The correct allocation is not Zero
Summary
Most US individual investors hold less than 5% of their portfolio in gold. Institutional allocations are at zero. We show that optimal allocation to gold is somewhere between 15-25% of a total portfolio, and review the characteristics of gold as an asset class that make it unique and different.
Gold as a new asset class
It may come as a shock to gold bugs but gold is a new asset class, with less th1an 60 years of history. Prior to 1971, gold was money and money was gold. The US dollar was convertible to gold—if you were a government or a central bank.
In 1963, France converted its US dollar holdings into 3,300 tons of gold and moved the gold to Paris in a secret operation codenamed “Vide-Gousset”. The US, despite some misgivings, honored its commitment to gold convertibility at that time but the strain of continued trade and fiscal deficits and increasing outflows of gold made it inevitable that the link would break. In 1968, Congress removed the gold backing of US dollar for domestic convertibility and in 1971 President Nixon ended international convertibility of gold. The US dollar and gold became unmoored from each other. Gold ceased to be money and became a monetary asset instead.
A good way to show the sharp break in the behavior of gold is to look at its volatility. Prior to the unlinking of US dollar from gold, the price of gold had minimal volatility of only 0.1% of year. The only volatility spike came in 1933 when President Roosevelt confiscated privately held gold and devalued US dollar from $28 to $35 per ounce.
After the end of domestic convertibility, gold price volatility jumped to 10% per year and rose even higher after the end of international convertibility. Since 1971, gold volatility averaged about 15% per year, similar to the volatility of the stock market.
Gold market history
Given the break in the behavior of gold price since the end of the Bretton Woods system, it makes sense to examine the history of the gold market since then.
Gold has gone through three bull markets and two bear markets.
Gold bull markets
| Driver | Start | End | Length | Price increase |
| 1970s inflation | Aug. 1971 | Sep. 1980 | 9 years & 1 month | 14.6x |
| 2000s stock market funk | Mar. 2001 | Sep. 2012 | 11 years & 6 months | 5.9x |
| Post-Covid inflation | Apr. 2019 | ? | 6 years & 9 months (ongoing) | 3.3x |
The first gold bull market took place during the 1970s, a period of rising commodity prices, oil shortages, and explosive inflation. Stock prices collapsed and interest rates rocketed up, destroying bond returns. The only asset that provided shelter was gold, and retail investors flocked to it. Physical gold became all the rage. This was the age of the Krugerrand and gold bars in safe deposit boxes.
The second gold bull market coincided with ten flat years of the US stock market. After the dotcom bubble deflated starting in 2001, US stocks went nowhere for a decade. At the same time, a new innovation arrived: gold ETFs which allowed investors to gain gold exposure without having the hassle of storing and protecting physical gold. Gold held by ETFs soared from zero in 2003 to 2,500 tons by late 2012.
The third gold bull market is ongoing. Unlike the previous gold bull markets, this one doesn’t correspond to weak US stocks markets; gold and stocks are going up in tandem which is historically unusual. Triggered by post-COVID inflation surge, it is now driven by central banks’ buying of gold to replenish their reserves.
Gold as a reserve asset
Central banks used to keep the majority of their reserves in gold. After the collapse of Bretton Woods, they stopped adding to their gold holdings and grew holdings of foreign currencies instead, keeping existing gold reserves flat. The US dollar became the preferred reserve asset.
When the Cold War ended in 1990, it led to a re-evaluation of economic policies around the world. Whatever concerns there may have been about the dependence on the US dollar, they were swept away by the triumph of the US economic and political system that seemed destined to dominate the world. Central banks around the world went on aggressive programs of reducing their gold reserves. Overall, the amount of gold held by central banks dropped by 20% during the 15 years from 1995 to 2010. The most aggressive sellers were European central banks, led by Switzerland, Belgium, the Netherlands, Austria, the UK, and Spain. Switzerland by itself sold almost 1,600 tons of gold.
