On 1/28/2026 gold touched $5,589/oz - an all-time record, doubling in just 18 months. The People's Bank of China (PBoC) bought for 17 straight months, taking reserves to 2,313 tonnes. Chinese gold ETFs pulled in $15.5B in 2025, then another $8.5B in just Q1/2026. This piece unpacks four layers: the global M2 expansion, China's real-estate dam breaking and 1.4 billion people hunting for a loophole, the Shanghai-COMEX carry-trade mechanism when the price gap can't close, and the weaponization of the USD pushing every non-allied central bank back toward metal.
Disclaimer: This article compiles public data from the World Gold Council (WGC), IMF, FRED, BIS, PBoC, LBMA, Shanghai Gold Exchange (SGE), Reuters, Bloomberg, FT, CNBC, Goldman Sachs, J.P. Morgan, Morgan Stanley, State Street, and Brad Setser/CFR. Correlation data and most historical figures come from the interactive "M2 vs Gold" widget from tapchiphowall.com embedded in this article.
On the latest figures: the 2026 gold price averaged YTD through March = $4,890/oz; the spot price on 4/24/2026 sat around $4,400-4,700 after correcting from the ATH on 1/28/2026. "Global M2" in this article = US + Eurozone + China + Japan (~$99T in 2026) plus the UK and several large emerging economies (India, Brazil, Korea, Canada…) converted to USD; the 2026 total is estimated at ~$118 trillion.
Scale - Gold Is Back At The Center
The Pearson r correlation between global M2 and the gold price over 2000–2026 is +0.96+ on nominal-level data (see the warning below - this figure largely reflects both series trending upward together over the decades). As paper money expands, gold - whose supply grows only ~1.5%/year through mining - must rise in price simply because the denominator is expanding. The two shocks of 2022 (the US freezing Russia's USD reserves) and 2020-26 (China's real estate collapse) mean this time isn't just an M2 story: gold is shifting from a "hedge asset" into "reserve structure."
According to the World Gold Council, in 2025 central banks bought 863 tonnes officially - the fourth straight year above 800 tonnes. Some alternative estimates (including "unreported" purchases) push the figure up to ~1,237 tonnes: a third straight year above 1,000 tonnes. WGC notes that 57% of total purchases go unreported - meaning more than half of the gold moving from Western banks into Asian vaults happens with no press release at all.
In January 2026, global gold ETFs took in a record month; for all of 2025, total assets doubled to $559 billion - the highest level in history. Note: within that doubling, net new money was only ~$89B (~32%); the rest (~$190B) came from the rising gold price. In other words, the 2025 ETF surge was both new money and a repricing effect. Total gold ETF holdings hit 4,025 tonnes - surpassing the 2020 COVID-era peak. And this is only the "paper" side. The physical side is flowing from London-New York to Shanghai-Hong Kong much faster.
Part 1 - Global M2, Or "Gold Stood Still, New Money Is What Got Cheaper"
The long-run correlation between global M2 money supply and the gold price is the highest of any macro factor: Pearson r ≈ +0.96 over 2000–2026 on nominal-level data (see the widget below). This doesn't say that M2 causes gold to rise, only that the two variables move almost in lockstep - so closely that you can't tell them apart just by eye.
When calculating correlation between two series that both trend upward over time (M2 and cumulative gold price), r at the nominal level is naturally close to 1 - this is a spurious correlation between two trending series, a classic trap in time-series analysis. Proof: switch to % YoY data in the widget below, and r drops to ~0.3-0.5 (still positive, but not "in lockstep").
The correct interpretation: in terms of level, M2 and gold both swell decade over decade - meaning gold holds its real value over time as more money is printed. But in terms of year-over-year rate of change, M2 is just one of many factors, and interest rates / risk premia / geopolitical shocks can matter more within any given short cycle.
What is M2? When does M2 rise?
M2 is a measure of the total money circulating in an economy, including: cash, demand deposits (checking accounts), and short-term time deposits (savings under 2 years). Simply put: M2 measures "how much money exists" within a country's financial system.
Global M2 is the sum of M2 across all major economies - the US, EU, China, Japan, etc. - converted to USD.
When does M2 rise?
