China’s 22-Month Gold Run: Diversification, Not a Dollar Exit

The conclusion: more gold, but not a new monetary order

China’s official gold reserves are rising, and this is not an isolated increase. The end-August figures released by the State Administration of Foreign Exchange (SAFE) on 7 September 2026 extend the run of monthly increases in reported quantity to 22, beginning in November 2024. The clearest conclusion is that additions to official gold holdings have persisted. The statistics do not, by themselves, establish a decision to abandon the dollar, a floor under the gold price, or a system in which renminbi can be exchanged for gold.^[1][2][3]

A central bank’s reserve portfolio is not the equivalent of a household investment fund. Expected returns matter, but so do confidence in the currency and financial system, and the ability to mobilise foreign currency when necessary. Adding gold therefore need not amount to a simple bet on a higher price. It can also reduce exposure to particular issuers, provide protection against different kinds of disruption, or change the portfolio’s composition. These are possible reserve-management objectives; the latest stock figures do not rank the Chinese authorities’ motives.^[6]

Look beyond the streak: three measures

The first measure is quantity: how much additional metal appears in the reported stock. The second is unit valuation: how much more or less each ounce is worth in dollar terms. The third is portfolio weight: gold’s share after allowing for all the other reserve assets as well. Treating these as interchangeable invites two errors—describing a modest physical addition as a huge new purchase, and treating a higher gold share as proof that dollar assets were sold. This particular release makes those distinctions unusually important.

In August, the reported quantity increased by about 0.85% from July, whereas its dollar value rose by about 14.27%. The latter is much larger because a change in unit valuation applies not just to the additional metal but also to the much larger pre-existing stock. Acquiring an asset and seeing an existing asset revalued are different events, even when they occur together. Reading the jump in value as expenditure on purchases would misstate the economic scale of this news.^[1]

The key distinction: the weight of the gold, its dollar valuation and its share of reserves can move differently.

SG Group reads the development more as a gradual expansion of alternatives within reserve management than as a wholesale exit from the dollar. That assessment is not permanent. Its significance changes if additions persist, if accumulation continues when prices decline, or if arrangements for turning gold into usable currency improve. The task is not to select the strongest political label. It is to identify the evidence that would warrant a different interpretation.

The same separation matters when reading markets. Central-bank accumulation explains one component of demand; it does not neutralise private outflows, real yields, exchange rates or profit-taking. A single headline is a poor basis for combining long-horizon reserve management with short-term price movements. The guide to gold, real yields, the dollar and demand provides a starting point for separating those influences.

What increased? Reading the end-August figures

For the official gold reserves associated with the People’s Bank of China (PBoC), SAFE’s monthly table reports both a valuation and a quantity in ounces. The subject is the monetary authorities’ official reserve assets—not the combined holdings of Chinese households, jewellers, mines and private companies. The shorthand “China’s gold holdings” should not be read as a measure of every ounce owned or stored in the country.^[1]

At the end of August 2026, reported holdings were 76.73 million troy ounces, up by 650,000 ounces from 76.08 million at the end of July. A troy ounce—the unit used in the gold market—is 31.1034768 grams. That converts to an addition of about 20.22 metric tonnes and a stock of approximately 2,386.57 tonnes. The decimal places are the result of converting the published quantity, not a new measurement of vault contents to that degree of precision. Unit conversion does not improve the precision of the original statistic.^[1][5]

The “increase” in the table is a difference between two month-end stocks. It does not provide gross purchases and sales during the month, or the dates on which contracts were agreed and settled. If several transactions occurred, the published stock records their combined outcome. The 650,000-ounce addition therefore cannot be assigned to a single order on a particular day, nor does it establish that the same amount was shipped into China from one particular overseas market during August.

An increase in foreign-exchange reserves is not the same as an inflow

Separately from gold, foreign currency reserves stood at $3,438.325 billion at the end of August, an increase of $19.549 billion, or about 0.57%, from July. SAFE’s explanation of the monthly change refers to exchange-rate translation and changes in asset prices, among other factors. Assets held in currencies such as the euro can change in dollar value solely because exchange rates move. An increase in dollar-reported foreign currency reserves is therefore not equivalent to an inflow of that many new dollars.^[1][4]

For the same reason, the combination of more gold and higher foreign currency reserves does not identify what was sold to acquire the gold. Changes in the overall portfolio involve multiple assets, exchange-rate translation, revaluation and transactions. Connecting an addition of gold to a sale of a particular government bond requires evidence beyond stock totals. The monthly table is a valuable starting point for understanding the portfolio, but it is not a transaction ledger.

