faultlinesweekly

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 ...

A Sign Change Is Not a State Change The spread between the 10-year and 2-year Treasury yields stood at +0.50 percentage points on 20 August 2026 — the 10-year at 4.69%, the 2-year at 4.19%, per the Federal Reserve's H.15 release. That is a normally sloped curve, and every monthly average from August 2025 onward is positive as well. Somewhere between April 2024, when the monthly average sat at −0.33 points , and September 2024, when it printed +0.10 , the sign flipped. The narrative attached to that fact runs like this: the inversion was the warning, the un-inversion is the all-clear, and the recession the inversion advertised either arrived quietly or was called off. The problem with that reading is not that it is optimistic. The problem is that it treats one number as one piece of information. The spread is a difference between two independently traded prices, and two prices moving in opposite directions, or in the same direction at different speeds, can produce identical spr...

A headline reporting that foreign central banks are "dumping" U.S. Treasuries almost always rests on one monthly number pulled from one table. The number is usually real. What it measures, how late it arrives, whether it describes reserve managers at all, and whether anything was actually sold are four separate questions — and they are answered by three different published series that frequently disagree with each other. As of mid-August 2026, all three point in the same direction for the first time in a while. Official holdings are down on the Treasury's own survey, down on the Federal Reserve's weekly custody line, and down over a multi-year window. That convergence is worth taking seriously. It is also the exact circumstance in which the reasons matter far more than the direction, because at least four unrelated mechanisms produce an identical-looking decline. Foreign-Held U.S. Treasuries TIC survey, June 2026 · total $9,299.0 billion Offic...