- Central banks print money (quantitative easing - QE): pumping money in by buying bonds, injecting liquidity into the system.
- Commercial banks lend more: every new loan effectively creates new money within the system.
- Government runs a large budget deficit: spending more than it collects, with the shortfall usually financed by issuing bonds - which the central bank often buys back.
From 2000 to 2026, M2 across the four major blocs (US + Eurozone + China + Japan, converted to USD) rose from ~$20 trillion to ~$118 trillion - nearly a 6x increase. Gold, which can't be "printed more of," became more expensive measured in that ever-growing pile of currency.
Every country calculates M2 differently - can that be gamed?
There is no "global M2 standard." The IMF deliberately does not impose a single M2 definition - each central bank decides its own components based on its domestic financial structure. The result: everyone calls it "M2" but they're measuring different things.
| Country / Bloc | What does M2 include? | Notable differences |
|---|---|---|
| US (Fed) | M1 (cash + checking accounts) + savings deposits + small CDs (<$100K) + retail money market funds | Excludes large CDs ≥$100K and institutional money funds. The $100K threshold has been unchanged since 1980, not adjusted even as deposit insurance rose to $250K (2008). |
| Eurozone (ECB) | M1 (cash + overnight deposits) + time deposits ≤ 2 years + deposits redeemable at notice ≤ 3 months | Collected from all monetary financial institutions (MFIs) across the 20 eurozone countries - relatively transparent thanks to a census-based system. |
| Japan (BoJ) | Cash + all deposits (demand + time + CDs) at licensed domestic banks | Does not segment by amount threshold like the US. Includes CDs - the US only counts small CDs. |
| China (PBoC) | M1 + personal savings deposits + corporate time deposits + foreign currency deposits + entrusted deposits. Since 2018: expanded to include principal-guaranteed WMPs (wealth management products) and money market funds. | The broadest of the four blocs. Even after the 2018 expansion, a large share of non-principal-guaranteed WMPs (~¥24 trillion as of 2017) still sits outside M2. |
So can China "print money but report a lower M2"?
Short answer: not easy to hide, but a direct comparison is misleading.
- China's M2 may actually be understated relative to reality - the opposite of the usual worry. China's shadow banking system, at its peak of ~¥100 trillion (2017), generated enormous credit through WMPs, entrusted loans, and interbank lending, most of which sat outside M2 before 2018. The PBoC expanded its M2 definition to "capture" this portion, but it's still not complete.
- GDP can be faked - M2 is much harder. With GDP, a government only needs to adjust a few input statistics (investment, public spending) to produce the desired number - a World Bank study (2018) compared satellite nighttime light data against reported GDP and found authoritarian regimes inflating GDP by 15–30%. M2 is different: it's calculated from the consolidated balance sheet of the entire banking system - every citizen's deposit is a liability line on a commercial bank's books. Faking M2 would require falsifying the books of thousands of banks simultaneously, and these figures cross-check against the BIS (Bank for International Settlements), cross-border capital flows, and CIPS/SWIFT data.
- But China has started hiding other data. From 2023–2025, Beijing stopped publishing hundreds of economic indicators it once disclosed openly (youth unemployment, detailed FDI, some export data). M2 is still published regularly every month, but the overall decline in transparency has forced analysts to use proxies (electricity consumption, rail freight, satellite imagery) to verify it.
- The real risk: not fraud, but different definitions. When we say "global M2 = $118T" by adding US M2 + Eurozone + China + Japan, we're adding four different yardsticks together - like adding kilograms to pounds and calling it "total weight." The number is meaningful in terms of trend (all of them are swelling), but the absolute value should be read with a margin of error of ±10–15%.
What is Pearson r?
Pearson r (the Pearson correlation coefficient) measures how much two variables move together or in opposite directions, on a scale from −1 to +1.
- r = +1: perfectly in the same direction - when A rises, B rises exactly in step.
- r = 0: no linear relationship.
- r = −1: perfectly in opposite directions - when A rises, B falls exactly in step.
r ≈ +0.96 means global M2 and the gold price have moved almost in lockstep for 26 years - every time money supply swells, the gold price ticks up with it. But high correlation isn't the same as causation: both could be driven by a third factor (for example: economic crisis → central bank prints money → investors buy gold), rather than M2 directly "pushing" the gold price.
