Week 6: Gold — The 5,000-Year Store of Value

---

1. Why This Is Important

Every textbook portfolio chapter eventually arrives at the same awkward question: should you own gold? And usually the answer is either an evangelical yes (the goldbug answer) or a dismissive no (the Bogleheads answer). Both are wrong because both are arguing about the wrong thing. Gold is not a stock. It is not a bond. It does not produce cash flow, will never pay you a dividend, and its "intrinsic value" — as J.P. Morgan testified to the Pujo Committee of the US House of Representatives in 1912 —

"Gold is money. Everything else is credit." — J.P. Morgan, House Banking & Currency Committee, 19 December 1912 (sworn testimony, Money Trust Investigation, p. 1084)

is exactly the same as the intrinsic value of every other store of value: consensus.

You need to think clearly about gold for five reasons.

**It is the cleanest test of what "store of value" actually

means.** Stocks compound earnings. Bonds compound coupons. Gold compounds (almost) nothing. If gold is worth holding, the reason cannot be cashflow — it has to be the durability of belief that other humans will accept it as money in fifty years. That same logic applies to fiat dollars, to bitcoin, and to every currency that has ever existed. Gold is the laboratory for thinking about belief-priced assets.

The inflation-hedge claim is half true and half marketing.

Gold has crushed inflation in three windows (the 1970s, the 2000s, the 2020s) and gone nowhere for decades in between. If you do not understand when the hedge works, you will buy at a peak and sell at a trough — exactly what the average gold ETF investor has done since 2011.

**Central banks have come back as the structural marginal

buyer.** After two decades of net selling (1990–2008), official sector purchases turned positive in 2010 and accelerated sharply after the 2022 freezing of Russian FX reserves. The World Gold Council reports central banks bought a record 1,082 tonnes in 2022 and another 1,037 tonnes in 2023 — the two largest annual totals on record. This is a different marginal buyer than the 2011 ETF cycle, and it changes how you should think about the 2020s gold rally.

The implementation choice is huge. Physical bullion ETFs

(GLD, IAU, GLDM, PHYS, MNT), gold futures, gold-mining equities, and physical coins are four very different exposures with four very different tax, fee, custody, and counterparty profiles. The default answer for a US-listed retail book is a physical-bullion ETF; everything else is a specialty tool.

"No cashflow" is no longer strictly true. Bullion banks

pay a small lease rate to borrow your allocated gold, prime-broker desks accept gold as eligible collateral, and under the Basel III NSFR rules that took effect in 2021 (EU) and 2022 (UK), allocated physical gold is treated as a high-quality liquid asset for banks while unallocated gold carries an 85% required stable funding charge. That regulatory shift quietly altered the cost structure of the paper-gold market.

This lesson covers what gold actually is, the 5,000-year history behind the belief, when its inflation hedge has worked, the role of central banks, the silver-and-platinum-group cousins, why industrial commodities are a different beast, and the small toolkit a retail investor should actually use.

A note on what gold is not primarily about. Gold has real industrial demand (electronics, dentistry) and large jewellery demand — together roughly half of annual flow demand per the World Gold Council's Gold Demand Trends series. But the investment thesis for holding gold in a portfolio is overwhelmingly about monetary belief and reserve demand, not jewellery fashion or industrial usage. When the gold price moves $200 in a quarter, the marginal driver is almost always central banks, ETF flows, or real-rate moves — not necklaces.

---

2. What You Need to Know

2.1 A Brief 5,000-Year History of Gold as Money

The reason "gold = money" feels obvious is that humans have been running the experiment, more or less continuously, since the Bronze Age. A short tour:

• c. 2500 BC — Egypt and Mesopotamia. Gold rings and bars

circulate as a unit of account in temple economies. Gold is chosen because it does not corrode, can be assayed by weight, and is rare enough that small amounts carry large value.

• c. 600 BC — Lydia (modern western Turkey). King Croesus

issues the first standardized gold-silver (electrum) coins — the first true coined money. The British Museum's Lydian-coin collection is the founding artifact of monetary history.

• 49 BC – AD 300 — Roman aureus. A 7.8-gram gold coin

anchors Roman commerce for three centuries. Successive debasements after AD 250 — emperors clipping gold content to fund military campaigns — are the textbook example of "fiat failure even when the fiat is metal."

• AD 1252 — Florence's fiorino d'oro. A pure-gold coin

becomes the reserve currency of Renaissance Europe. The Medici bank's correspondence network runs on it.

• 1717 — Britain's de facto gold standard. Sir Isaac Newton,

as Master of the Royal Mint, sets the gold-silver ratio at the level that made gold the dominant standard. The classical gold standard runs effectively from 1717 to 1914.

• 1944 — Bretton Woods. $35 per ounce, dollar convertible to

gold for foreign central banks, every other currency pegged to the dollar. The system runs for 27 years.

• 15 August 1971 — Nixon closes the gold window. Convertibility

ends. Gold floats. The 5,000-year monetary role of gold formally ends and the global financial system enters its first ever pure-fiat era — now in its 55th year.

Two takeaways for an investor in 2026. First, the current fiat era is historically unusual; every previous fiat regime ended in either restoration of metallic backing or hyperinflation. That does not predict the future, but it should temper anyone who treats fiat money as the natural state of affairs. Second, gold's "belief" is not a vibe — it is a 5,000-year continuous record of humans across every civilization independently selecting the same metal as the unit of last resort. That is the durability the chart below is implicitly pricing.

2.2 Why Gold Is Belief-Priced — and What "No Cashflow" Really Means

Stocks are claims on future earnings. Bonds are claims on future coupons. The discounted-cashflow framework that values both of them does not apply to gold because gold has no cashflow to discount. The price of gold is whatever the marginal buyer is willing to pay the marginal seller, and the marginal buyer's willingness to pay is set entirely by their belief that the next marginal buyer will pay at least as much.

