Gold ETFs vs Physical: The Cost Gap at $4,709

Four ETFs dominate US gold flows. Their published expense ratios as of this writing:

• SPDR Gold MiniShares (GLDM) — 0.10%
• abrdn Physical Gold Shares (SGOL) — 0.17%
• iShares Gold Trust (IAU) — 0.25%
• SPDR Gold Shares (GLD) — 0.40%

GLD is the largest by far (over 800 tonnes in HSBC's London vault), but it's also the most expensive. The four-times spread between GLDM and GLD is pure structural — same gold, same custody framework, different management fee. There's no reason to pay 0.40% unless you specifically need GLD's deeper options market for hedging strategies.

Tracking is tight. GLD has historically tracked the London PM gold fix within 0.01-0.02% on a daily basis, with the small drag coming from the expense ratio. Over a full year, you can expect returns to trail spot by approximately the expense ratio plus a few basis points of bid-ask friction. On a $50,000 position, that's $50/year in GLDM versus $200/year in GLD — a $1,500 difference over a decade that you keep entirely by choosing the right ticker.

Tax treatment is identical across the four. All are structured as grantor trusts, meaning long-term gains are taxed as collectibles at a maximum 28% federal rate — the same rate that applies to physical gold, and meaningfully higher than the 20% rate on stocks. Holding in a tax-advantaged account (IRA or 401(k)) eliminates this gap entirely and is the single biggest tax move available to gold buyers.

Gold pays no coupon. Every dollar you hold in gold is a dollar you didn't put in a Treasury, a money-market fund, or an I-bond. The opportunity cost is the real yield — the nominal Treasury yield minus expected inflation.

Here's the current snapshot from FRED on May 4-5:

• 10-year Treasury (DGS10): 4.45%
• 5-year TIPS yield (DFII5): 1.36%
• 10-year TIPS yield (DFII10): 1.95%
• 10-year breakeven inflation (T10YIE): 2.47%
• March 2026 CPI (FRED CPIAUCSL): 330.29 — 3.29% YoY

The textbook claim is that gold rallies when real yields fall and struggles when they rise — though as the inflation-hedge record shows, the relationship is messier than the simple version suggests. The 2009-2011 rally and the 2020 melt-up both came with deeply negative real yields. The 2013 crash (-28% in a year) came when Bernanke's taper talk pushed real 10-year yields from -0.7% to +0.9%.

What's strange about the current regime is that gold is sitting near $4,700 with positive 1.95% real 10-year yields. That's not a tailwind — that's a four-percentage-point swing from the last secular gold bull market. The metal is holding because three other forces are pulling harder than carry: central-bank buying (the structural floor that reset gold's price discovery), Iran-war tail risk (still bid into the May 6 Hormuz pause), and a USD index at 118.39 that's stalled rather than strengthening.

Why this matters for the ETF-vs-physical decision: when gold is fighting positive real carry, every basis point of cost matters more, not less. A 5% coin premium on a metal already running into a +1.95% headwind needs three years of price appreciation just to break even versus an ETF held in an IRA. In the negative-real-yield regimes of 2010 or 2020, that math was forgiving; in 2026, it isn't.

Physical gold means actual metal in your possession or a private vault — coins, bars, or rounds. The most-traded forms in the US are American Gold Eagles, Canadian Gold Maple Leafs, and LBMA-accredited gold bars from refiners like PAMP Suisse, Valcambi, and Royal Canadian Mint.

Purchase premiums are the headline cost. As of early May 2026, dealer pricing on standard products at major firms (APMEX, JM Bullion, Kitco) clusters around:

• 1 oz Gold Eagle (US Mint): 4.5-6.5% over spot
• 1 oz Gold Maple Leaf (RCM): 3.5-5% over spot
• 1 oz Krugerrand: 3-4.5% over spot
• 1 oz LBMA-accredited bar: 1.5-3% over spot
• 10 oz LBMA-accredited bar: 0.8-1.8% over spot
• 1 kilo bar: 0.5-1.2% over spot

On a $4,709 spot price, a 1 oz Eagle costs roughly $4,920-5,015 delivered. A 1 oz bar runs $4,780-4,850. Those premiums are non-recoverable. When you sell, you'll typically receive spot or 1-3% below — meaning the round-trip bid-ask spread on a coin can total 5-10%.

Storage and insurance add a second layer of friction. A bank safe deposit box runs $100-300 per year and is not FDIC-insured for the contents. Private vault storage (Brink's, Delaware Depository, Loomis) charges 0.3-0.5% of the gold's value annually, with a floor that makes small positions uneconomic. Home storage is free but introduces theft risk and policy complications — most homeowner's insurance caps unscheduled valuables coverage at $1,000-2,500 unless you schedule a rider, which adds another 1-2% per year.