The reason is simple: gold is priced in units of currency. When the denominator (total money in circulation) swells faster than the numerator (total gold in the ground), the price must rise. And the denominator has swelled very fast:
| Metric | 2000 | 2010 | 2020 | 2024 | 2026 (YTD) | × |
|---|---|---|---|---|---|---|
| US M2 | $4.9T | $8.8T | $19.1T | $21.5T | $22.7T | 4.6× |
| Eurozone M2 (USD) | $4.4T | $11.3T | $15.5T | $16.2T | $17.0T | 3.9× |
| China M2 (USD) | $1.8T | $10.5T | $31.5T | $43.0T | $50.0T | 27.8× |
| Japan M2 (USD) | $6.0T | $7.9T | $10.5T | $9.0T | $9.5T | 1.6× |
| Global M2 (4 blocs + UK/EM) | $20.0T | $53.0T | $96.0T | $104.0T | $118.0T | 5.9× |
| Gold ($/oz, average) | $279 | $1,225 | $1,770 | $2,386 | $4,890 | 17.5× |
Gold has risen faster than M2 for two reasons: (1) total gold in the ground grows only ~1.5%/year through mining - while global M2 grew ~7%/year over 2000–2024; (2) confidence in paper money has eroded with each shock (2008, 2020, 2022, 2024), repricing gold's "safe haven" premium. The ratio of gold market cap / global M2 rose from ~7% (2000) to nearly ~25% by end of Q1/2026 - and if this trend continues, some analysts think the ratio could return to the ~40% level seen in the 1980s (the stagflation peak).
China's M2 has grown 27.8× since 2000 - six times the growth of US M2. This isn't just "more money": it's 1.4 billion people (~494 million households) accumulating monetary assets at a pace unprecedented in economic history. When real estate - their primary M2-absorbing channel for two decades - broke, the pressure to find an alternative channel pushed gold into the role of the "channel a government can't print more of" (see Part 2).
Interactive Appendix - M2, Interest Rates, And Gold (2000–2026)
Interactive widget: correlation heatmap, central bank policy-rate chart (Fed/ECB/BoJ/PBoC), M2 and gold market cap, breakdowns by cycle (Dot-Com → Rate Hikes → 2008 Crisis → Normalization → COVID → Iran/Rate Cuts), and a raw data table. You can toggle between "Nominal Level" / "% YoY" and "Nominal Rate" / "Real Rate."
Part 2 - China's Real Estate Dam Broke, Gold Is The Main Loophole
This is the part that makes this rally different from every prior gold bull run in history. See the article "China: 33 Months Of Falling Home Prices": 1.4 billion people just watched their primary asset channel (~70% of household wealth) fall for 33 straight months. Meanwhile SAFE (China's foreign exchange regulator) caps $50,000 per person per year for converting to USD abroad, and any foreign-exchange transaction over $5,000 must be reported. Pressure builds up - with no official release valve.
Figures from the World Gold Council/Ray Jia:
- 2024: Chinese gold ETF flows ~RMB 30B (~$4.2B)
- H1 2025: Chinese gold ETFs +RMB 64B (~$8.8B), a first-half record at the time
- April 2025: a single month pulled in $6.8B net (65 tonnes) - the largest monthly record ever
- Full-year 2025: RMB 112B ($15.5B, 133 tonnes) - an annual record
- Q1 2026: Chinese gold ETFs +$8.5B in just 3 months - equal to 55% of all of 2025
- January 2026: RMB 44B ($6.2B, 38t) in a single month - the strongest start to a year on record
- March 2026 wholesale: 134 tonnes (physical gold flowing into the domestic market), +57% MoM, +12% YoY
On the central bank supply side: PBoC gold reserves rose from 1,842t (end of 2016) to 2,313t (3/2026) - a 26% increase in barely a decade. March 2026 marks the 17th consecutive month of buying - this streak follows an 18-month run that ended 5/2024 (the PBoC paused for 6 months, then resumed buying from 11/2024). Cumulatively across 2022-2026, the PBoC has stayed out of the market for only about 6 months total. And the unreported volume? According to WGC, 57% of total central bank purchases in 2025 came with no press release - and China is suspect number one (alongside Saudi Arabia, the UAE, and Kazakhstan).