This is not a slur. It is the same logic that prices the US dollar. A hundred-dollar bill produces no cashflow either. Its value is that the next person down the chain will accept it for $100 of goods. Voltaire's eighteenth-century one-liner captures the asymmetry:

**"Paper money eventually returns to its intrinsic value — zero."** — attributed to Voltaire, 1729

Gold's belief is 5,000 years old. The pure-fiat dollar's belief, in its current post-Bretton-Woods form, is younger than the Beatles. Bitcoin's belief is younger than the iPhone. **Every store of value rests on belief**; the only honest question is how durable the belief has been and how durable it is likely to remain.

Important caveat. Durability of belief is not the same as a guaranteed future return. Gold's 5,000-year record makes the belief extremely robust. It does not guarantee that the next ten years will look like any particular previous ten. Goldbugs who assume they will get the 1971–1980 return again because "gold always wins in inflation" are extrapolating one regime onto another. Treat the historical track record as a constraint on what is plausible, not as a forecast.

The practical implication of "no cashflow" is that gold cannot compound the way equities do. A diversified equity index delivered roughly 7% real per year over the long run because companies kept producing more goods, paying dividends, and buying back stock. Gold, over the long run, delivered roughly 1.0–1.5% real — slightly above inflation. It is *money that holds its value*; it is not capital that grows.

One important modern wrinkle on "no cashflow." Gold can in fact be lent out for a small yield. The London bullion market publishes the Gold Forward Offered Rate (GOFO) — the rate at which dealers will swap gold for dollars — and the implied gold lease rate is the difference between USD LIBOR/SOFR and GOFO. Bullion banks (HSBC, JPMorgan, UBS, ICBC Standard) borrow allocated bullion from large holders to satisfy short-term delivery and refining demand, paying a lease rate typically in the 0.1%–2% range, occasionally spiking higher in tight markets (it briefly went over 5% in early 2020 during COVID delivery chaos). That changes "no cashflow" to "very small cashflow if you are a large enough holder to access the wholesale market."

For retail investors, three things matter about this:

You will not get the lease rate directly unless you are

holding tonnes in an LBMA-recognised vault. The big physical gold ETFs do not lend out their bullion (GLD, IAU, GLDM, PHYS all explicitly prohibit it in their prospectuses). That is a feature, not a bug — see §2.10 on paper vs physical.

Some smaller, "yield-enhanced" gold products do lend. Read

the prospectus before buying anything that promises a yield on gold. The yield comes from counterparty credit risk — when the bullion bank fails, your "gold" is an unsecured claim, not a bar.

Gold is increasingly accepted as collateral. Prime brokers

will lend dollars against allocated bullion, typically at 80–90% loan-to-value. CME Group accepts gold as eligible margin collateral for futures positions. This makes gold a slightly "active" asset for sophisticated balance sheets — but only at scale.

Basel III and the quiet 2021–2022 regulatory shift. In June 2021 (EU) and January 2022 (UK), the Basel III Net Stable Funding Ratio (NSFR) rules came fully into force for the bullion market. Allocated physical gold held in a bank's vault qualifies as a high-quality liquid asset with no required stable funding penalty. Unallocated gold (paper claims on a pool, the form most over-the-counter gold trades take) carries an **85% required stable funding charge** — meaning the bank has to fund 85% of its unallocated gold book with long-term liabilities. The practical effect is that the cost of running the unallocated paper-gold market went up materially, and the cost of holding allocated physical gold went down. Several analysts (Alasdair Macleod at Goldmoney; Ronan Manly at BullionStar) have argued this is one of the structural reasons the gold price has held up better than the real-rate model alone would predict in 2022–2026. Whether you buy that thesis or not, the regulatory shift is real and is quietly reshaping the bullion market in the background.

2.3 The Inflation-Hedge Claim — When It Holds, When It Doesn't

The marketing claim is "gold hedges inflation." The historical record is more specific: **gold hedges sharp, persistent, and poorly-anticipated inflation, in regimes where real interest rates are falling.** When real rates are rising (Volcker 1981, the 2010s, briefly 2022 before the bond market broke), gold tends to go nowhere or down.

The chart below has two panels. The top panel plots the real (CPI-adjusted) USD gold price from 1971 — the year Nixon closed the gold window and floated the dollar — through April 2026. The bottom panel overlays headline CPI year-over-year against the real US 10-year yield, with the annual gold price change shown as bars, so you can see the relationship between the two macro variables and gold's response.

Three things should jump out from the top panel.

• Three real peaks at roughly the same level. The 1980, 2011,

and 2024–26 highs all sit between $2,200 and $2,500 in 2026 dollars. That repetition is striking — one interpretation is that there is a "marginal demand ceiling" above which gold gets expensive enough relative to other inflation hedges (TIPS, real estate, commodities) that the marginal buyer steps away. Treat that as a working hypothesis, not a proven law. Three data points are not a regime.

• Two lost decades. From 1980 to 1999 the real price fell

almost 80%. From 2012 to 2018 it lost about 40%. Holding gold through those windows tested every goldbug's faith.

• The rallies cluster around macro stress. 1971–80:

stagflation, oil crisis, end of Bretton Woods. 2001–11: dollar weakness, GFC, eurozone crisis, rates collapsing toward zero. 2019–26: COVID liquidity, post-COVID CPI shock, central bank buying as reserve diversification accelerated.

The bottom panel makes the real-rate relationship explicit. For most of the 1971–2020 sample, the simplest one-line model of gold is: real US 10-year yield down → gold up; real yield up → gold down. The mechanism is opportunity cost. If you can earn 2% real on a Treasury, the no-yield asset is expensive. If you can only earn -1% real on a Treasury, gold's zero yield suddenly looks competitive.

