the long-term return on gold has exceeded inflation by a factor of two

A lot of work has been done in science, the results of which show that gold helps manage and mitigate risk in an investment portfolio. At the same time, it has not been clearly determined what role gold plays in the development of the investment portfolio’s return. Frameworks for estimating the long-term performance of gold exist, but they are inconsistent with capital markets assumptions that apply to other asset classes. The report below provides this framework while considering that gold is both a real commodity and a financial asset.

Research on the expected return of gold has generally concluded that gold’s main function is to preserve the purchasing power of the asset. Thus, it is concluded that the gold yield moves hand in hand with the general price level (inflation) in the long term.

And despite the fact that existing studies are comprehensive, two factors are often found in them, which in our opinion distort the image of gold and have caused widespread prejudices:

• Studies use periods when the gold standard was in effect to analyze gold returns. This paints a misleading picture of gold and the general price level.
• Viewing long-term price dynamics only through demand in financial markets while ignoring other sources of demand. This is probably one of the main reasons why there is relatively less gold in investment portfolio allocations.

In most cases, the conclusion is reached that the expected long-term real annual return on gold (nominal return minus inflation) is between 0 and 1 percent.

The return on gold has been significantly higher than inflation

However, we show that the long-term return on gold has been significantly higher than inflation over the past 50 years. It is much more in line with the gross domestic product (GDP) of the global economy.

Our simple but robust approach uses as a starting point the distribution of on-the-ground stocks across different demand categories. Different categories of gold buyer demand are jewelry, technology, central banks, financial investments, and retail bars and coins. For many segments, the influence of buyers on price is much broader and more important than current theories assume. Although financial market investors tend to influence short-term price movements the most, their influence is smaller in the long term.

We show that the price of gold is most affected in the long run economic componentreflected by the nominal GDP, and financial componentwhich is reflected by the total value of global stock and bond markets. We use third-party inputs to estimate the long-term expected return on gold.

A challenge

Since gold has a dual nature, i.e. it is both a financial asset and a real commodity at the same time, it is not possible to estimate its value using traditional asset price models. This is further complicated by the fact that central banks also hold gold in reserves as a monetary asset. This is done despite the fact that the gold standard was abandoned five decades ago and there is no obligation to do so.

Since gold does not provide cash flow, traditional discounted cash flow models cannot be used. In general, in the case of gold, the price models applied to raw materials are also not sufficient, because the amount of gold on earth is constantly increasing. This and several other factors mean that the primary production of gold (in mines) does not limit the supply of gold. Unlike most commodities, such as oil and wheat, gold cannot be consumed until it disappears.

Several theories propose that the long-term return on gold should be equal to the rate of inflation. Among them are Hotelling’s works. In his work on exhaustible resources, he proposes that commodity prices are linked to interest rates, referring to the opportunity cost of production costs. Since interest rates and inflation move in sync over the long term, changes in raw material prices and production costs move hand in hand with interest rates and inflation, according to Hotelling.

That the price of gold is driven long-term by inflation, interest rates or mining costs is too narrow an approach for several reasons.

First, gold has significantly outperformed both inflation and the risk-free rate. Gold’s long-term average annual compound return (in US dollars) has been 8 percent in the years 1971-2023, while the US inflation rate has been 4 percent, and the yield on US 3-month government bonds has been 4.4 percent. The probability that such a much higher return is due to chance is very small.

These return numbers also disprove claims that gold returns are proportional to the risk-free interest rate. According to empirical data, the yield of gold is much higher than this. Because of this, this assumption cannot be used in gold price model frameworks either.

Second, some studies suggest that producers set a marginal gold price because they link it to mining costs. At the same time, it has become clear that miners respond to price increases by opening more expensive deposits, which increases mining costs and vice versa. Thus, causality appears to be the opposite of what the research suggests.

Finally, ground reserves of gold are steadily increasing, and all that gold is poised to return to the market, competing with mine production, which accounts for less than 2 percent of the world’s total. This reduces the sensitivity of the gold price in terms of production volumes, but at the same time it also differentiates gold from other commodities.

Cube

Existing research says that the gold price is most shaped by financial investments. Although the influence of financial markets cannot be denied in the short term, the influence of other categories of demand is greater in the long term.