There is also a timing distinction. The end-August observation date is not the 7 September publication date. For a reader on 8 September, the new information is the disclosure of the quantity—not a purchase of the same amount taking place on 8 September. Ignoring that distinction risks counting an action that may already have influenced market prices as additional demand still to come.

Inside the 22 months: persistence and scale are different

The current run began in November 2024. The preceding October stock was 72.80 million ounces—the same reported quantity as in each month from May through October that year. Holdings rose to 72.96 million ounces in November and 73.29 million in December, then increased in every month of 2025 and through August 2026. The number 22 counts positive month-to-month stock changes. It does not establish daily buying, a fixed purchase budget or a commitment to continue.^[1][2][3]

From the end-October 2024 baseline, the cumulative addition over the 22 months is 3.93 million ounces, or approximately 122.24 tonnes. That represents about 5.40% of the starting quantity. The run is long, but duration alone does not mean the core of the reserve portfolio has rapidly been replaced by gold. Persistence is one dimension; the size of the change relative to the existing stock is another.^[1][3]

In 2026, the change is more than mere continuity

Equally, dismissing the news as just another routine addition would overlook a change in scale. The increase during all of 2025 was 860,000 ounces, or about 26.75 tonnes. From January through August 2026, it was 2.58 million ounces, or approximately 80.25 tonnes—three times the 2025 total. The monthly increments within the 22-month run are not uniform. The volume of additions has become materially larger in 2026.^[1][2][3]

The monthly additions in 2026 were 40,000 ounces in January, 30,000 in February, 160,000 in March, 260,000 in April, 320,000 in May, 480,000 in June, 640,000 in July and 650,000 in August. From March through August, each increment exceeded the previous month’s. But August’s addition exceeded July’s by only 10,000 ounces, or about 0.31 tonnes. A continuing increase in monthly additions is not the same claim as a sharply accelerating rate of expansion.^[1][2]

Reading scale: January–August additions were three times the full-year 2025 increase, but August’s increment was close to July’s.

The history is consistent with changing the reserve mix through additions spread over time. It does not tell us whether the authorities are following a predetermined volume plan or exercising discretion in response to prices and market conditions. The same pattern of monthly increases can arise from different decision processes. Continued additions during a rising market would not establish an unlimited willingness to buy at any price. Persistence and sensitivity to price need to be assessed separately.

Interpretation should also be measured if the streak eventually ends. A flat month could reflect a pause, but stock data alone may not distinguish that from offsetting transactions or recognition timing. Conversely, a longer streak with very small additions would not carry the same policy significance as sustained larger increments. Looking at additions over several months is more informative than reacting to the binary question of whether the run continued.

Decomposing the valuation jump: purchases are not revaluation

The reported dollar value of gold rose from $306.354 billion at the end of July to $350.080 billion at the end of August, an increase of $43.726 billion. That increase must not be described as China’s expenditure on gold during August. Even without transaction prices and execution dates, combining quantity with valuation allows us to gauge the part that cannot be explained by additional ounces alone. This is where the first analytical framework—the three measures—becomes useful.^[1]

Dividing the reported value by the reported quantity gives an implied valuation per ounce: approximately $4,026.74 for July and $4,562.49 for August. These are not the authorities’ actual acquisition prices, nor are they an independently observed series of closing prices from a particular market. They are unit valuations inferred from two entries in the same statistical table. They can nevertheless be used for an accounting decomposition of the change in value.^[1]

About 94% sits on the valuation side—not a claim about price causation

Let Q be quantity, P the implied unit valuation and V the reported value, so that V equals Q multiplied by P. Under a convention that values additional ounces at the previous month’s unit valuation, the change in V splits into “additional quantity × previous unit valuation” and “current quantity × change in unit valuation”. For August, the first component is about $2.617 billion and the second about $41.109 billion. The latter accounts for approximately 94.0% of the increase. Under this convention, the interaction between changes in quantity and unit valuation is assigned to the valuation component.