But real rates are the main driver, not the only driver. The 2022–2026 leg is the cleanest counter-example. Real 10-year yields rose from -1% to over +2% during this window — the textbook setup for gold to fall. Gold rose anyway, by roughly 70% in real terms. What broke the model: record central bank buying (see §2.5), geopolitical reserve diversification after the freezing of Russian FX reserves in February 2022, US fiscal deficits running near 7% of GDP outside of recession, and a growing market discount for currency-debasement risk. As Ray Dalio put it in his 2024 LinkedIn essay *Why I Like Gold More Than Bonds Now*, gold is a "non-credit, non-fiat money" whose demand rises when both the credit-money system and the geopolitical system are under stress — even if real rates are also rising. Use real rates as the primary lens, but do not be a one-factor investor.

2.4 Gold vs Stocks vs 60/40 — Regime by Regime

The traditional way to compare gold and stocks is one cumulative chart from 1971 to today. The problem with that picture is that the long-run line is dominated by whatever happened last decade — you cannot see when gold won and when it lost. The chart below splits the half-century into five regimes and starts every line at $1 at the beginning of each regime, so you can see the shape of returns inside each window. A fourth line shows a 60/25/15 portfolio — 60% S&P 500, 25% intermediate Treasuries, 15% gold — rebalanced annually, alongside the 60/40 stock/bond benchmark.

Read the panels regime by regime.

• 1971–1980 (stagflation, end of Bretton Woods). Gold up

~16x in nominal terms, ~5x in real. S&P 500 essentially flat in real terms. 60/40 lost real value. The 60/25/15 portfolio ended materially ahead of 60/40, almost entirely because of the gold sleeve. This is the regime gold is paid for.

• **1981–1999 (the Volcker disinflation and the great bull

market).** S&P 500 compounded at roughly 14% real per year. Gold lost ~80% in real terms. 60/40 ran neck-and-neck with equities at lower volatility. 60/25/15 underperformed 60/40 by the gold drag — about 80–100 bps a year. This is the regime gold makes you pay.

• 2000–2011 (dot-com bust and GFC). Gold returned ~15% per

year compounded. S&P 500 was roughly flat for the full window (the "lost decade"). 60/40 outperformed 100% equities. 60/25/15 beat 60/40 again — gold was the winning sleeve.

• 2012–2019 (post-GFC disinflation). S&P 500 ran. Gold flat

to down. 60/40 came back. 60/25/15 lagged 60/40 modestly.

• 2020–2026 (COVID, reflation, central-bank gold cycle).

Stocks roughly doubled. Gold roughly doubled. 60/40 had its worst year on record in 2022 (stocks and bonds both down). 60/25/15 had a noticeably softer 2022 because gold rose while stocks and bonds fell — the regime in which a permanent gold sleeve earns its keep.

• Full 1971–2026 summary. S&P 500 has the highest end-value

by a meaningful margin. 60/25/15 has a higher Sharpe ratio than either 60/40 or 100% S&P 500, with a much shallower maximum drawdown. The price you pay for that smoother ride is ending with somewhat less wealth than 100% equities — a trade most investors should accept.

What the regime panels make obvious is the case that Harry Browne built his Permanent Portfolio (1981) on, that Ray Dalio later refined as Bridgewater's All Weather (1996), and that Christopher Cole at Artemis Capital articulated again in 2020 with the Dragon Portfolio white paper *The Allegory of the Hawk and Serpent*: gold and equities are different exposures. Holding a meaningful slice of each tends to produce a better risk-adjusted outcome across multiple regimes than either alone.

A note on language. Calling this a "free lunch" is too strong. What the data shows is a **realized diversification benefit** — over the specific 1971–2026 sample, the rebalanced mix had a higher Sharpe ratio than its components. There is no guarantee future regimes will deliver the same benefit. The case for the gold sleeve is "robust across many historical regimes," not "guaranteed in any future one."

2.5 Central Banks — The Structural Marginal Buyer Since 2010

For most of the post-Bretton-Woods era, central banks were net sellers of gold. Western central banks unloaded reserves in the 1990s and early 2000s — the "Brown's Bottom" episode in 1999, when UK Chancellor Gordon Brown sold roughly half of British gold reserves at $282/oz, is the case study for selling at the lows. That regime ended in 2010. Every year since 2010, the official sector has been a net buyer.

The numbers from the World Gold Council's 2025 *Gold Demand Trends* publication:

• 2010–2021: average ~480 tonnes/year of net official sector

purchases, dominated by Russia, China, Turkey, India, and Kazakhstan.

• 2022: 1,082 tonnes — the largest annual purchase since 1950.

The trigger was the freezing of $300 billion of Russian central bank dollar reserves after the invasion of Ukraine. Every non-aligned central bank suddenly had a vivid reason to want a reserve asset that no one else can switch off.

• 2023: 1,037 tonnes.
• 2024: ~1,045 tonnes.
• 2025 (full year estimate): still tracking above 900 tonnes,

with the People's Bank of China the most-watched buyer (PBoC reported gold reserve increases for 18 consecutive months through mid-2025).

For context, the entire annual newly-mined gold supply is roughly 3,500 tonnes. Central banks alone have been absorbing ~25–30% of new mine supply in the recent cycle. That is structural marginal demand of a kind the 2011 ETF cycle did not have.

The strategic implication is straightforward. The 2010s gold cycle was driven by Western retail and ETF flows, and it exhausted itself when real rates rose in 2013–18. The 2020s gold cycle is driven by non-Western central banks who do not care about real rates because their alternative — holding more dollars and Treasuries — was demonstrated in 2022 to carry political risk they had not previously priced. That marginal buyer does not turn off when real rates rise. It might slow down, but it is unlikely to flip back to the net seller it was in the 1990s.

2.6 Silver and the Other Precious Metals

Gold's "purity" as a store of value comes from a property the others lack: gold has essentially no industrial demand. Less than 10% of annual gold flow goes to industry (mostly electronics and dentistry). The rest is jewellery, investment, and central bank reserves — i.e., monetary or quasi-monetary demand. That is why gold's price tracks belief and real rates rather than the global manufacturing PMI.

Silver, platinum, and palladium are different. They blend a monetary identity with a much larger industrial demand component, and that mix is exactly why they are more volatile and behave less cleanly as portfolio diversifiers.

