A Foreign Reserves Total Is One Number. Its Disclosure Template Has Four Sections.

A central bank publishes a reserves figure once a month. Wire copy carries it as a single number with a month-on-month change attached, analysts turn it into months of import cover, and sovereign credit commentary cites it as a buffer. Very little of that engages with a basic fact: the number is one line of a four-section statistical return, and the other three sections exist because that one line was once shown to be misleading.

The return is the IMF's Data Template on International Reserves and Foreign Currency Liquidity, usually shortened to IRFCL. It was folded into the Special Data Dissemination Standard in March 2000, four years after the SDDS itself was established, and the reason is not obscure. In its own retrospective on the 1997–98 programmes in Indonesia, Korea and Thailand, the IMF wrote that gross reserves were a poor indicator of available international liquidity given the magnitude of liabilities set against these reserves, many appearing off-balance sheet, and singled out forward contracts outstanding as the item that mattered most in the Thai case. The template is the institutional answer to that finding. It does not replace the headline total. It surrounds it.

The quoted number is one line of a much longer return

The template has four sections. Section I covers official reserve assets and other foreign currency assets. Section II covers predetermined short-term net drains on those assets. Section III covers contingent short-term net drains. Section IV holds memo items. The figure that reaches a headline is Section I.A alone, valued at market prices, aggregated to a single total.

What matters for anyone reading a reserves release is that the sections are published alongside each other and are not netted into one another. A country can report a large Section I.A total and, on the same return, disclose obligations in Section II that consume most of it inside thirty days. Both statements are accurate. Only one of them travels.

Five lines inside Section I.A, five different speeds

Even before any drain is considered, Section I.A is not homogeneous. It contains five lines, and they behave differently under stress.

• Foreign currency reserves — securities, plus currency and deposits held with other central banks, the BIS, the IMF and commercial banks. This is the part that functions the way a headline implies: it can be spent or intervened with directly.
• Reserve position in the IMF — a claim arising from the country's quota subscription, drawable on the IMF, not a balance sitting in a correspondent account.
• SDRs — the IMF describes the SDR as a potential claim on the freely usable currencies of IMF members, rather than a claim on the IMF itself. Turning an SDR balance into spendable dollars requires a counterparty transaction.
• Gold — the template line reads, verbatim, Gold (including gold deposits and, if appropriate, gold swapped), with the volume also reported in millions of fine troy ounces. Gold is a reserve asset at market value. It is not a payment instrument, and monetising a material quantity of it at short notice is a market operation with its own price impact.
• Other reserve assets — financial derivatives and loans to nonbank nonresidents.

Section I.B then lists foreign currency assets deliberately not counted as reserves — securities, deposits, loans, derivatives and gold that fail the reserve-asset test. It forces a country to show what it holds in foreign currency but cannot claim as a reserve.

Section II is where a forward book becomes visible

Section II covers scheduled contractual obligations in foreign currency — things the authorities already know they must pay or will receive. Three item groups sit here: foreign currency loan, security and deposit flows; aggregate short and long positions in forwards and futures in foreign currencies vis-à-vis the domestic currency, including the forward leg of currency swaps; and residual items such as repos, trade credit and accounts payable or receivable.

Each of those lines is split into three maturity buckets: up to one month, more than one and up to three months, and more than three months and up to one year. That bucketing is the analytical payload. A forward book maturing across twelve months is a different proposition from the same notional maturing inside thirty days, and the template refuses to let the two look identical.

None of this is subtracted from Section I.A. A short forward position does not reduce the reported reserve total; it appears one section below, on its own line, in its own bucket. Whether a reader gets there is a matter of habit rather than data availability.

Section III prices what has not happened yet

Section III handles contingent flows — obligations that may or may not crystallise. It carries collateral guarantees on debt, foreign currency securities issued with embedded options such as puttable bonds, undrawn unconditional credit lines both received from and provided to other monetary authorities and financial institutions, and option positions in foreign currency. The option lines carry an unusual instruction: they are stress-tested for in-the-money value at exchange rate moves of plus and minus 5% and 10%.