Liquidating is the part that surprises people. Selling means finding a dealer, agreeing on a buyback price (usually quoted live but firm only for that day), and either physically delivering the metal or shipping it insured. The process takes hours-to-days, not seconds, and the dealer's offer is typically 1-3% below spot for bars and slightly closer to spot for sovereign coins. Compare that to an ETF, where you sell at market in seconds at the bid.

Authentication risk is small but non-zero. Counterfeit bars and coins do exist, particularly for products purchased online from non-major sources or overseas. Sticking with sovereign-mint coins (Eagle, Maple Leaf, Britannia) and LBMA-accredited bars from established dealers reduces this to near-zero, but it's worth a small premium to avoid the tail.

Once you've decided to buy physical, four custody choices remain — and they're not interchangeable.

1. Home storage. Free, fully private, fully under your control. The trade-off is theft risk plus the homeowner's-insurance gap. Practical only for small positions (under ~$25,000) or for investors who already have a high-grade home safe and a documented inventory.

2. Bank safe deposit box. $100-300 per year, but the contents are explicitly not FDIC-insured and the bank has no liability for loss. Access is during banking hours only — useless if your scenario for owning gold is a bank-system disruption. Some banks have begun refusing to rent boxes for precious metals storage; check the contract before assuming.

3. Private vault — allocated. The standard institutional structure. Specific bars or coins are assigned to you by serial number; the vault holds them as a bailee, not as principal. Annual cost is typically 0.5-1.0% of value with a $50-150/month minimum. Brink's, Loomis, Malca-Amit, and Delaware Depository are the credible operators. Allocated storage is the only physical structure that survives a vault bankruptcy — the gold is yours, not the vault's, and a creditor cannot claim it.

4. Private vault — unallocated. Cheaper (often 0.1-0.3%) because the vault commingles your claim with others. You own a pro-rata share of a pool, not specific bars. Do not confuse this with allocated storage. In a vault bankruptcy, unallocated holders are general creditors — they stand in line behind employees and secured lenders. The cost saving is real but so is the counterparty risk, which is what most physical-gold buyers thought they were avoiding by going physical in the first place.

For IRA-eligible gold, only specific products qualify under IRC § 408(m): American Gold Eagle, American Gold Buffalo, Canadian Gold Maple Leaf, Australian Gold Kangaroo, plus 0.995-purity bars from approved refiners. Self-directed IRA custodians (Equity Trust, Strata, Kingdom Trust) typically charge $200-300/year for the account plus depository fees of 0.4-0.6%, putting the all-in cost at roughly 0.8-1.2% annually before any premium paid on the metal itself. That's 8-12x the 0.10% you'd pay holding GLDM in a Vanguard IRA — a brutal comparison unless the physical-IRA buyer specifically wants metal in tax-deferred form.

A simple decision rule:

• Position under $25,000 or no specific crisis hedge in mind? Skip physical entirely. ETFs in an IRA dominate on cost.
• Position $25,000-$100,000 with an explicit insurance thesis? Allocated vault storage at a credible operator. Avoid unallocated.
• Position over $100,000 with a long horizon? Allocated vault, possibly with a small home-storage tranche for true tail-risk scenarios.
• Self-directed IRA holding physical? Justified only if you've priced the 0.8-1.2% drag and you specifically want tax-deferred treatment of metal you intend to hold for 20+ years.

ETF counterparty risk is real but bounded. GLD, IAU, GLDM, and SGOL are structured as grantor trusts, not ETPs that use swap counterparties. The gold is held by the custodian (HSBC for GLD, JPMorgan for IAU, BofA for GLDM, JPMorgan for SGOL) and audited semiannually by an independent inspector. The counterparty risk is therefore custodial, not derivative — closer to allocated vault storage than to a synthetic product. This is one of the more under-appreciated points in the ETF-vs-physical debate: the major US gold ETFs are structurally closer to physical-allocated than physical-unallocated.

The cost differential compounds. Take a $50,000 gold investment at the May 6 spot of $4,709, held for 10 years. Three structures:

Gold ETF (GLDM at 0.10% expense ratio, held in an IRA):

• Purchase: $50,000 at market, ~$0 commission
• Annual cost: $50
• 10-year cost: ~$500
• Tax treatment on sale (in IRA): zero capital gains drag at withdrawal age
• Sell: market price, ~$0 commission

Physical bars (1 oz LBMA-accredited at 2% premium, allocated vault storage at 0.5%):

• Purchase cost: $51,000 ($50,000 + 2% premium)
• Annual storage: $250
• 10-year storage: $2,500
• Sell: ~$49,000 (1-2% below spot)
• Effective drag: roughly $4,500-$5,500 vs ETF

Physical coins (1 oz Eagles at 5% premium, home storage):

• Purchase cost: $52,500 ($50,000 + 5% premium)
• Annual storage: $0 (assumes adequate home safe)
• 10-year storage: $0
• Sell: ~$48,500 (~3% below spot for coin buybacks)
• Effective drag: roughly $6,000-$7,000 vs ETF

Over a decade, the cost difference is roughly $4,500-$7,000 in favor of an ETF held in a tax-advantaged account. That's not a small number against a $50,000 position — it's a 9-14% headwind that the gold price itself has to overcome before the physical buyer even ties the ETF buyer.