Silver — the bimetallic ghost. From the late Roman Empire through the 19th century, silver was as much "money" as gold, and most of the world ran on a bimetallic standard. The British Crown's adoption of a de facto gold standard in 1717 (via Newton) and the US Coinage Act of 1873 (the "Crime of '73," demonetizing silver) were the two political events that ended silver's monetary role. William Jennings Bryan's "Cross of Gold" speech at the 1896 Democratic Convention was the last serious political fight to bring silver back. He lost. Industrial demand has carried silver's price ever since — today roughly 55% of silver demand is industrial (solar PV, electronics, brazing alloys), 25% is jewellery and silverware, and the residual ~20% is investment and coins.

The most-watched ratio is the gold-silver ratio — ounces of silver that buy one ounce of gold. Historically (Roman through 19th century) it sat in a 12:1–16:1 band, which is roughly the ratio of their crustal abundance. In the modern era it has ranged from ~15:1 (the 1980 Hunt brothers' silver squeeze peak) to over 120:1 (the COVID March-2020 spike). The chart below shows the full 1968–2026 history.

Two practical observations:

• The ratio mean-reverts but slowly and noisily. Silver bulls

treat ratio extremes (>100:1) as a signal to overweight silver vs gold, and historically the 1991, 2003, 2016, and 2020 extremes did each get worked off — but it took years, and silver's path during the wait was vastly more volatile than gold's.

• For a portfolio diversifier, gold is the cleaner instrument.

For a tactical trade on the next industrial-cycle turn, silver has more torque.

Platinum and palladium — the autocatalyst metals. Both are roughly 80% industrial-demand-driven, dominated by autocatalysts (palladium for petrol vehicles, platinum for diesel and hydrogen). Their prices track auto-production cycles, emissions regulations, and South African / Russian mine supply (those two countries produce ~75% of platinum-group metal output). They are not stores of value in any meaningful sense — palladium fell 65% in nine months in 2023 after the EV-driven structural de-rating finally caught up with the price. Treat them as specialty industrial-cycle bets, not as gold substitutes.

The investment-only takeaway: when the lesson talks about a "permanent metals sleeve," it means gold. Silver, platinum, and palladium can have a place in a tactical book but not in the permanent one.

2.7 A Note on Industrial Commodities — Why They're Different from Gold

Oil, copper, wheat, soybeans, natural gas — these are inputs to production, not money. They differ from gold in three structural ways.

They mean-revert toward the marginal cost of supply. When

oil hits $130 in 2008 or 2022, US shale wakes up, supply rises, and the price falls back toward the $50–$70 marginal-cost band. When oil drops to $30, marginal producers shut in, supply falls, price rises. Over a long horizon, an industrial commodity does not compound; it oscillates around a slowly drifting cost curve.

The real long-run trend is down, not up. Productivity

in extraction and farming has historically beaten demand growth. Real oil prices in 2026 are roughly the same as in 1973. Real wheat prices have fallen over a century. The "commodity supercycle" thesis is real for individual decades but not for a buy-and-hold compounder. Jeremy Grantham's GMO research papers on long-run real commodity prices are the standard reference.

**They are correlated with each other and with the business

cycle.** Owning a basket of industrial commodities is roughly owning a leveraged short on the dollar plus a leveraged long on global GDP. It is a factor exposure, not a diversifier.

For US retail investors, the implication is that industrial commodities are trades (positions sized for a specific macro view) and not holdings (permanent allocations). The mechanics of futures-based commodity ETFs — contango, roll cost, the USO case study — are covered in detail in **Week 39: Commodity Futures and Roll Yield**, where they belong. For this lesson, the takeaway is just that physical-bullion gold ETFs do not share that structural problem (see §2.8) — and that is the specific reason gold is investable as a passive long-term holding while most other commodities are not.

2.8 The Investable Toolkit — Bullion ETFs vs Gold Futures vs Coins

Per the US-only investable mandate, here is the gold-specific shortlist for a US-listed retail book.

Vehicle	Type	Expense	Notes
GLD	US-listed physical-bullion ETF (trust)	0.40%	Largest, most liquid, deepest options chain
IAU	US-listed physical-bullion ETF (trust)	0.25%	Cheaper than GLD, slightly less liquid
GLDM	US-listed physical-bullion ETF (trust)	0.10%	Cheapest US bullion ETF; smaller options market
SGOL	US-listed physical-bullion ETF (Swiss vault)	0.17%	Bullion held in Zurich, not London/NY
BAR	US-listed physical-bullion ETF	0.18%	Allocated, audited monthly
IAUM	US-listed physical-bullion ETF	0.09%	iShares' answer to GLDM
GDX	Gold mining equities	0.51%	Levered, operational risk; ordinary stock tax treatment
GDXJ	Junior gold miners	0.52%	Higher beta, much higher idiosyncratic risk
/GC futures	CME 100-oz gold futures	n/a (margin)	Section 1256 60/40 tax treatment; full physical-delivery contract
MGC futures	CME 10-oz micro gold futures	n/a (margin)	Same Section 1256 treatment, smaller contract size

For investors with access to Canadian listings (Interactive Brokers, Questrade, most international brokers), the toolkit expands materially:

Vehicle	Type	Expense	Notes
PHYS	Sprott Physical Gold Trust (NYSE Arca / TSX)	0.41%	Closed-end fund, allocated bullion, physical redemption available in 400-oz bars
KILO	Sprott Physical Gold Trust (kilobar)	0.35%	Same Sprott trust structure, kilobar redemption
MNT	Royal Canadian Mint Gold ETR (TSX)	0.35%	Receipts backed 1:1 by allocated bullion in RCM vaults; physical redemption
MNS	Royal Canadian Mint Silver ETR (TSX)	0.45%	Silver equivalent of MNT
CGL.C	iShares Gold Bullion ETF, CAD-hedged (TSX)	0.55%	CAD-hedged exposure for Canadian-dollar accounts

The Sprott (PHYS) and Royal Canadian Mint (MNT) products are worth flagging specifically because they are the two US-tax-friendly ways for an American holder to escape the 28% collectibles rate (see §2.11 on tax treatment): both qualify for Passive Foreign Investment Company (PFIC) QEF election treatment, which converts gains to long-term capital gains rates when properly elected on Form 8621 each year. The QEF election is real paperwork — most retail investors will not bother — but it exists and is used by sophisticated holders.

