Does gold really preserve wealth? 50 years of data, three calculations, one uncomfortable conclusion

Almost everyone who sells gold says "gold preserves wealth". Is that true? Yes — but it omits the three most important words: depends on your entry. The same metal bought in 1971, in 1980 or in 2015 delivers completely different life experiences fifty years later: one is 14x, one goes nowhere, one is 4x.

This article deals in arithmetic, not mystique: using 55 years of real gold prices, real US CPI and real exchange rates, it takes "store of value" apart and runs the numbers three different ways. By the end you will know two things — what gold actually preserved, and under what conditions it manages to preserve anything.

1. First, be clear: "store of value" means three different things

The argument never ends because people are not talking about the same thing.

• Nominal preservation: the number neither grows nor shrinks. Gold cannot do this; it moves every day.
• Purchasing-power preservation: fifty years from now it buys the same basket it buys today. This is the definition gold comes closest to satisfying.
• Relative preservation: beating stocks, beating property, beating inflation. On this measure gold loses most of the time — just not by an embarrassing margin.

So the right question is not "does gold preserve wealth" but "from what entry price, over what holding period, against what benchmark". The next three sections give three sets of answers.

2. Fifty-five years of real prices: four phases, two fates

In August 1971 Nixon closed the gold window and the price began floating freely from its fixed 35 USD/oz. Measured from that year to today, gold has passed through four clear phases:

Prices are representative quotes for each phase, rounded to whole dollars; annualised figures are compound rates over the actual number of years.

Put the four phases side by side and the pattern is clear: gold does not rise steadily; it drifts sideways for years and then spikes. From 1980 to 1999 — nineteen years — it fell 70%, with not one year worth celebrating. From 2001 to 2011 it rose sixfold, and almost everyone felt like a genius. One metal, two faces.

3. Three calculations, three conclusions: entry matters more than the asset

This is the core of the article. Same metal, same today, only the entry changes — and the annualised return differs by a factor of thirty.

To read the table, look only at the last two lines:

• Starting from 35 USD in 1971, gold beat inflation, with a real annualised return of 4.98%. That is respectable — close to long-term government bonds, but far more volatile.
• Starting from 850 USD in 1980, the real annualised return after 46 years is just 0.29%. In other words, someone who bought at the panic peak in 1980 barely preserved purchasing power across two generations — and did not even match a savings account.

4. Inflation adjustment: what 850 USD in 1980 is worth today

Many people say "gold was 850 in 1980 and 4,131 now, nearly 5x" — which sounds great. But a 1980 dollar is not a 2026 dollar. At roughly 4.24x cumulative US CPI:

In other words: someone who bought gold in January 1980 has, by October 2026, gained 15% in purchasing power. The average coupon on 10-year US Treasuries over the same period was far higher.

Take the 2011 high of 1,920. Adjusted by CPI that is roughly 2,817 in today's money, and the current 4,131.76 is 1.47x that — about 2.6% real annualised over fifteen years. Not a loss, but hardly the "store of value" legend either.

5. The local-currency view: the numbers that matter more to overseas Chinese

If you ultimately spend in local currency, the USD gold gain is only half the story. Add the exchange rate:

From 2000 to today the CNY gold price went from 72.4 to 889.18 per gram — 10.13% annualised over 26 years. Note the counter-intuitive part: over that period the CNY appreciated against the dollar (8.28 → 6.69), and appreciation depresses the CNY gold price. Without that FX headwind, the CNY gain would have been even larger.

Run it the other way: buy in China near 400 per gram in 2011, and by 2015 the price was 210 per gram — a 47% loss, with break-even only in 2019, seven years later. That timing is almost identical to the USD series, which shows that the exchange rate changes the size of the move, not its direction. The full conversion formula is in USD gold price vs CNY gold price.

6. Three conditions that must all hold for "store of value" to work

Compress all the arithmetic above into an actionable test and only three rules remain:

Condition 1: hold through a full cycle (15 years or more)

Gold's cycles run a decade or longer. 1980–1999 was a nineteen-year bear market, 2001–2011 a ten-year bull, 2011–2015 a four-year bear, 2016 to today a ten-year bull. Any holding period shorter than a decade is a directional bet, not wealth preservation. Pull any start date through the historical gold price lookup and you will feel what "cycle" means.

Condition 2: do not buy at an emotional peak

How do you spot a peak? Ignore headlines and watch three numbers: whether real rates are deeply negative, whether price is more than 20% above its 200-day moving average, and whether the local premium is abnormally wide. At 4,131.76 today, gold is 6.15% below the September high of 4,402.73 — high-level consolidation rather than a panic top, but not cheap either.

Condition 3: hold it through a low-cost instrument

This is the most overlooked condition and often the decisive one. Look at the arithmetic below.

7. The maths: what ten years of holding 1 kg of gold costs

At 889.18 per gram today, 1 kg of gold is worth 889,179. Across three holding methods, the ten-year friction cost differs by more than 4x:

ETF estimated at a 0.25% annual fee on unchanged opening value; physical spread taken at 6%, the low end of the common industry range.