The decomposition shows that most of the increase in reported value cannot be explained by additional ounces alone. It does not estimate how much PBoC demand caused the world gold price to rise. Nor is the $2.617 billion quantity component the actual amount paid for the gold: transaction prices during the month need not equal the previous month’s implied valuation. An exact accounting identity is not a reconstruction of the underlying trades.

For reserve management, this is more than an arithmetic distinction. If a price increase raises gold’s portfolio weight, concentration can rise without any additional acquisition. Further accumulation can diversify exposure to issuers while simultaneously increasing exposure to a volatile asset. A valuation gain on past acquisitions does not establish that additional purchases will produce the same result. Public institutions, too, have to reckon with portfolio weights changed by market prices.

Gold’s share is about 9.08%—use total reserves as the denominator

Total official reserve assets at the end of August were $3,854.885 billion. Dividing gold’s $350.080 billion value by that total gives a share of about 9.08%, up from approximately 8.08% in July. Dividing instead by foreign currency reserves alone produces about 10.18%, but that is a different ratio: gold relative to foreign currency reserves. For gold’s share of total reserves, the denominator must include gold itself, Special Drawing Rights, the IMF reserve position and the other components of official reserve assets.^[1]

The rise of roughly one percentage point is not evidence that an equivalent share of dollar assets was sold. If the numerator—the value of gold—rises substantially, the share can increase without sales of other assets. Conversely, additional purchases can coexist with a falling gold share if the gold price declines. Quantity is the starting point for assessing accumulation; valuation incorporates market movements; portfolio weight reflects both of those and changes elsewhere. The right statistic depends on the question being asked.

From ownership to payment: gold has four gates to pass

Gold’s status as an asset that is not the debt of a particular government or company is important for reserve management. Unlike a bond, it does not depend on an issuer repaying principal. But “no issuer” must not be translated into “usable anywhere, at any time, with any counterparty”. Custody, bar specifications, title, market access, a buyer or collateral counterparty, and the currency of final settlement are separate operational conditions.^[6][10]

The second framework is therefore a four-gate path from ownership to payment: title, custody and delivery, monetisation or financing, and final payment. An increase in ounces chiefly deepens the stock at the first gate. Assessing resilience in a crisis also requires examining the route by which that asset can be mobilised, traded, converted into the currency needed and delivered to a recipient. The distinction applies to official gold reserves generally, not only to China.

Custody diversification and ease of use need not move together

Domestic custody can reduce direct dependence on an overseas custodian or foreign legal decisions. Holding gold close to an international trading centre may, on the other hand, make sales, collateral use and delivery easier. Which arrangement is preferable depends on the disruption being considered and the transactions required. Safety cannot be measured by location alone, or liquidity by weight alone. China’s monthly reserve table does not provide a custody-location breakdown or the terms governing individual holdings.^[1][10]

A useful contrast is the reserve adjustment announced by De Nederlandsche Bank (DNB) on 2 September 2026. To improve tradability, DNB said it had shifted 86 tonnes of holdings from New York and Ottawa to London between March and August 2026. Of this, 59 tonnes involved sales and repurchases in London, while 27 tonnes were transported. Total holdings remained 612.4 tonnes. This illustrates how usability can change without an increase in quantity; it does not suggest that China has undertaken the same operation.^[11]

The reverse test: more gold need not mean more usability, and usability can improve without more gold.

Gold can reduce credit exposure to a particular issuer, but it does not automatically remove relationships with custodians and transaction counterparties. Depending on the disruption, a willing buyer might be unable to transact, a settlement bank might be unavailable, or delivery specifications might matter. Calling gold completely immune to sanctions is therefore too strong. Legal rights over an asset and the ability to convert its value into an international payment are different properties.

This distinction is not intended to diminish accumulation. It is a way of assessing reserve management more substantively than by weight alone. If China expands its usable custody, settlement and financing options, the same quantity could have a different strategic value. If access to needed markets deteriorates while holdings rise, the gap between accounting wealth and resources that can actually be mobilised could widen.

How far does gold accumulation amount to de-dollarisation?

The term “de-dollarisation” covers several distinct developments: reducing the dollar share of reserves, changing the currency in which trade is invoiced, borrowing internationally in another currency, and changing payment routes. Gold accumulation is an asset-side development. It does not directly tell us how companies price transactions, how banks fund themselves or how import invoices are settled. Combining reserve assets and currency networks into one metric makes it easy to overstate the change.