Gold futures versus physical-bullion ETFs. These are two very different exposures even though both reference the same spot price.

• Bullion ETFs (GLD, IAU, GLDM, PHYS, MNT) hold *physical

gold bars* in vaults. There is no roll cost, no contango. The only drag is the management fee (9–55 bps depending on the fund). Tracking to spot is tight — 0.1–0.5% per year in practice.

• Gold futures (/GC, MGC) are dated contracts. The holder

must close, take delivery, or roll into the next contract every two months. Gold's futures curve is *normally in mild contango* — the cost of carry (storage + insurance + financing) is positive, so deferred contracts trade slightly above spot. The roll cost on gold is modest (typically 1–3% annualised in normal regimes, occasionally higher in tight markets) — far smaller than the 5–10% annual contango drag on oil futures — but it is not zero. The compensating advantages are leverage (a /GC contract notional of ~$240,000 in April 2026 requires only ~$11,000 in margin) and Section 1256 tax treatment: 60% long-term / 40% short-term capital gains regardless of holding period, marked to market at year-end.

For a buy-and-hold gold sleeve, a physical-bullion ETF is the right tool. For a tactical or leveraged gold position, gold futures or options on GLD are usually better. The two coexist for different jobs.

Physical coins and bars. For investment purposes the mainstream coin products (American Gold Eagle, Canadian Maple Leaf, South African Krugerrand, Austrian Philharmonic) trade at 3–6% premium to spot, plus storage and insurance (0.5%/year for an insured commercial vault) and a 1–3% bid-ask on resale. The all-in friction wipes out the GLD/IAU expense advantage over any holding period under 10–15 years. Physical is the right answer only if you have a specific use case that paper gold cannot serve — primarily off-grid jurisdictional risk or estate planning across multiple legal systems.

2.9 Paper Gold vs Physical Gold — Allocated, Unallocated, and Lending Risk

Not all "gold" is gold. The bullion market distinguishes three custody models:

• Allocated, segregated, audited. Specific bars (with serial

numbers) belong to you, held at a named LBMA-recognised vault, audited at least annually by an independent firm. You have a property right to the metal. If the custodian fails, the bars do not enter the bankruptcy estate. This is what GLD, IAU, GLDM, PHYS, SGOL, BAR, MNT all use.

• Allocated but pooled. A specified weight of gold is yours,

but not specific bars — your claim is satisfied from a collective pool. Slightly weaker but generally adequate for retail.

• Unallocated. A book-entry credit for a quantity of gold

on a bullion bank's balance sheet. You are an unsecured creditor of the bank. If the bank fails, your "gold" is a bankruptcy claim, not a bar. Most over-the-counter gold trading in London happens in unallocated form — it is what makes the market liquid — but it is not the same product as owning a bar.

Why this matters for ETF selection. Read the prospectus for any gold ETF you are considering and check three things:

Custody form. The major US bullion ETFs (GLD with HSBC,

IAU with JPMorgan, GLDM with HSBC, SGOL with JPMorgan in Switzerland) all hold allocated gold and disclose bar lists. PHYS and MNT also hold allocated and offer physical redemption, which is the strongest possible form of "I really own this."

Lending policy. Does the trust permit lending out its

bullion? GLD, IAU, GLDM, SGOL, PHYS, and MNT all explicitly prohibit lending. Some smaller "yield-enhanced" gold products do lend — typically those that quote a positive yield on a non-dividend-paying asset. The yield comes from counterparty exposure to bullion banks. There is no large, broadly-marketed US ETF that lends its gold en masse; the risk lives more in unallocated wholesale products and in small-issuer "yield-enhanced" structures aimed at hedge funds.

Audit cadence and auditor. Annually at minimum, by a

recognised firm (Inspectorate International, Bureau Veritas). The bar list should be public.

The general rule: **for a retail core gold sleeve, pay the few extra basis points to hold an ETF that owns allocated, unlent, audited bullion**. The cost difference between GLDM at 10 bps and a hypothetical "yield-enhanced" product at 0 bps is trivial relative to the counterparty risk you take on by accepting the yield.

2.10 Sizing — How Much Gold Should You Own?

The honest answer is more than 0%, less than 100%, and the exact number depends on your view of the next ten years' regime. Three reference points from named investors:

• *Harry Browne, Fail-Safe Investing* (1981) — Permanent

Portfolio:** 25% gold, 25% stocks, 25% long Treasuries, 25% cash, rebalanced annually. The first systematic case for a large permanent gold sleeve from a quantitatively-minded retail framework.

• **Ray Dalio, Bridgewater Associates — All Weather (1996,

refined 2010s):** roughly 7.5% gold (and another ~7.5% commodities), inside a risk-parity framework where each asset is sized to contribute equally to portfolio volatility. In Dalio's 2024 LinkedIn essay *Why I Like Gold More Than Bonds Now*, he argues gold should be a meaningful portion of any diversified portfolio "in the 10-15% range" given current fiscal and geopolitical conditions.

• **Rick Rule (former CEO Sprott US Holdings) — recurring

interview guidance, 2020–2025:** "10% of your investable net worth in gold and silver, with two-thirds gold and one-third silver, as catastrophe insurance — and you should hope the insurance never pays off." Rule explicitly frames the allocation as insurance against the failure of the fiat-currency system, not as a return-seeking position.

How big is the global gold market relative to financial assets? The chart below puts the current allocation in context.

