Stores of Value: Part 3 - Is there a role for stores of value assets in diversified portfolios?

Section 2: Forward-Looking Analysis

In order to make a reasonable forecast of the expected return and volatility of a portfolio of assets we require 3 sets of data for all assets: 

• Forecasts of volatilities, 
• Forecasts of pair-wise correlations between assets, 
• Forecasts of returns. 

Volatility and correlation forecasts are readily available as they are typically calculated from historical data, and there are well-established methods for forecasting the expected returns of most assets. There are however (and somewhat surprisingly in the case of gold!) no widely agreed methods for forecasting the expected returns of gold and Bitcoin. 

Price discovery tends to take place somewhere between that of a currency and a collectible (e.g., Fine art, vintage watches, BMX bikes, trading cards), as explored in detail in a fascinating recent blog post from Aswath Damodaran (Professor of Finance at Stern School of Business of New York University). Both gold and Bitcoin trade in high turnover, liquid markets with  their price set against fiat currencies (typically the US Dollar) but there is no objective basis upon which one could argue that the price is either cheap or expensive – the price is the price.  The price is simply where willing buyers and sellers agree to trade.   

Our colleagues from the Macro Strategy and Economics team have provided us with their latest capital market forecasts, extending their volatility and correlation analysis to include gold and Bitcoin: 

Which leaves the challenge of generating return forecasts for the last two rows.

Challenge accepted! 

Let’s start with gold, borrowing again from the blog post of Professor Damodaran, the main reason to hold gold in a portfolio is as a hedge against crises and high rates of inflation, particularly if the bulk of a portfolio is invested in financial assets. (Real assets like land, property, and infrastructure should do relatively well in the presence of high inflation and therefore also make useful hedges.)

Starting with the assumption that the subset of investors who hold gold for these purposes hold a constant proportion of their portfolio in gold, this implies that the value of gold should rise and fall along with the overall nominal value of financial assets. However, given that this relationship is somewhat behavioural (i.e., the preference of investors to hold some proportion of their portfolio in gold for hedging purposes) we can expect that the proportion will vary with the perceived desirability of gold as a hedge or investment, which will be driven by factors such as gold’s recent price performance, fears of recession, inflation, and other crises. 

Complementing this viewpoint, the World Gold Council published research in 2024 on Gold’s long-term expected return (GLTER) which explains changes in the value of gold as being positively related to nominal economic growth (i.e., inflation     plus real GDP growth) and negatively related to market returns. Its conclusions are broadly consistent with the intuition that gold’s long-run performance is linked to inflation and macro regimes, though estimates vary across models and samples.

Gold as a safe haven

As a proxy for economic and market crises we have identified 9 occasions where the closing price of the S&P-500 Index declined by at least 15% from high to low since May 1982 (the first date for which we have data for US 10-Year Note futures). Included in this selection are the Global Financial Crisis, September 11, Covid-19, the Iraq invasion of Kuwait, the 2018 hawkish Fed and US-China trade war escalation, the 1987 crash, the LTCM hedge fund blow up, and “Liberation Day”. 

For each of these ‘crisis’ periods we calculate between the dates of the S&P-500’s high and low: 

• The change in the price of the S&P-500 Index 
• The return of US 10-Year Note futures 
• The return of gold. 

Our choice of US 10-Year Note futures is because they are usually considered the go-to safe haven asset. They are highly liquid, perceived as likely to rise in a crisis, and are permitted investments in most diversified professional investment portfolios. However, we note that in inflationary or fiscal-stress regimes, bonds may be a less reliable hedge, which is one reason investors also consider complementary diversifiers. 

Gold outperformed 10-year note futures in 5 of the 9 crises captured in the table above, and one further occasion was a tie. Both hedging assets recorded positive returns in 6 of the 9 crises and the average return of 10-year note futures was marginally higher across the sample.

In comparison, the average return of bonds and gold was superior to either bonds or gold alone: 

• The average return of both assets was positive in 7 of 9 crises, compared to 6 for either of bonds or gold = more reliable. 
• The R² between the returns of the S&P and the choice of hedging asset in the list of crises above was higher for the average of bonds and gold than for either asset individually = more predictable. 