The International Monetary Fund’s Currency Composition of Official Foreign Exchange Reserves database, COFER, is not a measure of total official reserves including gold. It covers the currency composition of foreign-exchange reserves, and individual country breakdowns are confidential. Placing the global dollar share beside China’s gold share does not allow China’s dollar holdings to be calculated: the coverage, country aggregation and denominators differ. Inferring an equal reduction in dollar assets from an increase in gold is a step beyond the evidence.^[8]

Trade and funding create persistence in reserve-currency demand

If companies make payments and borrow in a particular currency, the monetary authorities retain a reason to have access to that currency. BIS research on reserve composition highlights links to factors such as trade invoicing. Increasing gold holdings does not instantly eliminate those operational requirements. Where gold is sold to obtain dollars for a final payment, diversification at the asset-holding stage can coexist with continuing dollar use at the payment stage.^[9]

Two interpretations can therefore coexist. A higher gold share can matter as a reduction in dependence on currency issuers, while remaining insufficient evidence of a change in the global monetary order. There is no contradiction. Reserve managers can continue to use the existing payment system while building protection against relying on it exclusively. Both “gold has no purpose while dollar dependence remains” and “buying gold makes the dollar unnecessary” impose an overly narrow choice.

What the evidence measures: accumulation concerns assets. A change in settlement currency needs separate evidence on invoicing, borrowing and payment.

The rise in reserves also does not establish a transition to a gold-backed renminbi. Owning gold is different from accepting an obligation to convert currency into gold on specified terms. Such an arrangement would raise separate questions about eligibility, conversion conditions, pricing and institutional operation. The monthly statistics concern ownership of an asset, not the creation of a redemption right.

Conversely, the persistence of the dollar’s role does not make the accumulation meaningless. Changing the reserve mix can be understood as building options in normal conditions rather than attempting a rapid adjustment after a crisis begins. A meaningful change need not take the form of a dramatic institutional replacement. Assessing China’s actions requires allowing for both the strength of the existing system and attempts to adjust dependence on that system at the same time.

Misreadings and alternatives: one stock series, several hypotheses

A geopolitical explanation for gold accumulation has intuitive appeal. But official reserves serve multiple purposes, potentially combining diversification, contingency protection and adjustments to risk and return. ECB research also examines several motives behind official-sector demand. China’s run of increases cannot, on its own, be treated as a forecast of a particular military action or the timing of financial sanctions. Buying more insurance is not the same as knowing when an accident will occur.^[7]

An alternative hypothesis is reserve management responding to prices and portfolio weights. The terms on which metal can be acquired, and the timing of transactions, may also affect the reported sequence. These explanations are not equally established, but the stock series does not uniquely select one of them. A useful analysis sets out what further evidence each hypothesis would lead us to expect, rather than simply choosing the most appealing story.

A central-bank survey is not China’s purchase schedule

The World Gold Council’s central-bank survey published in June 2026 offers context for official-sector interest in gold. But survey responses are not commitments to future transactions, and the respondents’ collective views do not identify the intentions of the PBoC. Several central banks valuing gold does not mean they will all buy at the same time or for the same reason. Expectations surveys and reported stocks are different kinds of evidence.^[13]

Claims that the “real” holdings must be larger than the reported number require care as well. Official reserves, assets of other public entities, commercial-bank or private inventories and domestically mined gold are not the same statistical category. The existence of gold in an adjacent category does not justify automatically adding it to PBoC reserves. Equally, the reported reserve figure is not an upper limit on all public and private Chinese holdings. Preserving the boundaries is more informative than presenting a precise-looking total assembled from incompatible categories.

A key misreading: the existence of a buyer does not establish a price floor. Other buyers, sellers and their terms still matter.

Another argument is that gold should be difficult to push lower while central banks are accumulating it. Persistent demand can provide support, but one group of buyers need not absorb all selling at every price. If rates or the dollar move adversely, or private holders need cash, prices can fall even while official demand continues. The reserve-diversification conclusion and the short-term market-direction conclusion need to remain separate.