Three observations from the chart that frame the sizing question:

• The long-run average is around 5%. In other words, if every

investor on the planet held the global market portfolio of financial assets, they would each hold roughly 5% in gold.

• The current global allocation is roughly 1.7%. Investible

gold is meaningfully underweight its long-run average. If global allocations mean-reverted to the historical 5% level, with the supply of gold roughly fixed (mine supply growth ~1.5%/year), the implied gold price would be roughly **2.5–3x current levels** — call it $7,000–$8,500/oz. This is not a forecast; it is a sensitivity calculation that goldbugs cite often. Take it as illustrative.

• The 1980 peak hit ~13%. A return to that "panic" level,

rather than the long-run average, would imply gold prices in the $15,000+ range. Again — illustrative, not predicted.

For a US retail investor, three honest sizing answers:

• 0% is the Bogleheads answer. It is internally consistent

if you believe inflation will stay tame and real rates will stay positive forever. The 60/40 backtest from 1981–2021 supports it. The 2022 result does not.

• 5–15% is the institutional consensus — Harry Browne's

Permanent Portfolio at the high end (25%, but his framework assumes equal weighting of four uncorrelated regimes), Ray Dalio's All Weather at the low end (~7.5%), Rick Rule's insurance frame at 10%, and the "global market portfolio" argument at ~5%. This is the right answer for most US retail investors who want a diversified portfolio that performs acceptably across multiple regimes.

• 20%+ is the goldbug answer. It is internally consistent

only if you have a strong conviction view on monetary regime change. Most people who run this allocation have not honestly stress-tested what happens if they are wrong for a decade — see the 1980–1999 panel of the regime chart in §2.4.

The interactive at the end of this lesson lets you slice the historical record by inflation regime to see when the hedge actually paid off and when it did not.

2.11 Tax Treatment and Non-US Investor Notes

US investors. GLD, IAU, GLDM, SGOL, BAR, IAUM are all structured as grantor trusts holding a "collectible," which means **long-term capital gains are taxed at the 28% collectibles rate** in the US, not the standard 15–20% LTCG rate. GDX miners, by contrast, are ordinary stock and qualify for the standard rates plus qualified dividends.

There are three honest ways for a US investor to mitigate the 28% rate, each with caveats:

Hold the bullion ETF inside a tax-advantaged account (Roth

IRA, traditional IRA, 401(k)). This is by far the cleanest and most-recommended approach. Most retail brokers permit GLD, IAU, GLDM, PHYS inside an IRA.

**Use Sprott PHYS or Royal Canadian Mint MNT with a QEF

election.** As Canadian-domiciled funds, they qualify as Passive Foreign Investment Companies (PFICs) for US tax purposes. With a QEF election filed annually on Form 8621, gains are treated as long-term capital gains at 15–20% rather than collectibles at 28%. The election creates ongoing paperwork and requires the fund to provide annual statements (PHYS does; verify with each fund). Talk to a tax preparer familiar with PFIC rules before relying on this.

**Long-dated equity options on GLD instead of holding GLD

spot.** Because the option contract is a separate security, the holding-period rules apply to the option itself, not to the underlying trust units. Long-dated calls (LEAPS) can be used to express a directional view on GLD with capital-gains tax treatment on the option contract. However — and this is important — equity options on GLD are not Section 1256 broad-based-index options; they are equity options taxed under the standard rules with wash-sale, straddle, exercise, and assignment complications. Do not assume options "automatically solve" the 28% collectibles issue. They are a tool for the right situation, not a blanket workaround. Anyone running meaningful size in this structure should work with a tax preparer familiar with section 1234, section 1092 straddle rules, and the specific GLD trust mechanics.

There is also one common myth worth correcting: a few US states (notably Texas, Wyoming, Utah, Oklahoma, Louisiana, and several others) have passed legislation declaring gold and silver specie to be legal tender for state purposes, and exempt those metals from state sales tax. That legislation is real — Texas opened a state bullion depository in 2018 — but it operates only at the state-tax level. **Federal capital gains treatment is unchanged**, and the 28% collectibles rate still applies on realized gains regardless of which state you live in.

Non-US investors — the picture is very different.

• Hong Kong. No capital gains tax and no GST on physical gold

or gold ETFs. HK-domiciled investors can hold US-listed GLD directly with no withholding on (non-existent) dividends. The HKMA does not maintain monetary gold reserves, but Hong Kong is one of the world's deepest physical gold trading hubs (Chinese Gold & Silver Exchange Society, established 1910). Local bank-vaulted bullion via Hang Seng or Bank of China is widely available.

• Taiwan. No capital gains tax on securities. Bank gold

passbook accounts at Bank of Taiwan, Mega Bank, and Cathay United are the most common retail vehicle — they are unallocated claims on the bank, so the §2.9 caveats about custody form apply. For allocated, audited exposure, Taiwanese investors using international brokerage accounts can hold GLD, PHYS, or MNT.

• Mainland China. Cross-border gold ETF access is restricted

by SAFE rules; mainland investors typically use the Shanghai Gold Exchange (SGE) and bank gold accumulation accounts (招商 / 工行 / 建行 黃金積存). Physical gold sales are subject to a 13% VAT but investment gold for SGE-eligible products is VAT exempt. For mainland investors with Hong Kong or Singapore brokerage access, the same offshore toolkit is available.

• Estate-tax warning for non-US holders. US-listed ETFs

including GLD, IAU, GLDM are US situs assets for US estate tax purposes. Non-US-domiciled investors die holding more than $60,000 of US-listed securities run into the US estate tax regime — up to 40% on the excess. Canadian-domiciled funds (PHYS, MNT) and physical gold held outside the US sidestep this. For non-US investors with significant gold allocations, this is a real consideration.