To date, gold reserves on the ground reach 212,582 tons. If we imagine this quantity as a cube, then we can divide this cube into parts according to the owners. The emerging picture is extraordinary for many reasons.

The cube illustrates how such a common metal is so scarce that the volume of the cube can barely fit three Olympic swimming pools. It also reveals how small the share of financial investments, i.e. physical gold-backed exchange-traded funds, and OTC (gold acquired through over-the-counter transactions) is. In some ways, this is a misleading statistic, given how large the daily flow of gold is through financial centers.

Since such a large part of the hypothetical cube has not been acquired through financial instruments, other factors that are not related to the daily decisions of the financial markets must also be taken into account. The flow of gold to investors through financial instruments is more than twice as volatile as net consumption (jewelry + technology – recycling), while accumulation is much slower.

It is this accumulation – be it individuals, central bank reserves or even long-term investments – that is for us economic component. Financial component represents more tactical factors, such as the demand for risk management, be it individuals or institutional investors.

These components are consistent with the key drivers of the gold market that we have identified in our previously established GRAM and Quarum pricing models.

Abandoning the gold standard

Calculating expected gold returns based on historical data generally falls into the same trap. In general, the longer history you look at, the better, because more data allows you to be more confident in your analysis. Data from 1900 or earlier is generally used to look at the long-term performance of bonds and stocks.

Transferring this logic to gold creates one glaring problem: for the lion’s share of the 20th century, gold prices were fixed by central banks and governments. This means that gold was money, pegged to the dollar at a fixed price. It was changed only in special cases. Because of this, investors could not always use gold to hedge against inflation or equity markets. In addition, gold was prohibited as an investment in the USA between 1933 and 1974.

Although gold can be seen as an asset during the gold standard periods, the most important is what happened after 1971.

If we analyze, for example, a listed company and try to understand what the expected return of the stock could be, we start from the current state of the business, not from what the business has been like in the past. The past and the future are different, and the past cannot be transferred to the future. Let’s look at the Finnish company Nokia, which produced rubber cables and shoes until the early 90s. It later became one of the world’s leading companies in the telecom industry. The fact that Nokia is a footwear manufacturer could not be taken into account in the price level metrics at the beginning of the 90s. Similarly, in the case of gold, 2024 cannot be compared to the first half of the 20th century, when the gold standard was in effect.

Analysis results

For economic and financial components, we use several economic and financial variables. For the economic component, we use nominal GDP in dollars. Nominal GDP has real GDP, an inflation component (the GDP deflator), and a currency component – ​​the latter we use to convert GDP into US dollars. It reflects the flow of capital from earnings to gold.

For the financial component, we use the total value of global stock and bond markets, or the global portfolio in US dollars. It reflects how much investment is available to investors. The important thing here is that we look at the total market value, which takes into account both the quantity available and the quantity added, not just prices.

We use regression analysis to assess the effect of these variables. The analysis reveals that GDP is the main driver of the gold price in the long term.

The table below shows the results of the regression analysis of the two models. Model 1 is a simple regression that examines how the price of gold moves with GDP. This model gives us a positive statistically significant relationship – 79 (R 2 ) percent of gold price movements are described by changes in GDP.

However, Model 2 uses both economic and financial components. According to this model, 92 percent (R 2 ) of gold price movements are described by financial and economic components together. At GDP of 2.8, it means that, all else being equal, a 1 unit increase in GDP leads to a 2.8 unit increase in the price of gold. Since both are logarithmic, it can also be described as a percentage change. The negative coefficient of the global portfolio (-1.07) moderates this relationship because gold competes with other assets for savings.

It is important to mention that the negative coefficient of the global portfolio does not reduce the price of gold, but makes it grow more slowly. For Model 2, the Johansen cointegration and Phillips-Perron tests clearly show that there is a long-run relationship between the two components and gold prices.

Graph 2 shows the result of these regressions. The purple dot shows the gold price modeled on GDP growth only. In this case, clear deviations can be seen both in the 80s and in the 2000s. With both components (economic and financial), the modeled gold price increase is indicated by the black dotted line. The two variables are much better for modeling gold price movements.

In our opinion, any model that does not consider economic growth in addition to financial factors is not suitable for estimating the long-term expected return of the gold price. Our work shows how important economic growth and the role of gold in global portfolios are in the long-term development of the price of gold, both theoretically and empirically.