Ask two allocators why they hold gold, and the answers often have little to do with each other. One talks about correlation matrices and drawdown behavior. The other talks about reserve currencies over multi-decade stretches. Both describe real phenomena, and the gap between them is why so many hard-asset positions end up without a clearly defined mandate.
The stated reason for holding metal determines which vehicle fits, how much carrying cost is defensible, and what evidence would justify resizing or exiting the position.
A metals allocation is unusual because it produces no earnings and collects no rent. Every contribution it makes has to come from price behavior and from how that behavior interacts with the rest of the portfolio.
That makes the position simpler to model than most alternatives and harder to justify, because there is no cash flow to fall back on when the price moves sideways for a decade.
What Problem Is a Precious Metals Allocation Meant to Solve?
Three distinct portfolio jobs often get assigned to the same line item. One is ordinary diversification: an asset whose returns are not driven by corporate earnings or the discount rate. Another is tail-risk protection, where the position is expected to hold up during sharp equity drawdowns. The last is a hedge on the purchasing power of the currency in which the portfolio reports, operating over decades rather than quarters.
Evidence for tail protection is more qualified than most summaries suggest. Baur and McDermott's Is Gold a Safe Haven?, published in the Journal of Banking & Finance in 2010, tested 1979 through 2009 and found that gold acted as both a hedge and a safe haven for the U.S. and major European equity markets, but not for Australia, Canada, Japan, or the BRIC markets.
Same metal, same window, different results depending on where the investor sits. The paper also found gold was a strong safe haven for most developed markets specifically at the peak of the 2008 crisis, a narrower claim than "gold protects portfolios."
The third job is the one official institutions discuss most openly. The World Gold Council's central bank survey of 76 reserve managers in 2026 found 89% expecting global central bank gold reserves to rise over the following twelve months, and 74% expecting global U.S. dollar holdings to fall over five years.
For a decision-maker, that is a reserve-management judgment about currency composition, not a price forecast. It sits alongside the same tactical investment strategy questions allocators face when deciding whether to adjust portfolio weights.
What Does the Research Say About Gold as an Inflation Hedge?
Less than the marketing implies. Erb and Harvey's Golden Dilemma examined gold against unexpected inflation from 1975 to 2011 and found effectively no correlation, noting that any apparent positive relationship was driven by a single year, 1980.
Their broader finding concerns mean reversion in the real price: the ratio of gold's price to the U.S. consumer price index averaged 3.2 since 1975, hit 8.73 in January 1980, and fell to 1.46 by March 2001.
For an allocator, those two dates matter. Anyone buying at the 1980 real-price peak waited twenty-one years to reach the trough of the real decline, while the consumer price index kept climbing to 334.1 by August 2026 against a 1982–1984 base of 100.
Erb and Harvey's conclusion is that gold works as a long-run inflation hedge mainly if the horizon is measured closer to centuries than to normal investment mandates.
None of that makes the position useless. It does make "inflation hedge" the weakest available justification, leaving correlation and tail behavior to carry more of the analytical weight.
What Does It Cost to Hold Physical Bullion?
The carrying cost of precious metal investments is spread across five or six separate line items rather than concentrated into one expense ratio, which is why it is routinely underestimated at the point of purchase.
Start with the acquisition premium. The US Mint does not sell bullion coins directly to the public, distributing them through authorized purchasers instead, and prices them based on the prevailing market price of the metal plus a premium to cover minting, distribution, and marketing costs. Every intermediary between the Mint and the buyer adds to that figure.
Then there is the round trip. The Mint's own consumer awareness guidance tells buyers to understand current market premiums before transacting and to scrutinize any coin offered below the prevailing market price.
Premiums move with product form and retail demand, and the spread between what a dealer pays and what a dealer charges never appears as a separate line item on a statement.
Storage and insurance follow, and for tax-advantaged accounts the rules are specific. Section 408(m) of the tax code treats an IRA's acquisition of collectibles as a distribution, with a carve-out for certain coins and for bullion meeting exchange fineness standards, provided that bullion remains in the physical possession of the account trustee. Home storage does not satisfy that condition.
On disposal, the IRS taxes net capital gains from collectibles, including coins, at a maximum rate of 28%, compared with 0%, 15%, or 20% for most other long-term capital gains.
How Do Bullion ETFs Compare to Holding Metal Directly?
An exchange-traded trust consolidates most of those frictions into a single annual fee and a bid-ask spread on a listed security. SPDR Gold Shares carried a 0.40% gross expense ratio and roughly $145.5 billion in assets under management as of September 23, 2026, while the iShares Gold Trust lists a 0.25% sponsor fee.
Those charges package storage, insurance, audit, and administration into the vehicle. They are paid out of the trust's own metal, so the ounces backing each share decline gradually even when the investor does not trade.