---

3. Common Misconceptions

1. "Gold has intrinsic value." It doesn't. Neither does the dollar, the euro, sterling, or bitcoin. All five are belief-priced. The most concise statement of the point is John Maynard Keynes's 1924 line about gold being "a barbarous relic" — by which he did not mean gold was worthless, but that all monetary anchors are social conventions and treating any one of them as metaphysically "real" is a mistake. Five thousand years of belief makes gold's convention extremely durable; it does not make it intrinsic. The honest framing is that gold is a Schelling point in monetary coordination — humans across every civilisation independently selected the same metal. That is robust, and it is also still just consensus.

2. "Gold always rises with inflation." It rises with *poorly- anticipated, sustained* inflation, especially when accompanied by falling or negative real rates. In a Volcker-style "rates ahead of inflation" regime (1980–82, parts of 1994, briefly 2022 before the bond market broke), gold falls in nominal terms even while CPI is high — because the opportunity cost of holding a non-yielding asset has gone up faster than the inflation it is supposedly hedging. The right one-line framing: gold hedges *real-rate suppression*, not inflation per se. The two often coincide in crisis regimes, which is why the "gold = inflation hedge" shortcut is approximately right most of the time and exactly wrong in the windows that matter most.

3. "Gold is a bad investment because it has no yield." Gold is a bad cashflow asset — it is not capital that compounds. Treating it as a bond substitute is a category error. As a portfolio diversifier with negative real-rate sensitivity and positive crisis-regime correlation, its long-run Sharpe addition to a 60/40 has been positive across the full 1971–2026 record (see §2.4). Warren Buffett's 1998 Harvard speech mocking gold (you dig it up, ship it across the world, dig another hole, put it in, then "pay people to stand around guarding it") is the canonical version of this misconception. Buffett is one of the greatest equity capital allocators of all time and his point about cashflow superiority is correct. He is not, however, a portfolio constructor managing for multi-regime survival — Berkshire holds ~$300B in cash and Treasuries instead, which is its own form of non-yielding-relative-to-equity drag for a different reason.

4. "Gold miners are leveraged gold." They are levered to the gold price minus all-in sustaining costs (AISC) and carry operational risk, country/jurisdictional risk (many large miners operate in regime-risky jurisdictions like Russia, Mali, the DRC), hedging-program risk (miners that hedged forward sales in the late 1990s missed most of the 2001–2011 rally — Barrick Gold's hedge book is the textbook case), management capital-allocation risk, and equity-market beta that can dominate gold beta during broad sell-offs. The empirical record is striking: GDX has underperformed GLD over most multi-decade windows since GDX inception in 2006, despite the theoretical operating leverage. Picking individual miners (Newmont, Agnico Eagle, Franco-Nevada on the streaming side) has historically worked better than the GDX index, but that is stock-picking risk, not gold exposure.

5. "Bitcoin replaces gold." Bitcoin's belief is 16 years old (Satoshi's white paper, October 2008). Gold's is 5,000. They may both belong in a portfolio — many investors run both as small parallel sleeves — but the substitution claim ignores the durability gap. The relevant test is not "what has bitcoin done in the last decade" (during which it has obviously crushed gold). The relevant test is "what is the conditional-on-failure outcome, and how confident are we that the belief endures across the next multi-generational regime." On those questions, gold has the empirical record and bitcoin has only an asymptote. Treat them as two different positions, not two implementations of the same position.

6. "You should own a basket of commodities for diversification." Industrial commodities (oil, copper, base metals, agricultural softs) are pro-cyclical risk assets, not diversifiers. They correlate with global growth, the dollar, and equity beta in any period long enough to matter. Gold is the diversifier; oil and copper are factor bets on global growth. The DBC and PDBC diversified-commodity ETFs are not substitutes for gold.

7. "GLD costs the same as buying physical." GLD has a 0.40% annual fee, GLDM is 0.10%, IAU is 0.25%, and bid-ask is small (typically 1–3 bps for a retail order). Coins from a dealer carry a 3–6% premium plus storage and insurance (roughly 0.5%/ year for an insured commercial vault) plus a 1–3% bid-ask on resale. For paper exposure, GLD/IAU/GLDM win on cost over any horizon under ~10 years. For doomsday physical-delivery use cases, the comparison does not apply because the products are not interchangeable.

8. "The 1980 high was the all-time high." Not in real terms. The 1980 peak of $850 nominal is roughly $2,400 in 2026 dollars, which is roughly the same level as today. That observation matters because it reframes the "gold is at all-time highs" media narrative — in real terms, gold spent most of 2024 *just back to* its 1980 high after a 44-year wait. Always use real-terms charts when judging "is gold expensive."

9. "Central banks selling gold means gold is dead." Central banks were net sellers every year from 1990 through 2008, and during that exact window gold was weak in real terms. They have been net buyers every year since 2010 — a regime change discussed in §2.5 — and during that window gold has been strong. Central bank flow is a price-setting trade, not a sign of irrelevance. The Bank of England's "Brown's Bottom" sale (1999, 395 tonnes at $282/oz) is the textbook cautionary tale: the official sector sold at the bottom and bought back at the top, which is what official-sector flows tend to do during regime changes.

10. "Gold has too small a market to matter." The above-ground investible gold stock is approximately 213,000 tonnes worth roughly $20 trillion at April 2026 prices. That is meaningfully larger than the global high-yield bond market and roughly half the size of the US municipal bond market. Daily LBMA OTC gold trading volume is ~$160 billion, comparable to the daily trading volume of US 10-year Treasuries. Gold is one of the largest and most liquid asset classes on the planet — the "small market" claim usually conflates flow (annual mine supply ~3,500 tonnes, ~$300B/year) with stock (the existing 213,000 tonnes), which are two different things.

11. "Gold is a useless rock that pays no income." As §2.2 discusses, gold can be lent through the bullion-bank market for a small lease yield, and it is accepted as collateral by prime brokers and the CME. Under Basel III's 2021–22 NSFR implementation, allocated physical gold qualifies as a high- quality liquid asset for bank balance sheets. The "useless rock" framing is a half-century out of date.