WGC officially recorded 863t of central bank buying in 2025, but the total flow of metal moving into central bank vaults (per refinery and storage data from Metals Focus, GFMS) was ~1,237t. The gap of ~374t is roughly equivalent to the reserves of Austria, Spain, or Portugal. Most of it falls under "Asia-Middle East." In other words: the PBoC may have bought an additional ~100-200t that has yet to be disclosed.
Part 3 - The Mechanism: SGE Is Physical, COMEX Is Paper
To understand why "every time Shanghai's session opens, gold jumps," you need to understand how the world's two largest gold exchanges operate differently.
A physical exchange. Established in 2002 under PBoC supervision. Every transaction must settle with delivery of real metal. Trading hours: 9:00–11:30, 13:30–15:30 plus a night session 20:00–02:00 Beijing time.
Supply is tightly controlled: only certain banks are licensed by the PBoC to import gold into China. When domestic buying exceeds the permitted import volume → the SGE gold price runs above the world price. This gap cannot be arbitraged away simply by shipping gold from London to Shanghai for a quick profit - because importing gold into China requires a PBoC license.
A derivatives exchange. Mostly futures and options contracts. Fewer than 1% of contracts settle with actual delivery. The rest are rolled over or cash-settled.
This is where the global gold price gets set during US hours. The share of COMEX trading volume happening during Asian hours keeps rising - the US reacting to whatever just happened in Shanghai.
"More than 36 tonnes of gold bars were registered for delivery at Shanghai Futures Exchange (SHFE) warehouses, doubling within a single month - an all-time record. Traders and banks bought cheap gold on the spot market and delivered it into SHFE vaults to offset short futures positions, locking in profit from the gap between the futures price and the physical price."
John Reade (WGC): "So many people were piling into futures that prices shot up above physical gold. That created an opportunity for others to step in and deliver gold into the system."
The Mining.com article above is about futures vs internal China spot arbitrage (the SHFE futures price runs richer than domestic spot due to speculation). This is different from the Shanghai Premium (China's gold price runs above the international London/COMEX price because the PBoC restricts import quotas). Both phenomena coexist and reflect China's "gold fever" from different angles.
When the SGE price gap is wide enough (historically $10–120/oz, once peaking at $120), a trader chasing that ideal arbitrage would in theory:
- Buy gold on COMEX or in London (where the price is low)
- Sell gold on the SGE (where the price is high - the Shanghai Premium)
- Ship physical gold from the West (US/Europe) to China to deliver and lock in the profit
In practice, step (3) gets blocked because: (a) the PBoC tightly controls import quotas for gold entering China - only certain commercial banks hold quotas, and quotas get tightened whenever the PBoC wants to protect the CNY exchange rate; (b) shipping + insurance + tax + assay costs; (c) SGE gold and COMEX/LBMA gold aren't always the same specification (SGE standard bars are 99.99% purity, COMEX accepts 99.5% 100oz bars). Result: foreign traders can't simply pump more gold into China to capture the spread, which keeps the Shanghai Premium persistently positive, sometimes for months - rather than self-closing the way theory says it should.
And this is exactly where the "when Shanghai's session opens" mechanism kicks in: when the SGE's 9:00am Beijing session (= 9:00pm ET the prior day, or 8:00 GMT) sets its opening price, if that closing print exceeds the prior day's LBMA London close, COMEX's overnight session has to chase it. That's why gold traders now watch the SGE morning fix more closely than the US CPI print - especially since Q4/2025, when the SGE price gap has stayed persistently wide.
The reverse case: in H2 2025, Shanghai gold was cheaper than London gold
To keep the picture from being too one-sided: from September through December 2025, the Shanghai gold price was actually lower than London's - instead of higher as usual. To be clear: Chinese buyers were still buying gold very heavily throughout 2025, but mostly through ETFs and investment bars. The jewelry segment alone was weak - gold prices had gotten so high that consumers were reluctant to buy jewelry. Meanwhile, Western investors poured money heavily into gold ETFs, pushing the London price higher. In September 2025 the gap briefly fell to −$37/oz - an unusually low level. By January 2026, the balance flipped back as Lunar New Year gold-buying demand pulled the Shanghai price back above London's. The gap between the Shanghai and London prices reflects which type of demand is in the driver's seat: jewelry and physical gold in Asia, or financial investment in the West.
