the implied price of gold in a matrix of long-dated real interest rates and oil prices[2]. While real interest rates have been declining of late, supporting the gold price, long-dated oil prices have also been in a downtrend, exerting downward pressure on gold. At present, these factors imply a fair-value gold price of just under $1,500, below the current level of about $1,560.

[2] This model update can be found at the Goldmoney website here

Chart 1: Long-term US Real Interest Rates and Oil Prices

Source: Macro Hive

However, it is important to note that the rate of central bank gold buying has been elevated over the past year, helping to explain the difference. The model does suggest, therefore, that were central banks to step back from the market, the price of gold would be likely to decline slightly, holding other factors equal. As it stands, however, the gold price is more or less where the model would place it.

It should be noted that the model, as specified above, is somewhat regime-dependent. For example, in the model, binary ‘plug’ variables for specific events are highly statistically significant. Such events include the global financial crisis of Oct-Nov 2008; Fed announcements of QE1 and QE2, and operation TWIST; and the subsequent ‘taper’ announcement when these measures were scaled back. While such events no doubt have an impact on investors’ expectations and risk preferences generally, they cannot be adequately captured with reference to observed real interest rates, oil prices or other model variables alone

This apparent regime-dependency is of importance when assessing the future demand for gold as ‘insurance’ for a portfolio of financial assets. As we explored in Part I, due to gold’s stable supply, shifts in the demand function for gold can have a large impact on price beyond what is implied by real interest rates, energy prices or central bank purchases. The behaviour of gold in response to major policy announcements helps to demonstrate this. It also highlights that there are limitations to the fair-value approach to modelling gold, assuming as it does that the demand function for gold is continuous, when it clearly is not. That there is now talk of the Fed and possibly other central banks implementing new, unconventional policies to support demand and/or their respective financial systems, the current model results should be interpreted with care.

In Part III, I will drill down into the outlook for gold’s primary drivers, long-term real interest rates and oil prices, and explore some additional factors that could shift the demand function for gold in future. As I will explain, I believe that gold is likely to strongly outperform financial assets in the months ahead.

John Butler has 25 years experience in international finance. He has served as a Managing Director for bulge-bracket investment banks on both sides of the Atlantic in research, strategy, asset allocation and product development roles, including at Deutsche Bank and Lehman Brothers.

(The commentary contained in the above article does not constitute an offer or a solicitation, or a recommendation to implement or liquidate an investment or to carry out any other transaction. It should not be used as a basis for any investment decision or other decision. Any investment decision should be based on appropriate professional advice specific to your needs.)