12. "Gold and silver move together." They have positive long- run correlation but the relationship is unstable and silver is materially more volatile. Silver's industrial-demand share (~55%) makes it as much a tactical industrial-cycle bet as a monetary asset. The gold-silver ratio (§2.6) has ranged from 17:1 to 125:1 in the modern era — that is not "moving together."

---

4. Q&A Section

Q1: If gold has no cashflow, how can I justify owning it? A: The same way you justify holding US dollars in a savings account. You are not holding either for cashflow; you are holding them as a store of purchasing power and an option on tail-event demand. Gold has held purchasing power across regimes that flattened multiple national currencies (Weimar, Argentina, Zimbabwe, Venezuela, Lebanon — pick your decade). That is the value proposition. It is small relative to equity compounding, which is exactly why a 5–15% allocation is the institutional consensus for most investors.

Q2: Is bitcoin a better gold? A: Maybe in 50 years; not yet. Bitcoin's belief is 16 years old. Gold's is 5,000. Bitcoin has higher upside if the belief endures — but the conditional-on-failure outcome is materially worse than gold's, because gold has 5,000 years of cross-civilisational empirical evidence and bitcoin has only Satoshi's white paper and a series of price charts. Running both as small parallel sleeves is more honest than picking one. Many of the "gold versus bitcoin" debates conflate "which has done better in the last decade" (bitcoin, by a mile) with "which is more likely to still be money in 2075" (gold, by a similarly wide margin). The two questions have different answers.

Q3: What is the right benchmark for "did gold work"? A: Real interest rates, not nominal CPI. The cleanest one-line test: did real 10-year yields fall during the period? If yes, gold should have risen, and if it did, the real-rate hedge worked. Use FRED series DFII10 for a quick read. But — referring back to §2.3 — the 2022–2026 leg shows that the real-rate model is the primary driver, not the only driver. Central bank flow, geopolitical reserve diversification, and fiscal-deficit risk premia all matter at the margin.

**Q4: Why does GLD use a 28% collectibles tax rate, and is there a way around it?** A: The IRS classifies gold bullion as a collectible regardless of whether you hold a coin or a unit of a grantor trust that holds bullion on your behalf. Three honest workarounds, none of them automatic:

Hold the bullion ETF inside an IRA or other tax-advantaged

account. Cleanest and most-recommended.

Use Sprott PHYS or Royal Canadian Mint MNT with an annually-

filed QEF election (Form 8621). Real paperwork; only worth it for meaningful size.

Express the position via long-dated options on GLD instead of

spot GLD — but be aware these are equity options under standard rules (wash sales, straddles, exercise/assignment all apply), not Section 1256 broad-based-index options. Do not assume options "automatically solve" the 28% issue.

Note on the "states that don't tax gold" claim you may have seen: about a dozen US states (Texas, Wyoming, Utah, Oklahoma, Louisiana and others) have passed legislation declaring gold and silver legal tender for state purposes and exempt from state sales tax — Texas opened a state bullion depository in 2018. That legislation is real but operates only at the state-tax level. Federal capital gains treatment is unchanged, and the 28% collectibles rate still applies regardless of which state you live in. Anyone telling you "gold is tax-free in Texas" is mixing up state sales tax (true) with federal capital gains tax (unchanged).

Q5: Should I buy physical coins or a bullion ETF? A: For investment purposes, GLDM, IAU, or PHYS. For doomsday preparedness or estate planning across multiple jurisdictions, physical coins with off-site storage. The two are different products. The premium on coins (3–6%) plus storage (~0.5%/year for an insured commercial vault) plus the bid-ask on resale wipes out the structural cost advantage of physical over many years.

Q6: What about gold mining stocks? A: GDX is a leveraged-but-noisy gold proxy. It has higher beta to gold but also picks up operational risk, country risk, hedging- program risk, and management risk. Over the long run, GDX has underperformed GLD despite higher operating leverage on paper — see misconception #4 above. If you want gold exposure, own gold; if you want a tactical trade on gold moving sharply higher, GDX is a higher-torque short-duration position, not a long-term holding.

Q7: How do I use options on GLD for tax efficiency? A: Three common structures: (1) covered calls on existing GLD to generate income with a price ceiling — covered in detail in Week 27: Covered Calls; (2) cash-secured puts to accumulate GLD at lower prices — Week 28; (3) deep-ITM long-dated GLD calls (a "stock-replacement" structure) to hold gold exposure with less capital deployed — Week 38: LEAPS Replacement. The tax angle is not that options magically dodge the 28% collectibles rate — exercise into the underlying inherits the trust's tax treatment. The angle is that the *option contract itself* is taxed under standard equity-option rules with its own holding period, so position management (rolling, closing, restriking) creates taxable events at standard capital-gains rates rather than at 28%. The wash-sale, straddle, and constructive-sale rules all apply. Anyone running meaningful size in this structure should work with a tax preparer familiar with section 1234, section 1092, and the GLD-specific mechanics.

Q8: When should I trim my gold sleeve? A: When real 10-year yields are above +2% and rising, when central bank net buying slows materially below the 2022–2024 1,000-tonnes-per-year pace, and when the gold-to-S&P 500 ratio has compressed below its 20-year median. None of those conditions hold in April 2026. The "permanent allocation" is permanent; the size is what you adjust at the margin in extreme regime windows.

Q9: Where does this lesson connect to the rest of the course? A: Week 4 introduced the 60/40 framework, and Week 5 covered the bond leg in depth. This lesson adds gold as a third asset that diversifies both stocks and bonds in regimes where bond/equity correlation breaks down (2022 being the textbook example). Week 11 covers rebalancing as part of the behavioural discipline lesson — it is the mechanic that turns the gold/stock/bond mix into the realized diversification benefit shown in §2.4. Week 15 puts the multi-asset book together, and Week 27 (Covered Calls), Week 28 (Cash-Secured Puts), and Week 38 (LEAPS Replacement) cover the option overlays that the §2.11 tax discussion alludes to. Week 39 covers commodity futures, contango, and the USO case study in detail.