The 45,000 difference is 5% of principal. If gold compounds at 5% a year for ten years, total return is about 55% — and cost eats close to a tenth of it. Half of "preserving wealth" is the gold price; the other half is not overpaying in fees.

As for leveraged spot gold: one lot is 100 ounces, notional 413,176 USD, and at 1:100 leverage the margin is about 4,132 USD — an adverse move of just 1% wipes out the margin. Swap at 18 USD per lot per night is 540 USD a month, close to 13% of capital. This is a trading instrument, not a preservation instrument — put the swap into the P&L calculator and you will see the real cost of holding a leveraged position long term.

8. So should you hold gold, and how much

Treat gold as insurance, not an engine of return, and everything falls into place. Insurance is priced to cost you a little in normal times (opportunity cost) and save you in extremes (a hedge against currency and credit risk).

• 5%–10%: suits most people who treat gold as insurance. Build the base with an ETF or a bank accumulation plan — low cost, good liquidity.
• 10%–15%: for those with a clear concern about inflation or currency debasement who can also tolerate a drawdown of 30% or more.
• Above 20%: needs a very strong reason. Historically, the cost of a permanently heavy single-asset position only shows up at the end of the learning curve.

On execution, lump sum versus instalments matters less than people think, but it matters a lot psychologically. Fixed monthly buying (dollar-cost averaging) removes single-point risk such as "buying the 1980 top"; for the mechanics see running a gold accumulation plan. If you are still torn between an ETF and physical metal, this comparison costs out both tax and liquidity.

9. Frequently asked

Q: Does gold really hedge inflation over the long run?
A: Depends on the entry. From 35 USD in 1971 to 4,131 in 2026, the nominal annualised return is 9.06%; adjusted by US CPI the real rate is about 4.98%, so yes, it beat inflation. But start from the January 1980 high of 850 — equivalent to about 3,605 today — and the current price is only 15% higher, a real annualised 0.29% over 46 years, worse than a term deposit. The accurate statement is: gold preserves purchasing power over long cycles, but the price you pay decides whether you get to enjoy that.

Q: Why did someone who bought gold in 1980 wait 28 years to break even?
A: Because 850 in 1980 was panic pricing, not cost pricing. US inflation was above 14%, real rates were deeply negative, money flooded into hard assets and pushed gold two to three times above its long-run average. Volcker then raised rates to 20%, real rates turned positive, and gold fell for 19 years to 252 — a 70% drop. The nominal price did not regain 850 until 2008, and purchasing power took even longer. That is not gold failing; it is mean reversion.

Q: Is gold's return in local currency the same as in dollars?
A: No, and the gap is large. In 2000 gold was about 272 USD/oz with USD/CNY at 8.28, or 72.4 per gram; in October 2026 the price is 4,131.755 with the rate at 6.6937, or 889.18 per gram. That is about 10.13% annualised in CNY over 26 years, below the 11.03% in USD — because the CNY appreciated against the dollar over the period, and the exchange rate absorbed part of the gain. The formula is in USD gold price vs CNY gold price. For overseas Chinese, your return depends on the currency you eventually spend.

Q: If I buy physical bars, how much of the return do costs eat?
A: Take a 1 kg bar at the current 889,179: a 6% bid/ask is about 53,000, plus storage and insurance at roughly 0.15% a year, about 13,000 over ten years, for a total holding cost near 67,000 — 7.5% of principal. The same 1 kg held for ten years in a gold ETF at a 0.25% fee costs about 22,000. The 45,000 gap is the premium you pay for metal you can see. The itemised breakdown is in ETF or physical bars.

Q: How much gold should an ordinary investor hold?
A: There is no single answer, but here are reference ranges: people who treat gold as insurance rather than a return engine typically hold 5%–10% of total assets; those with a clear inflation or currency concern who can tolerate a 30%+ drawdown can go to 15%. A single asset above 20% needs an exceptionally strong case in any historical period. What matters is not the percentage but whether the money can stay invested for a full cycle — gold has fallen for 19 consecutive years before.

Q: Is 4,131 a high level?
A: On this article's snapshot (2026-10-20 22:40 UTC+8, 4,131.755): 6.15% below the 21 September high of 4,402.73 and 1.42% above the 7 October low of 4,073.93 — consolidation within a historically high zone. Whether it is "expensive" should be judged against real rates, not against past prices: price keeps making new highs, and real rates are the valuation anchor. The method is in the five indicators to watch.

Q: Does gold jewellery count as preserving wealth?
A: It is consumption, not investment. Jewellery carries a fabrication fee of 30–80 per gram and a brand premium of 5%–15%, both of which go to zero on resale; buy-back is at a discount to the metal price. Round trip, the cost is 15%–25%. Buy it because you like it, not as portfolio allocation — details in is gold jewellery an investment.

Q: Central banks keep buying gold — doesn't that mean the price cannot fall?
A: Central bank buying changes the long-term demand structure, not the short-term price. Central banks were buying in 2013 too, and gold still fell 28%. Treating official buying as a "floor guarantee" is a classic narrative trap; the mechanism is explained in central bank buying and de-dollarisation.