Layer 4 - Weaponizing The USD: When Currency Becomes A Punitive Weapon
In February 2022, the US and EU froze ~$300 billion of Russia's foreign exchange reserves held at Western financial institutions. That was a clear signal to every non-allied central bank: USD reserves held at Western institutions can be seized at any time, without going through a court.
The reaction: from 2022 to 2026, central banks in emerging economies bought gold at an unprecedented pace. Four straight years above 800 tonnes (per official WGC figures) - 2-3 times the historical average (~400t/year over 2010-2018).
| Year | Official CB buying (WGC) | Estimated true total (Metals Focus) | Biggest buyers |
|---|---|---|---|
| 2018 | 657t | ~657t | Russia, China, Kazakhstan |
| 2019 | 605t | ~605t | Russia, China |
| 2020 | 255t | ~255t | Turkey, UAE |
| 2021 | 450t | ~450t | Thailand, India |
| 2022 | 1,082t | ~1,136t | Turkey, China, Singapore, Egypt |
| 2023 | 1,037t | ~1,090t | China, Poland, Singapore |
| 2024 | 1,089t | ~1,150t | Poland, Turkey, India, China |
| 2025 | 863t | ~1,237t | Poland (102t), Kazakhstan, China (27t disclosed) |
The February 2022 event was a historic turning point: before it, central banks bought gold for traditional reasons (diversification, hedging against USD risk). After it, they bought for strategic reasons - not wanting to become the next Russia.
"When the US-EU froze Russia's assets in 2022, China accelerated its drawdown of US Treasury holdings - from ~$1.3T (the 2013 peak) to under $1T (4/2022) and then to ~$759B by end of 2024. In parallel, PBoC gold rose from ~1,948t (end of 2021) to ~2,010t (2022), 2,235t (2023), and then 2,312t (3/2026). This is no coincidence: the PBoC is restructuring its reserves into a form that can't be seized."
BRICS and the "Unit" - the next step, if it actually happens
On 10/31/2025, BRICS+ piloted Unit settlement: an intra-bloc settlement unit structured as 40% gold + 60% a basket of member currencies, adjusted daily. It's still only at the pilot stage, not yet an official operating system. But it's the first signal of metal being written back into cross-national monetary architecture since Bretton Woods in 1971.
Results from the WGC's 2025 central bank survey:
- 95% expect global gold reserves to rise over the next 12 months (a record since the survey began in 2018)
- 43% of central banks plan to increase their own gold holdings (vs 29% in 2024)
- 0% plan to decrease
- 62% rank "risk of USD weaponization" among their top 5 reasons for buying gold
The trend of central banks selling USD to buy gold is real, but there are two self-braking mechanisms preventing it from turning into a simultaneous collapse:
1. USD-denominated debt is a "natural short" against abandoning the USD. As of 2025, total USD-denominated debt held outside the US is ~$13 trillion (BIS Quarterly Review). Every dollar of USD debt is effectively a hidden short USD position: the borrower needs to buy back USD when the debt comes due. If the USD weakens, they benefit - but if everyone sells USD reserves at once and dollar liquidity dries up, the USD will spike (a short squeeze), and it's precisely those countries in USD debt who get crushed by the higher cost of repaying it. Central banks understand this perfectly well: they can't abandon the USD entirely because they still need it to service their own debt obligations and their domestic banking system.
2. A prisoner's dilemma - dumping together is collective suicide. Most emerging-market central banks still hold 50-60% of reserves in USD. If everyone sold USD at once, the value of their own remaining USD reserves would evaporate - meaning the faster they sell, the more they destroy their own balance sheet. This is a classic game theory problem: every country wants to reduce USD exposure gradually while hoping others reduce theirs more slowly. The Nash equilibrium: everyone reduces gradually, but no one dares to cut abruptly. The USD loses its role as the absolute reserve currency - but through slow erosion, not collapse.