The global financial crisis of 2008 put an abrupt end to the selling. Since then, central banks bought back all the gold they sold in the 1990s and 2000s. Gold’s share of reserves stopped declining and stabilized at around 10%. Since 2020, central banks have been buying more gold than foreign currencies: the share of gold in total reserves went from 10% to 25% and there is no sign of slowing down. Much has been made of the Russian response to the freezing of its foreign exchange assets but the shift began before the invasion of Ukraine. At this point, it is hard to see why central banks would reverse course and start piling into US dollars again. This gold bull market still has legs.
Gold ETFs
One of the barriers to greater gold allocations in investor portfolios has been the hassle of dealing with physical gold. It’s easy to buy a few ounces of gold or a few Krugerrands and hide them in a safe next to the guns, but this approach doesn’t scale well. Any meaningful gold holding needs to be secured, insured, and periodically audited. This process is costly and cumbersome.
Gold ETFs solved this problem twenty years ago. The first one, GLD, launched in 2004, and has been followed by a number of imitators. Gold ETFs take care of storage and insurance. It’s also cheaper to buy and sell gold via ETFs than in physical form. For anyone other than hardcore survivalists, gold ETFs are a superior format for investment.
The cost of gold ETFs has declined. The oldest ETF, GLD, has an annual performance drag of 0.4%, due to its high expense ratio. Its biggest competitor, IAU, was launched in 2005 and has a smaller performance drag of 0.29% per year. Finally, GLDM launched in 2018 with a performance drag of only 0.15% per year.
Ironically, the higher the cost drag, the more assets the ETF has. Investors aren’t paying enough attention to fund costs: they should transfer their gold exposure from GLD and IAU to GLDM. There are also a few small ETFs that have even cheaper fees: as an example, IAUM has about $5 bn in assets and charges 0.09%.
Gold ETFs
| ETF | Expense ratio | Performance drag | Assets | Launch date |
| GLD | 0.40% | 0.41% | $132 bn | Nov. 2004 |
| IAU | 0.25% | 0.29% | $62 bn | Jan. 2005 |
| GLDM | 0.10% | 0.15% | $23 bn | Jun. 2018 |
Optimal allocation to gold
We started this post by noting that most investors have zero gold exposure in their portfolios. Very few have more than 5% invested in gold. But what is the optimal number?
Let’s examine the historical record to see what would have worked. We’ll look at two different starting points: the first is 1971 when gold ceased to be money and became an investment asset, and the second is 2000 when stocks were trading at similar valuations to today (40x cyclically-adjusted PE for stocks).
Assume we have a standard portfolio of 60% stocks and 40% bonds, and we start adding gold to it.
The standard portfolio delivered a return of about 7.3% since 1971 and less than 6% since 2000, with volatility of about 9.5% for both starting points. Adding gold would improve returns and reduce volatility. Taking gold allocation to 15% would reduce volatility to below 9% while improving returns. The minimum volatility combination has had 15% gold since 1971, and 25% gold allocation since 2000.
The optimal allocation to gold maximizes the Sharpe ratio of the combined portfolio. The optimal allocation has been 25% since 1971, and 40% since 2000. With this allocation, investors would see their returns increase by 1% per year since 1971 and more than 2% per year since 2000. This may not seem like much, but 2% annual difference in returns compounds to 50% higher wealth after 20 years. And this improvement in returns comes without an increase in volatility: the volatility of the combined portfolio is lower than the standard 60/40 portfolio.
Worst-case scenario
What would have happened if we picked the worst possible starting time to allocate to gold, like at the middle of a gold bear market and right at the beginning of a massive stock and bond bull market?
Using 1990 as the starting point, the combined portfolio with 25% gold allocation lagged during the 1990s but then made back all the underperformance, while suffering from smaller drawdowns and less volatility. The worst case scenario wasn’t so bad.
The rational conclusion
If investors were rational, nobody would have less than 15% of their portfolio allocated to gold. But investors are humans and susceptible to fads, manias, and crowd psychology. It may feel more comfortable to follow the crowd and put all chips on stocks after the longest bull market in history.
A rational investor would behave like Mr. Spock: look at the data, consider the evidence, and take the optimal course of action. What would Mr. Spock do?
