JUDGMENT判断On October 13, in Bangkok, the IMF raised its global growth forecast by 0.1 percentage points. Another line in the same report reads: the economies accounting for 95 percent of global growth have fallen from 54 in April to 32. One number revised up, one list revised down, both changes landing on the same table. Most of the 22 economies that dropped out over six months were not downgraded. They simply stopped being large enough to appear in that row.

FACT事实The report is titled Resilience Under Strain, Urgent Choices. Its central forecast: global growth slowing from 3.5 percent in 2025 to 3.1 percent in 2026, then back to 3.4 percent in 2027; advanced economies at 2.1 percent in 2026, emerging market and developing economies at 3.7 percent. Inflation runs the other way: 4.1 percent in 2025, rising to 4.7 percent in 2026, easing to 4.1 percent in 2027. Growth concentrates along two axes — energy exporters outside the Middle East conflict, and economies embedded in artificial intelligence investment. China and the United States together account for more than 75 percent of net imports of AI-related computing hardware in 2025 and 57 percent of new data centres built since 2024. Another set of books landed the same week. The Bureau of Labor Statistics published September non-farm payrolls on October 2: 29,000 jobs added, under a third of the market expectation; July and August revised down by 60,000 combined, July itself flipped from +21,000 to −10,000. After revisions, the past three months average roughly 17,000 a month. Average hourly earnings came in at $37.81, up five cents; the average workweek held at 34.4 hours, unchanged.

SPECULATION推演Two documents of different kinds, one accounting method: protect the aggregate first, then adjust its members. The IMF kept 3.1 percent and paid for it with a list narrowed to 32 economies. The BLS kept the positive print of 29,000 and paid for it by rewriting two months it had already published. This is why the macro reading keeps failing to explain the micro experience — the total did not break; the constituents changed. IMF chief economist Silvana Tenreyro attributes the divergence to three things: exposure to commodity prices, participation in AI activity, and borrowing costs. All three sit on the structural ledger. None of them turns over next quarter.

OPINION观点One line in the report is easy to skip: the estimates and projections in Chapter One "are based on statistical information available as of October 1, 2026" — twelve days before publication. The BLS release covering September, carrying the negative 60,000 in revisions, arrived on October 2. Inside the global forecast treated as the market's freshest judgment, the American labour market still looks the way it did before revisions. Read any macro forecast by turning first to its cut-off date, then to its conclusion. The second question belongs to the concentration row: are the remaining hundred-plus economies subtracting, contributing nothing, or merely being diluted? The IMF does not distinguish, and this measure cannot. Concentration is not recession. What it does say is this: when economies outside the list deteriorate, nobody is keeping their books.

---

Further Reading