The only scenario where the USD truly collapses? Everyone defaults simultaneously - meaning the $13 trillion of USD debt held outside the US gets wiped out all at once, destroying the short-squeeze mechanism described above. That is almost only conceivable in a scenario of major war or the collapse of the international order. The nearest precedent: the British pound losing its role as the global reserve currency - but even that took ~50 years (1914-1960s), not a single event. After WWII, Britain owed its colonies about £3.5 billion (sterling balances); Britain wanted to write off that debt but in practice only froze it, then let inflation erode its real value - meaning even as the empire dissolved, the debt wasn't defaulted on, it was diluted slowly. Bretton Woods in 1944 formalized the USD's role, but the Sterling Area persisted until 1972. In other words: even history's most famous "collapse example" was slow erosion, not an overnight rupture. Absent an event on the scale of a world war forcing everyone to default simultaneously, $13T of USD debt is the chain binding everyone to the dollar system whether they like it or not.
The paradox: the more the US runs deficits, the stronger the chain gets. Every time the US runs a budget or trade deficit, USD flows outward → the cost of borrowing USD abroad falls → countries and companies find it easier to borrow in USD → total global USD debt rises → structural demand for USD grows even larger → the US, in turn, becomes even more comfortable running further deficits. This is a self-reinforcing loop that Valéry Giscard d'Estaing (France's finance minister in the 1960s) called the "exorbitant privilege": the more "irresponsible" the US becomes, the more tightly the world gets bound to its currency.
(An intuitive comparison: the JPY operates on a similar mechanism at a smaller scale. Japan is the world's largest creditor nation - net foreign assets of ~$3.3T. When a crisis hits, Japanese investors repatriate capital, companies need JPY to repay domestic debt → the JPY strengthens amid the storm - right when every other currency is falling. The USD works on the same logic but at global scale: every crisis triggers a USD short squeeze because everyone needs dollars to repay debt at the same time - the 2008 GFC, COVID 3/2020, both times the DXY spiked right in the middle of the collapse. This is exactly why de-dollarization during a crisis is nearly impossible - that's precisely when everyone needs USD the most.)
In other words: gold benefits from the de-dollarization trend, but don't expect the USD to evaporate one day with gold shooting to $20,000. This process takes decades - and the structure of global debt itself is the system's "shock absorber."
Layer 5 - Real Rates And The Q2/2026 Correction
Even though the long-term story is very positive (M2, China, USD weaponization), gold doesn't go straight up. Over a span of 3 months - from the ATH of $5,589 on 1/28/2026 to $4,503 by end of Q1/2026 - gold fell −19.4%. This is one of the sharpest corrections in the modern gold bull run.
Middle East tensions escalated, with Brent crude above $90/barrel. US inflation expectations rose back up - markets repriced the expected number of Fed rate cuts from 4 down to 1-2 for 2026.
The 10-year TIPS real yield (expected, long-dated) reached ~1.94% in early April 2026 - the highest since 2008. When real rates are high and positive, the opportunity cost of holding gold (which pays no yield) rises → money flows out of ETFs.
Global gold ETF holdings fell ~63 tonnes in March 2026 - the largest outflow since 2024. Profit-taking after clearing $5,000.
After four straight years of net buying above 800t, IMF/WGC data for Jan-Feb 2026 shows the pace of buying slowing (the PBoC added only +5t in March, the smallest since Q3/2024). There is a hypothesis that some Middle Eastern central banks reduced or paused buying due to Iran-related costs, but there's no public confirmation yet of net selling.
But viewed through the lens of real rates, the picture is still positive:
| Year | Nominal FFR | US CPI | Real FFR | Gold ($/oz) | Gold YoY |
|---|---|---|---|---|---|
| 2022 | 1.68% | 8.0% | -6.32% | $1,800 | +0% |
| 2023 | 5.33% | 4.1% | +1.23% | $1,943 | +8% |
| 2024 | 5.33% | 2.9% | +2.43% | $2,386 | +23% |
| 2025 | 4.58% | 2.8% | +1.78% | $2,900 | +22% |
| 2026 (YTD) | 3.64% | 2.5% | +1.14% | $4,890 | +69% |
The core axis: the actual real rate (FFR − CPI YoY) is falling (2.43% → 1.14%) as the Fed starts cutting - but not fast enough to explain gold's +69% YoY move. The rest comes from M2 (global +12% YoY 2025-26), Chinese retail demand, and central bank buying. Gold has decoupled from the traditional real-rate model - the "geopolitical premium" plus the "de-dollarization premium" is estimated by Goldman/Amundi in 2025 at roughly 20-30% of the current price (by comparison against a macro fair-value model based purely on the USD index + real rates + ETF flows).