The conclusion that follows is not that one hedging asset is better than the other, but that a portfolio holding both gold and US 10-year notes would have been (and, we think, is reasonably likely to be) more effective in a crisis compared to a portfolio holding only one, or neither. 

Towards a return forecast for gold

Bringing the above concepts together we can argue that: 

1. Gold has a useful role to play in a portfolio as a hedge against adverse events, 
2. There is a dollar value of gold that will be held at any given time for this purpose, 
3. This amount will be some function of the total value of all financial assets, the percentage of investors who choose to hold gold for its hedging qualities, and their desired portfolio allocation to gold for hedging purposes. 

The first term in point 3 is an observable variable: the total value of investible assets. The second and third terms will be subjective, driven in part by collective concerns about events for which gold is considered to be a good hedge (e.g., inflation, currency debasement), and recent changes in the gold price. 

This idea seems to have some validity based on existing data. For example if we compare the value of above-ground gold with the value of total financial assets (St Louis Fed data) the regression has an R² of 0.90 indicating a strong relationship between the two over the long run. 

We observe a similar relationship between aggregate global money supply and the total value of gold. These relationships lend support to the idea of a central portfolio allocation to gold based on global financial asset values around which the total gold allocation moves depending upon the subjective factors described above.

From 1971 to the end of 2024 financial assets have increased at around 8% p.a. on average while OECD M3 Money supply and the value of above-ground gold have both increased at 10% p.a. on average. 

Given that the value of gold appears to be a combination of the increase in financial asset values and a rational behavioural tendency, we think that it is reasonable to argue that the future will not differ meaningfully from the past and therefore the past should provide a good guide for the future in this regard. In a typical year, gold mining results in an increase in above-ground supply by less than 2%, contributing to the total value of gold in a way that doesn't benefit holders of gold, so we subtract this from the return forecast, and so…  

We will therefore use the conservative forecast of 8% p.a. long term price returns for gold. 

As a sense check, based on the other capital market forecasts above, 8% nominal annual return gives an excess return of 4.25% p.a. above cash. Dividing this by a forecast volatility of 13.5% gives a Sharpe ratio of ~0.3 which is a reasonable assessment of the rate at which gold converts risk into excess returns when compared to risk premia of other asset classes. 

Forecasting Bitcoin’s returns 

Now to Bitcoin’s return forecast, which requires more assumptions. Here we will rely on the assertion made in Part 2 that Bitcoin is an emerging store of value asset, with the implication of ‘emerging’ being that its share of overall store of value assets will therefore grow. For the reasons we covered in Part 2, we think that this is a reasonable assumption to make which then feeds into our forward looking statements below.  

At the end of 2025, the combined value of gold and Bitcoin is ~$US33 trillion. Based on the 8% price return assumption for gold (as a proxy for store of value assets in total) plus around 1.75% annual mine supply this would increase to ~$US84 trillion by 2035. About $US13 trillion of this increase would arise from mining of gold and Bitcoin with the remainder of $US38 coming from price appreciation.

Assuming Bitcoin maintains its end-2025 5.3% share of store of value assets through to 2035 would imply annualised returns of 8.7% for Bitcoin and 8.0% for gold based on December 2025 prices. Under a constant-share scenario, Bitcoin’s implied price return is slightly higher because its future supply growth is lower than gold’s - meaning more of the required market-cap growth must occur through price rather than quantity.

By the time we get to 2035 Bitcoin will likely have either failed as a store of value asset or will have become more successful in capturing a greater share of the total store of value market. The table below reveals an interesting possibility, which is that if Bitcoin is able to become an equal partner with gold in the market for store of value assets, the price of gold could be little changed a decade from now. The implication is even if Bitcoin is successful as a store of value asset over the longer term, it may be prudent to hold both assets. 

Source: AMP

Taking a reasonably conservative approach in assuming that by 2035 Bitcoin doubles its share of the Store of Value market to 10.6% implies a return of 17.1% p.a. based on the method above, and so… 

We will therefore use the forecast of 17% p.a. long term returns for Bitcoin¹. 

Again, applying the Sharpe ratio sense check, a 17% return with 50% volatility implies a Sharpe ratio below 0.3, which is consistent with other risk premia and similar to gold above.

