Nigeria's National Bureau of Statistics published the July 2026 consumer price index on 17 August: headline inflation of 15.43 per cent year-on-year and 1.57 per cent month-on-month. Two tables in that report, headed “DIVISIONAL LEVEL” and labelled contribution year-on-year and contribution month-on-month, purport to show which of the thirteen spending divisions drove the number. Every one of their 26 cells is that division's fixed basket weight multiplied by the single headline rate. Food and non-alcoholic beverages carries a weight of 40.02, and 40.02 times 15.43 divided by 100 is 6.18, which is exactly what the table prints; restaurants and accommodation weigh 12.92 and the table gives 1.99; transport weighs 10.66 and the table gives 1.64. The ranking is therefore the weight ranking, and it carries no information about what any price did.
The same downloadable zip contains the workbook whose sheet holds the thirteen divisional indices — NBS ships the raw material for the real calculation alongside the table that does not use it. Weighting each division by its own index move and dividing by June's all-items level sums to 1.572 against a published 1.5722, and it tells a different story: food contributed 1.49 percentage points of the 1.57, not 0.63, while restaurants contributed 0.05 rather than 0.20. Clothing and footwear is the clean test. The July table credits it with pushing prices up by 0.08 of a point; NBS's own clothing index fell 1.43 per cent over the month, so its true contribution was minus 0.06. Across the 18 releases since the January 2025 rebasing, 16 are weight-scaled, and 415 of those 416 cells match the formula to the digit.
The bureau knows how to do it properly, because for two months it did. In the report published on 15 April for March data, and again on 15 May for April, the contributions tables are genuine decompositions: all 52 cells reconcile with the index workbook shipped in the same zip. No spreadsheet is needed to see the difference — the March table ranks health fourth and housing fifth, and the April table ranks education services third, because education prices rose 4.92 per cent that month. The weight vector cannot produce either order. Then the May edition, published 15 June, reverted to weight times headline, and so did June and July, none of the three carrying a note explaining the switch or the reversion. The tables are read as findings regardless: on 18 August Investors King reported that four divisions made up “about 72%” of the headline rate, and those four weights sum to 72.02. The August report is due around 15 September, before the Central Bank of Nigeria's Monetary Policy Committee meets, and will show which of the bureau's two methods is in use.