Likewise, using COT percentiles and z-scores to compare futures positioning is not the same as measuring central-bank physical reserves. Commitments of Traders data describe positions under participant categories in particular markets, with different coverage and frequency from monthly reserve statistics. The value of considering both is not that they validate the same number. It is that long-term official holdings and shorter-term private-market positioning may point in different directions.

SG Group View: assess the news with three clocks

SG Group’s central assessment is that China is expanding its reserve-management options, but the speed and practical reach of that adjustment should not be overstated. The larger additions in 2026 matter more than the extension of the streak alone. At the same time, most of August’s valuation jump cannot be explained by additional ounces, and a gold share of roughly 9% is not a basis for declaring a new monetary order. Real accumulation and the impression amplified by prices should be considered separately.

The third framework uses three clocks: the clock of transactions and their recognition in holdings, the clock of month-end disclosure, and the multi-year clock of reserve reallocation. Markets can respond immediately to a release, but the action disclosed has already occurred. How far that action changes the long-run reserve mix depends on accumulation over much longer periods. Different questions about the same gold headline therefore require different time horizons.

Immediacy is overstated; the value of persistent optionality is understated

Two things are easily overstated: the ability of a streak to predict the next price rise, and the extent to which a higher gold valuation constitutes immediate de-dollarisation. What is more easily understated is the way repeated decisions in normal conditions can alter the options available in a crisis. Reserves do not have to become entirely gold for that to matter. Holding an asset with different characteristics alongside needed foreign currencies can itself be valuable. A gradual change is not necessarily an unimportant one.

Optionality also has costs. Gold itself does not pay interest; storage and transactions carry expenses; and its value at the moment it is needed is not guaranteed. Holding it can also forgo interest that a foreign-currency bond might have earned. These are not estimates of the costs China incurred in this episode. They are necessary comparisons when assessing gold holdings. An account that emphasises safety while omitting opportunity cost is incomplete.^[6]

SG Group’s judgment: important as persistent diversification, not as a short-term price guarantee or a completed replacement for dollar settlement.

The conditions that would weaken this view are clear. If additions stop for several months while the share continues rising solely because of prices, the assessment of accelerating substantive diversification should be scaled back. If gold proves difficult to convert into the currencies required, its value as a crisis option also needs reassessment. Conversely, continuing additions combined with stronger custody and monetisation channels would support a greater practical role for gold.

For short-term analysis, the distinction between publication delays and lead–lag relationships is especially important. Treating August holdings as information available at the start of August introduces knowledge that became public later. Even when month-end prices and holdings move together, sequencing and common causes remain unresolved. The proposition that the release is valuable for interpreting policy and the proposition that it is a leading trading indicator require separate tests.

More gold can expand the alternatives to exclusive dollar dependence without eliminating the need to use dollars.

How could this affect households, businesses and Japanese markets?

For households, the most immediate connection is to the prices of gold products and assets they may own. But the yen price faced by a Japanese consumer is not determined by the international dollar price alone. Exchange rates, retail mark-ups, fabrication and distribution costs also matter. A single fact about Chinese official holdings cannot quantify the effect on Japanese retail prices. Nor does this release, by itself, justify a broad claim about direct effects on everyday food or electricity bills.

Existing holders benefit in valuation terms if gold prices rise. For someone buying jewellery or gold for industrial use, the same price increase is a cost. The division between beneficiaries and those bearing costs is therefore not just between countries; it is also between existing owners and new purchasers. A higher valuation is not equivalent to cash income before an asset is sold. Assessing its contribution to household spending or business funding also requires looking at conversion terms and costs.

For businesses, contract terms matter more than the headline

For jewellers, fabricators and businesses using gold in electronic components, the delay between input-cost changes and selling-price adjustments matters. If fixed-price sales contracts remain in place while material costs change, margins can be squeezed temporarily. Conversely, higher inventory valuations may improve the accounts while increasing the cash needed for the next purchase. A higher gold price need not affect reported profit and working-capital requirements in the same direction.

Miners and other resource businesses do not automatically receive a proportional increase in profit when gold prices rise. Output, energy and labour costs, exchange rates, capital expenditure and sales contracts all intervene. The conditional beneficiary is a business whose realised selling-price improvement exceeds its cost increases and contractual constraints. Translating a global demand headline into an equity earnings claim requires that business-level bridge.