Section III is symmetric, and that symmetry is easy to miss. Undrawn credit lines provided to others are a potential outflow. Undrawn credit lines received from others are a potential inflow that is not in the headline at all. A country with a modest Section I.A total and a large committed facility on the receiving side sits in a materially different position from one with the same total and no such line.

Swap lines are dollar liquidity, and they are not reserves

The clearest demonstration that usable foreign currency and reported reserves are separate quantities is the Federal Reserve's swap network. Six central banks — the Fed, the ECB, the Bank of Japan, the Bank of England, the Bank of Canada and the Swiss National Bank — converted their temporary bilateral arrangements into standing ones on 31 October 2013, arrangements that remain in place until further notice. On 19 March 2020 the Fed added temporary lines with nine more central banks, sized at USD 60 billion for Australia, Brazil, Korea, Mexico, Singapore and Sweden, and USD 30 billion for Denmark, Norway and New Zealand, for at least six months. On 28 July 2021 the FIMA repo facility, which lets foreign official institutions raise overnight dollars against Treasuries held in custody at the New York Fed, was made standing, with a per-counterparty limit of USD 60 billion and an initial minimum bid rate of 25 basis points.

None of that capacity is a reserve asset for the borrowing central bank in the sense of Section I.A. Undrawn, a committed line belongs in Section III. Drawn, it creates a foreign currency liability that shows up in Section II as a scheduled repayment. The chart below shows the Fed side of the 2020 episode.

The shape is the point. Outstanding swaps stood at USD 45 million on 18 March 2020, the day before the nine temporary lines were announced. One week later the figure was USD 206.1 billion. It peaked at USD 448.9 billion on 27 May and had fallen to USD 6.8 billion by 28 October, before rising back to USD 17.9 billion on 30 December as year-end funding pressure reappeared. Several hundred billion dollars of genuinely usable liquidity moved into and out of the system inside nine months, and none of it entered a reserves headline anywhere.

The mirror image sits off-balance-sheet in private markets. In its December 2022 Quarterly Review, the BIS estimated that as of mid-2022 non-banks outside the United States carried roughly USD 26 trillion of dollar obligations from FX swaps and forwards against about USD 13 trillion on balance sheet, and non-US banks roughly USD 39 trillion against about USD 15 trillion, inside a global stock of some USD 97 trillion across all currencies. The BIS called it missing debt because standard debt statistics do not capture it. The reserves template captures the official sector's share of the same phenomenon — but only in Sections II and III.

The calendar matters as much as the composition

Reserves data arrives on two different clocks, and the faster clock carries the thinner information.

Series	Frequency	Timeliness	Basis
Official reserve assets (the headline)	Monthly	Within one week	SDDS prescription
Full IRFCL template, Sections I–IV	Monthly	Within one month	SDDS prescription
COFER currency composition (world total)	Quarterly	2025Q3 released 19 December 2025	Voluntary IMF survey, not an SDDS category

The first two rows are prescriptions of the Special Data Dissemination Standard, binding on subscribing economies. The third is not; it is treated separately below. Markets therefore routinely react to the headline weeks before the drains that qualify it become public. In a calm month that gap is trivia. In a stressed month it is the whole story, because the headline moves first and the explanation arrives later.

The currency-composition picture is slower and blunter still, and it sits outside the standard. COFER data are reported to the IMF voluntarily, and data for individual countries are kept strictly confidential. What the dataset publishes is a world total of official foreign exchange reserves by currency — the dollar, euro, renminbi, yen, sterling, Australian dollar, Canadian dollar and Swiss franc, plus an other-currencies residual. The release of 19 December 2025, covering the third quarter of 2025, put total foreign exchange reserves at USD 13.0 trillion, with the dollar share at 56.92%, the euro at 20.33% and the renminbi at 1.93%. It also carried a methodology change: from 2025Q3, with revisions back to 2000Q1, the IMF eliminated the unallocated portion and now imputes it, producing a composition covering 100% of the world total. The advanced-economy and emerging-economy breakdown was removed at the same time, back to 2000Q1. Anyone comparing a current dollar share against an older vintage is comparing two different constructions.

