IMF Eliminates 'Unallocated' Category to Provide 100% Currency Composition of Global FX Reserves

When analysts debate the decline of the U.S. dollar, they often point to precise decimal points in global reserve data, assuming central banks publicly declare every currency they hold. The reality is far messier. For decades, the International Monetary Fund’s official tally contained a massive black box—a category of "unallocated" reserves that at one point obscured $5.5 trillion in global wealth. Now, that black box has been mathematically dismantled.[3]

In a major methodological update, the IMF has eliminated the "unallocated" category from its Currency Composition of Official Foreign Exchange Reserves (COFER) dataset. For the first time, the dataset now provides a 100% currency composition for the world's $13.1 trillion in foreign exchange reserves.[1][2]

The change relies on advanced statistical imputation to fill the gaps left by countries that do not confidentially report their exact holdings to the IMF. By using a combination of stratified mean imputation and carry-forward imputation, IMF statisticians have retroactively revised the entire dataset back to the first quarter of 2000.[1]

To understand the scale of this revision, one must look back to the peak of the data gap. In 2013, a staggering 47% of global foreign exchange reserves were classified as unallocated. This meant the IMF simply did not know the currency breakdown for nearly half of the world's official reserve wealth.

The massive blind spot was primarily driven by a handful of large reserve holders. Historically, advanced economies reported their holdings reliably, but major emerging economies—most notably China and Taiwan—either did not report their composition or were slowly phasing their data into the system to protect state confidentiality.

Over the past decade, the unallocated portion shrank organically as more countries began confidentially reporting their data to the IMF. By late 2025, the unallocated share had dropped to roughly 7% of global reserves.[1]

With the gap narrowed to a single digit, the IMF determined that mathematical imputation was now statistically robust enough to estimate the remainder without distorting the global aggregates. The new methodology assigns the missing 7% based on the known behaviors of similar economies and the past behaviors of the non-reporting countries themselves.[1]

The newly complete data provides a clearer picture of the global financial hierarchy. According to the Q1 2026 release, the U.S. dollar accounts for 57.13% of all global reserves, followed by the euro at 20.03% and the Chinese renminbi at 1.99%.[2]

However, the evidence still has limits. Because the final 7% is imputed rather than directly reported, the exact decimal-point precision of the global aggregates remains an estimate. Furthermore, the COFER dataset only tracks official central bank reserves, excluding the massive foreign currency holdings managed by sovereign wealth funds.[1][3]

Despite these limitations, eliminating the unallocated category removes a major source of analytical distortion. Previously, researchers tracking de-dollarization had to guess whether changes in the dollar's share were due to actual central bank selling or simply the mechanical effect of a large country moving in or out of the reporting pool.[1]

The update also strengthens the confidentiality of individual countries. By publishing a fully allocated 100% aggregate, the IMF eliminates the "residual disclosure risk" that occurred when a single large country's reporting status changed, which previously allowed analysts to reverse-engineer that specific nation's holdings.[1]

Ultimately, the methodological shift transforms COFER from a partial survey into a comprehensive global baseline. While the exact contents of a few central bank vaults remain state secrets, the global financial system finally has a complete map of its reserve wealth.[3]