The Japanese transmission path: dollar gold price → exchange rates and contract terms → retail prices, input costs and asset valuations. Each stage has other drivers.

For investors and traders, the form of exposure matters as well. Physical gold, official reserve holdings, futures, exchange-traded products and contracts for difference (CFDs) may share some price exposure, but differ in custody, costs, maturities, margin requirements and counterparty relationships. A central bank’s reason for holding gold over a long period cannot simply be transferred to a short-horizon instrument with financing costs. Even with the same price direction, costs and time horizons can produce different economic outcomes.

For policymakers and corporate treasury teams, the lesson is to separate a broader asset mix from dependable payment capacity. Whether the right currency is available at the right time is a more operational question than whether gold has appreciated. China’s latest statistics are important evidence of reserve accumulation, but they are not instructions for the cash management or asset allocation of an individual company or household.

Conditional scenarios: what would change the assessment?

For this news, it is more useful to vary the combination of quantity, unit valuation, liquidity and settlement needs than to attach a single target to the gold price. The reference scenario is that additions continue without quickly eliminating the role of foreign currency reserves or established payment currencies. Persistent accumulation would support the diversification interpretation, while shorter-term market direction would still depend on rates, private flows and other influences.

In a stronger-accumulation scenario, the first test is whether additions remain elevated over several months. The next questions are whether the higher gold share reflects more than prices, and whether foreign-currency liquidity remains adequate. Larger additions would make the diversification interpretation more persuasive, but they would also raise exposure to a non-interest-bearing asset and its price volatility. Greater quantity should be assessed as a reserve-management trade-off, not an unqualified positive.

A pause and a liquidity shock challenge different parts of the view

In a pause scenario, continuing valuation gains would no longer support the claim that diversification is advancing through additional quantity. A single flat month, however, would not establish abandonment of a long-term approach. Looking across several months at holdings, prices and other reserve assets would make it easier to distinguish price sensitivity, allocation adjustments and temporary timing effects. Small changes in stocks call for proportionately cautious conclusions.

A more demanding scenario is one in which gold holdings rise while access to foreign-currency funding or settlement becomes harder. Even a higher gold price and a larger stock may do little for immediate payment capacity if the required currency cannot be obtained by the required date. The test is then the ability to pass the four gates, not the weight of the metal. Large gold reserves and the absence of short-term foreign-currency shortages are not equivalent assurances.

How to use the scenarios: they are branches for updating an assessment, not assigned probabilities or price targets.

If additions continue while the dollar or real yields move against gold, that would provide evidence about the persistence of official demand. If prices fall at the same time, however, it would also demonstrate that official accumulation does not guarantee a price floor. Evidence strengthening a policy interpretation need not strengthen a bullish market interpretation. Keeping those propositions separate helps prevent selective use of favourable price movements as evidence.

This approach is consistent with building macro scenarios around explicit conditions. Define in advance what would change the assessment rather than replacing the explanation after the outcome is known. A story that claims to be vindicated whether gold rises or falls cannot be tested. What this release calls for is not a narrative capable of accommodating every result, but a way to distinguish which parts of the hypothesis gain support and which weaken.

What the monthly statistics cannot tell us

First, actual acquisition prices and funding sources cannot be identified from month-end stocks. The implied unit valuation discussed earlier is value divided by quantity, not a purchase price. Identifying which currency funded an acquisition, which existing assets were moved, or which domestic or overseas markets were used requires separate information in each case. Two reported totals should not invite an imagined single transaction linking them.

Second, quantity does not reveal custody locations, delivery conditions, dependence on particular counterparties or the availability of particular holdings for use. Assets with the same accounting value may produce different amounts of usable funding at a given time because of location and market-acceptance conditions. The stronger a geopolitical claim, the more important these non-weight conditions become. Ownership is not a guarantee of frictionless mobilisation under every crisis scenario.

Comparing China with the world introduces another problem

Third, international comparisons need aligned observation dates. The WGC’s country data compile reports from the IMF and other sources, with lags between website updates and the months covered for individual countries. A page updated in September does not necessarily contain end-August observations for every country. Mixing China’s latest figure with older observations elsewhere risks presenting differences in timing as differences between countries.^[14]

Fourth, the monthly table does not provide a future purchase schedule or target portfolio weight. August’s 650,000-ounce addition does not imply the same increment in September, and a share of about 9% does not identify any intended destination. Past increments can be mechanically extrapolated, but such a calculation would rest on an assumption; it would not represent an official plan.