The math gets worse for smaller positions. A $10,000 physical coin allocation at 5% premium plus 1% storage minimums looks structurally broken. A $200,000 allocated-vault position at sub-1% all-in is competitive with a taxable-account ETF (since the 28% collectibles rate applies to both) but still loses to an IRA-held ETF.

The math gets better in true crisis scenarios. If your thesis is system-level financial disruption — bank holidays, capital controls, custodian seizures — the ETF advantage evaporates because the entire ETF wrapper depends on the same plumbing your thesis assumes will fail. This is the only scenario where physical gold's cost premium reliably pays. If you can't articulate the specific failure mode, you're paying 5-7% to hedge a risk you haven't sized.

The case for physical isn't zero — it's narrow. Three scenarios where physical's cost premium genuinely pays:

Counterparty-risk elimination during financial-system stress. Physical metal in a vault under your name is yours regardless of what happens to your broker, the ETF custodian, or the federal regulatory framework. ETFs are claims on a custody chain. In normal markets the chain is robust. In a 2008-style stress event with custodian-level fragility, the chain matters. This is the only scenario where physical's premium reliably pays. If you can't name the specific failure mode you're hedging — "my broker fails and the SIPC bridge breaks" or "the IRS reclassifies ETF gains mid-cycle" — then you're paying 5-7% upfront for a tail you haven't sized.

Privacy below reporting thresholds. Physical gold transactions below $10,000 don't trigger automatic IRS Form 8300 reporting. ETF transactions appear on every brokerage 1099-B. For investors who specifically value financial privacy on a small portion of their wealth, physical offers a disclosure profile that paper gold cannot match. The $10,000 limit is small, the use case is narrow, but it's real.

Estate planning leverage. Physical gold has a step-up in basis at death like any other asset, and physical coins at fair-market value can be distributed in-kind without forcing a sale. ETF shares are simpler administratively — they sell, distribute cash, and rebalance — but physical gold occasionally fits an estate plan that wants asset-specific bequests.

What physical gold is not good for: tactical positioning, rebalancing, dollar-cost averaging, monthly income strategies, or any allocation under ~$25,000. The friction makes those use cases structurally uneconomic.

The tangibility argument deserves an honest mention. Some investors are simply more comfortable holding gold they can touch, and a comfortable investor is a better long-term holder than an uncomfortable one. If a 5-7% upfront premium is what it takes for you to keep a 5-10% gold allocation through a 20% drawdown, that premium might be the cheapest insurance in the portfolio. Just don't confuse the comfort dividend with a financial argument.

Most diversified investors hold 5-10% of their portfolio in gold. On a $500,000 portfolio, that's $25,000-50,000. The tax structure of that allocation matters more than the metal-vs-paper choice for almost every investor below the $250,000 gold-allocation threshold.

Step 1: Default the bulk of the position to a low-cost ETF in a tax-advantaged account. GLDM at 0.10% in an IRA is the structural starting point. The 28% collectibles rate disappears, the cost drag is minimal, and the position rebalances cleanly with the rest of the portfolio.

Step 2: Decide whether you want a physical insurance tranche. If yes, size it deliberately — typically 10-20% of the gold allocation, or roughly 1-2% of total portfolio. On a $500,000 portfolio, that's a $5,000-10,000 physical position.

Step 3: Pick the physical structure that fits the tranche size. Under $25,000 → home storage with sovereign coins. $25,000-100,000 → allocated vault. Over $100,000 → split between allocated vault and a smaller home-storage tranche.

Step 4: Don't trade the physical tranche. Premiums and friction make tactical physical gold a losing structural bet. If you want to add or trim gold exposure based on macro views, do it through the ETF tranche.

On the macro setup right now: the gold market is digesting a 10% pullback from February's peak against a backdrop of Fed funds at 3.64%, 10-year Treasuries at 4.45%, and 10-year TIPS yielding +1.95% in real terms. The metal is fighting positive real carry — historically a structural headwind — but holding because central-bank demand and geopolitical tail risk are pulling the other way. The Gold ETF VIX (GVZ) at 26.61 is elevated but well below the March panic prints above 35 — implied volatility is expecting the chop to continue, not resolve.

In that environment, picking up 30-40 basis points a year by switching from GLD to GLDM, or skipping a 5% coin premium on a tactical position you're going to flip in 18 months, is among the few high-confidence gold trades available. The metal might rally; it might roll over. The cost wedge is the part you control.