Note that, for consistency this requires that we lower our forecast return for gold to 7.5% to reflect gold’s loss of store of value market share, which we have reflected in the analysis below. 

Portfolio Construction 

Now we are in possession of all the required inputs, we can generate an efficient frontier to help us understand the payoff between risk and return from adding gold and Bitcoin into our portfolio. 

Our optimiser started with a base portfolio of the model SRP defined earlier and was allowed to substitute up to 5% each of gold and/or Bitcoin.  For example, a portfolio with 4% gold and 2% Bitcoin would also contain 94% of the SRP. 

Two distinct phases are visible in the optimal allocations, with ‘optimal’ defined as the portfolio with the least volatility for a given level of expected return (equivalent to maximising the Sharpe ratio at a given level of expected return). 

Phase 1 is the gently upward sloping straight line moving left from the SRP. As we progress further left along this part of the efficient frontier, the portfolio allocation to gold is increasing, expected returns are increasing and expected volatility is decreasing. At the far-left extent of this section of the efficient frontier the portfolio is allocated 5% to gold and 95% to the SRP. 

Phase 2 is the curved section where the Bitcoin allocation is gradually increased from 0% to 5%, with gold remaining constant at 5%. At the very top of the efficient frontier the portfolio is allocated 5% to gold, 5% to Bitcoin, and 90% to the SRP. 

Interestingly, adding Bitcoin does not immediately increase the forecast volatility of the portfolio: the minimum volatility portfolio subject to our constraints contains 5% gold and 0.75% Bitcoin. Furthermore, even with 0% gold in the portfolio, an allocation of up to 1.5% Bitcoin would result in lower forecast portfolio volatility than a 0% gold and 0% Bitcoin allocation.

But wait, isn’t this a bit biased? 

A reasonable question at this point might be to ask whether we have rigged the numbers to make gold and Bitcoin look good in this framework. 

To hopefully allay concerns of bias, we explored what returns one needs to believe are reasonable to justify any holding of gold and/or Bitcoin. It turns out that our optimiser will hold non-zero allocations to gold and Bitcoin, even when their forecast returns are lower than that of the SRP (which is 6.32%). Specifically, the optimal allocations to gold and Bitcoin rise above zero when forecast returns are equal to or greater than 4.2% and 5.8% p.a. respectively. Recall that these are less than the steady state returns of 8.0% and 8.7% calculated above using the assumption that the total market for Store of Value assets continues to increase at its historical rate over the coming decade. 

The optimiser will reach the 5% maximum allocation when the forecast return for gold is 4.5% or more, and for Bitcoin when the forecast return is 9.4% or more. We believe these return expectations to be well within the realms of possibility. 

Additionally, we would not be putting in the time and effort to write this series of notes if we did not strongly believe that store of value assets deserve a place in the portfolios of many investors.   

You may also ask: isn’t this result just because your equity returns are too conservative?  

We looked at that too. It turns out that with 12% as the return forecast for Australian and international equities, the optimal portfolio will hold the 5% maximum allocation to both gold and Bitcoin. Even if equities are forecast to return 15%, the optimal portfolio holds 5% gold and over 3% Bitcoin.   

It turns out that gold’s diversifying characteristics are such that even if we forecast that equities would return an astounding 30% p.a. the optimal portfolio would still allocate a portion to gold. 

In summary 

The forward-looking analysis illustrates how incorporating gold and Bitcoin into a simulated diversified portfolio could deliver meaningful benefits, even under conservative return assumptions. We've demonstrated that both assets can be reasonably expected to enhance absolute and risk-adjusted returns while potentially also reducing portfolio volatility. This supports the case for including store of value assets within a diversified portfolio, acknowledging their distinct roles in hedging risks and capturing potential upside.  

Looking Ahead: The Role of Alternative Diversifiers  

In Part 4, the concluding part of this series, we will explore the potential role that alternative diversifiers can play alongside traditional diversifying assets within an investment portfolio. This discussion is particularly relevant for investors who are wary of the possibility that governments might pursue currency debasement to reduce their debt burdens. By examining these alternatives, we aim to provide insights into how portfolios can be further strengthened against such macroeconomic risks.