The boundary: reported accumulation is clear. Ranked motives, transaction details, custody-specific usability and future targets are separate questions.

Fifth, the isolated price impact cannot readily be reduced to a single number. Other buyers and sellers, interest rates, exchange rates and funding conditions move at the same time. A price increase in a month with large official additions does not, by itself, allocate causation. A price decline does not prove the demand was irrelevant either: the counterfactual price without that demand is not directly observed.

These uncertainties do not prevent useful analysis. Distinct questions about the quantity history, valuation components, portfolio weight and usability make it clear what each additional piece of information could resolve. Updating the assessment in stages is a stronger approach to both policy and markets than forcing all the uncertainty into one supposedly complete narrative.

The next dates and the numbers that matter

The next focal point is the end-September 2026 quantity. SAFE’s 2026 release calendar schedules the next official-reserve-assets release for 7 October 2026. This is a planned date and may change. The first entries to examine are ounces, the month-to-month difference and the comparison with recent additions—not the dollar valuation alone. Whether the streak reaches 23 months matters less than whether the scale of accumulation is maintained.^[12]

The related international-reserves and foreign-currency-liquidity template is scheduled for 30 September 2026. The calendar indicates that releases generally cover the preceding reporting period, so a document released on 30 September should not be mistaken for an end-September stock report. Its role is to supplement headline reserve totals with liquidity-related information, not to provide a complete account of individual gold transactions or custody arrangements.^[12]

Repeat the same quantity–valuation comparison

After quantity, compare the implied unit valuation—reported value divided by ounces—with the previous month. A flat quantity and higher valuation would not be the same kind of accumulation news. More ounces but a lower value would indicate that the adverse valuation movement outweighed the quantity effect under the relevant decomposition. Then calculate the gold share using total reserves as the denominator. Repeating this sequence helps avoid selecting only the statistic that supports a preferred de-dollarisation story.

On the market side, examine US real yields, the dollar, private investment demand for gold and futures positioning separately. These do not replace Chinese reserve statistics; they provide context for whether official demand and the broader market are moving together. It would be as mistaken to insist that higher real yields must always depress gold as to ignore rates because central banks are buying. The strength of those relationships can itself vary across periods and market conditions.^[6]

The next sequence: change in ounces → implied unit valuation → share of total reserves → foreign-currency liquidity and usability.

Updating the political interpretation requires separate evidence of specific changes to settlement currencies, custody or trading arrangements. General diplomatic rhetoric is not the same as an operational system being introduced. What would strengthen the interpretation is not a growing number of similarly worded headlines, but information showing which of the four gates has changed in practice.

Finally, revisions to historical observations would require recalculating both the streak and monthly increments. Tracking monthly data means keeping the observation month, release date and data version consistent. This preserves what could have been known at the time rather than merely making a forecast look more precise after the event. A repeatable method is the foundation for following reserve reallocation without being pulled around by headlines.

Final assessment: look beyond weight to the change in options

The 22-month increase is clear evidence of persistent additions to reported gold reserves, and the larger increments in 2026 matter. But the sharp jump in valuation that makes this release look especially large owes much to factors other than additional ounces. Without separating weight, dollar valuation and portfolio share, changes in official behaviour become confused with changes in market valuation.

SG Group’s assessment is of gradual but meaningful reserve diversification. It does not imply that dollars will cease to be used tomorrow or that gold prices cannot fall. Accumulating an asset with different characteristics while retaining access to existing foreign currencies is a better fit with both the quantities and the institutional context. A stronger conclusion would require evidence not just of continued additions, but of a change in how the asset can be used.

The question that survives the next headline

Whether the next release extends the streak or ends it, the central questions remain: how much did quantity change, what changed the valuation, and can the gold be connected to the payment needed at the time it is needed? Read together, those questions turn “China bought again” from a recurring headline into evidence about the structure of reserve management.

An increase in gold is an entry point for thinking about the future of reserves, not a statistic that can supply every conclusion. The news value lies not just in the run, but in the scale of additions and the options they may create. A serious question about de-dollarisation must look beyond the vault to the connections among trade, funding, settlement and custody.