The trust structure improves liquidity and price transparency but removes direct possession, which for some holders is the entire point of the exercise.
Metal behind these vehicles sits in the wholesale market that LBMA vault data tracks: 9,632 tonnes of gold in London vaults in August 2026, equivalent to about 770,571 bars and valued at roughly $1.4 trillion.
For an investment committee, willingness to accept custodial claims on that pool produces a very different cost and governance profile from wanting bars in a named account.
The choice resembles the questions raised by structured products, where the payoff and the wrapper have to be assessed separately. Two instruments can reference the same underlying asset while exposing the holder to very different counterparty, tax, liquidity, and operational outcomes.
Why Do Mining Equities Behave Differently From the Metal?
Because miners sell metal but own a cost structure that moves on its own schedule. Newmont's fourth-quarter results filed with the SEC put full-year 2025 gold by-product all-in sustaining costs at $1,358 per ounce against an average realized gold price of $3,498 per ounce.
That is a margin of roughly $2,140 an ounce, and a 10% move in the metal price translates into something closer to a 16% move in that margin. Operating leverage cuts in both directions.
The same filing shows why that leverage is not clean. Newmont guided to by-product AISC of $1,680 per ounce for 2026, a rise of roughly 24%, attributing it to lower gold production, weaker by-product credits from copper, lead, and zinc, and higher royalties and production taxes tied to a higher expected gold price.
A rising metal price can therefore raise the cost of producing it. Layered on top are the ordinary hazards of owning an operating company, from permitting and jurisdiction to capital allocation and equity issuance.
A mining position is a bet on management as much as on bullion.
How Do Futures Change the Risk Profile?
Futures replace storage cost with margin and roll mechanics. The COMEX gold contract, according to CME contract specs, is 100 troy ounces, physically settled against gold assaying to a minimum of 995 fineness, with a minimum price fluctuation of $0.10 per ounce.
That tick is $10 per contract, and a single contract controls notional value in the hundreds of thousands of dollars at recent prices.
Leverage of that order changes what a drawdown means. What an unlevered holder can wait out can become a margin call for a futures holder, making the timing of an exit less discretionary.
The trade-off is transparency. The CFTC publishes Commitments of Traders data, usually every Friday at 3:30 p.m. Eastern using the prior Tuesday's open interest, broken out by trader category wherever 20 or more traders hold reportable positions.
Nothing equivalent exists for identifying who is buying coins.
Comparison of four ways to hold a precious metals position: physical bullion, bullion ETF, mining equity, and futures, across what you own, annual carry, how it trades, and the case against each.
What Is the Case Against a Metals Allocation?
The strongest decision-making case against the allocation is opportunity cost, and it can be quantified rather than treated as theoretical.
The 10-year TIPS yield stood at 2.76% on September 23, 2026, representing an inflation-adjusted yield available from a Treasury security. A hard-asset sleeve funded from that allocation gives up that real yield for price behavior that can run against the holder for two decades.
Demand composition is the second caution. The World Gold Council's Gold Demand Trends for the second quarter of 2026 put total demand, including over-the-counter activity, at 1,269 tonnes, flat year on year, with central bank net purchases of 289 tonnes.
ETF flows over the same quarter were negative 44.8 tonnes, compared with positive 171.1 tonnes a year earlier. Central bank demand of 345 tonnes in the first half of 2026 was the weakest first-half total since 2022.
The investment portion of the buyer base can reverse quickly, and a thesis resting on other investors continuing to buy is a different thesis from one resting on portfolio mathematics.
Then add the frictions already covered: no cash flow, a 28% ceiling on collectibles gains, custody requirements, spreads on both sides of the trade, and correlation properties that vary by market and period.
How Do Allocators Size and Review a Hard-Asset Sleeve?
The sizing decision should follow from the job description rather than precede it.
A position held for equity-drawdown protection should be measured against equity drawdowns, and an allocator can state in advance what it needs to do during a 20% index decline to count as working.
A currency-composition hedge is reviewed on a multi-year cycle and would not necessarily be judged broken by two flat years. Conflating the two mandates in one line item makes the position difficult to review, because almost any outcome can be rationalized after the fact.
Vehicle selection follows the same logic. Custody-driven reasons push toward allocated physical metal, with its premium and storage costs. Liquidity-driven reasons push toward listed trusts and their 0.25% to 0.40% annual drag.
A view on operating leverage belongs in equities, where management and jurisdiction come attached whether the holder wants them or not.
What separates a considered hard-asset sleeve from a reflexive one is usually documentation. The thesis gets written down, the vehicle is matched to it, the carrying cost is calculated across all its components rather than assumed, and the review date is set before the position opens instead of whenever the price next moves sharply.