A falling inflation rate is not always disinflation: Romania and the New York Fed, same week

Victoria Shinder11 October 2026, 09:393 min read

Two central-bank texts from the same week, one from Bucharest and one from New York, are about the same problem from opposite ends: how to read an inflation number that has just moved a lot.

Bucharest: a big fall that the bank does not trust. On 8 October the National Bank of Romania held its policy rate at 6.50%. Annual inflation had dropped from 10.42% in June to 6.17% in August — a fall of more than four points in two months. The bank's own explanation leaves little room for celebration: the drop came from base effects, as last summer's removal of the electricity price cap and increases in VAT and excise duties fell out of the twelve-month comparison. A year-on-year rate compares today's price level with last year's; when last year contained a one-off jump, the rate falls mechanically once that jump is twelve months old, whether or not prices today are rising slowly. And the BNR's outlook is that inflation will rise again until year-end, on fuel, energy and food prices, the drought, and the leu.

New York: why some price jumps can be ignored and others cannot. A day later, Kartik Athreya, the New York Fed's director of research, published a short essay on the Fed's view of prices. His two examples are a spike in the price of pumpkins — a tiny share of budgets that barely moves the average — and a persistent rise in fertilizer costs that pushes up all food prices, around 10% of the typical household budget. The first can be looked through. The second raises inflation at least temporarily, and the central bank's task is to stop it from changing what unaffected sellers expect, by being clear about its goal and setting the policy rate appropriately relative to the neutral rate.

Put together, the two texts give a usable reading rule. Romania's 2025 tax and energy measures were the policy equivalent of a one-off level shift: a large jump in specific prices on a specific date. Such a shift lifts the annual rate for exactly twelve months and then drops out. That is what the BNR describes — and also why a falling annual rate in that situation says little about underlying momentum. Athreya's framework points to what does matter: whether the shock spread to prices it did not directly touch. The BNR's figure for that is adjusted CORE2, which fell from 8.3% to 6.2% over the same months — also helped, the bank says, by base effects from last year's indirect-tax increases, but showing some softening from weaker consumer demand and lower food-commodity prices.

Where the two disagree in emphasis. The New York Fed essay is about keeping expectations anchored so that broad shocks fade; the BNR statement notes that short-term inflation expectations remain high and that new supply pressures — fuel, drought, the exchange rate — are arriving just as the base effects run out. That is the situation in which a central bank holds rather than cuts despite a headline rate that has fallen by about two-fifths.

The practical takeaway for anyone reading a monthly inflation release: before treating a big fall as news, check what happened twelve months earlier, and look at the measure that excludes the items the shock hit directly. Both banks, this week, did exactly that in public.

Sources: National Bank of Romania, decision of 8 October 2026 — https://www.bnr.ro/25929-2026-10-08-hotarari-ale-ca-al-bnr-pe-probleme-de-politica-monetara ; Federal Reserve Bank of New York, Liberty Street Economics, 9 October 2026 — https://libertystreeteconomics.newyorkfed.org/2026/10/prices-prices-prices-overall-inflation-and-the-costs-you-care-about/