Adequacy ratios inherit whatever numerator they are handed

The standard adequacy rules all divide reserves by something: roughly three months of import cover; the Greenspan-Guidotti benchmark of 100% of short-term external debt; a ratio of around 20% of broad money for financially open economies. The IMF's ARA metric combines four weighted components — short-term external debt, broad money, export income and other liabilities — with heavier weights for fixed and managed-float regimes than for floaters, adjustments for dollarisation, capital controls and commodity exposure, and an adequacy band commonly cited as 100–150% of the metric.

Each ratio is only as good as its numerator. If that numerator is Section I.A gross, a large short forward book, a pledged tranche of securities and a gold-weighted portfolio are all invisible to it. Two countries at the same ARA coverage can sit in quite different places once Sections II to IV are read. That is not a criticism of screening tools; it is a caution against treating one as a liquidity assessment.

Where this reading breaks down

The template can mislead in the other direction too.

• A large forward book is not automatically a hole. Section II reports aggregate short and long positions. A central bank running a matched book for exporters may carry a large gross short position offset by scheduled inflows. Reading the short leg alone overstates the drain.
• Gold-heavy portfolios are not uniformly weak. Gold is slow to monetise, but it does not carry the settlement, sanctions or counterparty risk attached to holding another sovereign's securities. Which weakness matters depends on the scenario being tested.
• Small headline totals can conceal strength. A standing swap line with a major reserve-currency central bank does not appear in Section I.A at all. A country with a modest total and reliable access can be better placed than one with a larger total and none.
• Disclosure is not uniform. The full template applies to SDDS subscribers and to SDDS Plus adherents. Economies outside those standards may publish a headline with nothing comparable behind it, and an undisclosed drain is not an absent one.
• None of this is a timing signal. A stretched Section II raises the conditional probability that a defence of an exchange rate runs into constraints sooner than the headline implies. It does not date the event, and positions of this kind have persisted for long stretches without incident.

What to Watch Next Week

• The template release, not the headline release. Note the two publication dates separately and see whether the market reacted to the first and ignored the second.
• The up-to-one-month bucket in Section II. Its ratio to the foreign currency reserves line in Section I.A is the single most compressed statement of near-term room.
• Section III credit lines, both directions. A newly disclosed line received, or a line provided that has grown, changes the contingent picture without touching the headline.
• Section IV pledged assets and securities lent or on repo. A rising encumbered share inside a flat headline is a change the total will not show.
• Fed H.4.1, line “Central bank liquidity swaps.” Weekly, Thursdays. Any sustained move off a near-zero base indicates offshore dollar funding stress before it appears in monthly national data.
• The next COFER release. Quarterly, and now on the post-2025Q3 methodology, so check the vintage of any comparison figure before drawing a trend.

Concrete Framework

A repeatable order of operations, using only what the publishing authority already discloses.

1. Locate the actual return. Find the IRFCL template on the central bank or IMF data portal rather than the press summary, and confirm the reference month against the publication date.
2. Split Section I.A into its five lines. Separate foreign currency securities and deposits from the IMF reserve position, SDRs, gold and other reserve assets. Only the first line is directly spendable; note the share it represents.
3. Read Section II by bucket, not in total. Record the net drain due within one month, then within three months, then within one year. Compare each against the foreign currency reserves line, not against the full headline.
4. Check both sides of Section III. Note undrawn lines received as potential inflow and lines provided plus guarantees and puttable bonds as potential outflow. Read the option stress-test columns at plus and minus 5% and 10% as a rough sensitivity to a currency move.
5. Subtract encumbrance from Section IV. Deduct pledged assets and securities lent or on repo from the spendable line. Note the currency composition disclosure while there.
6. Recompute one adequacy ratio twice. Run the chosen ratio once on the gross headline and once on the adjusted figure from steps 2 to 5. The gap between the two answers is the quantity the headline was hiding.
7. Add external access separately. Record standing or committed swap and repo access as a distinct item outside the reserve stack, with its size, tenor and conditionality, rather than folding it into a reserves number.
8. Track the series, not the level. Repeat monthly. A stable headline with a rising one-month drain or encumbered share is exactly the configuration a single-number reading cannot detect.

The headline total is not wrong. It answers a narrow question — what the authority owns in reserve assets at market value on a reference date — and answers it accurately. Three further sections were bolted on after 1997 because that narrow question is rarely the one being asked when the number starts to matter.