Methodology note: the table above uses the actual realized real rate (FFR − trailing CPI YoY) - suitable for long-run comparisons. The Q1/2026 correction card above uses the 10-year TIPS yield (expected, long-dated real rate) - reflecting market expectations, more sensitive to the Iran shock. These are two different indicators and shouldn't be conflated.
Outlook - 2026 Forecasts
Q4/2026 target: $5,400/oz (raised from $4,900 after clearing $5,000 in January). A bull-case scenario could push it to $6,000. Main drivers: continued central bank accumulation + ETF flows.
"A new peak?" - expects 250 tonnes of ETF inflows in 2026 (~$40-50B). Bull scenario: $5,000+. Risk: real rates rising fast.
The "structural bull cycle" continues. Forecasts an upward range without a specific number. Drivers: de-dollarization + Fed rate cuts + Chinese demand.
"Forward or back" - base case $5,000, bull case $6,000, bear case $4,000. The "structural" factor beats the "cyclical" one if real rates fall.
The long-run bull case rests on the four layers analyzed here: (1) global M2 keeps expanding from $118T toward $130-140T by 2027, (2) China keeps buying ETFs while the PBoC keeps accumulating, (3) the SGE-COMEX spread structure keeps Asia-US pressure alive, (4) non-US-allied central banks keep diversifying away from the USD. The near-term bear case: real rates rise (Iran, the Fed pausing cuts), some central banks selling due to cash shortfalls, and profit-taking after clearing $5,000. The structure still tilts upward - but the path will be very volatile.
Vietnam Connection - SJC Gold, The Premium, And Households
Vietnam has no official gold derivatives market. People hoard physical gold through shops like SJC, Phu Nhuan (PNJ), Mi Hong, and Bao Tin Minh Chau. Based on estimates compiled from years of gold-consumption data, Vietnamese households hold ~400-500 tonnes of physical gold (some higher estimates from VAFI put it at ~2,000 tonnes) - one of the highest ratios relative to GDP anywhere in Asia. The familiar reasons: memories of the hyperinflation of the 1980s, low trust in banks, and a tradition of hoarding gold as a "dowry" asset.
The consequence: when the world gold price rises, SJC gold rises with it but is always more expensive by a margin - because import quotas are tightly restricted. The gap between SJC gold and world gold (shorthand: the SJC premium) has at times reached 15-20%, peaking in panic around mid-2024 at ~25% (a gap of ~20 million VND/tael).
Two sequential interventions followed: (i) in June 2024, the SBV stopped auctions and switched to selling gold directly through the Big Four (Vietcombank, BIDV, VietinBank, Agribank) - the premium immediately fell from ~25% to 3-7% and stayed stable through the second half of 2024 and most of 2025; (ii) Decree 232/2025/NĐ-CP (issued 8/26/2025, effective 10/10/2025) formally abolished the state monopoly on producing SJC gold bars established under the earlier Decree 24.
Moving into 2026, as the world gold price kept breaking ATHs at a runaway pace, the SJC premium widened back out to ~12-13% (on 4/26/2026, a gap of ~20 million VND/tael) - even though Decree 232 was already in effect. The reason: the SBV still restricts raw-gold import quotas to protect the USD/VND exchange rate (which was showing signs of renewed tightness in early 2026), and domestic supply lags and can't keep pace with the world price - scrapping the production monopoly doesn't mean opening up imports.
A 12-13% premium is still high, but it's much healthier than the ~25% panic level of mid-2024 - and it's still higher than China's peak SGE-LBMA premium in 2023 (~$121/oz, ~6% over the London price), which reflects that Vietnam's gold import quota is even more restrictive than China's.
03 Discussion
Leave a note
A considered space for questions, counterpoints, and useful additions. Civil, on-topic, signed.
Reader notes
...Loading notes...