What Pakistan’s Weekly SPI Inflation Actually Measures

Every Friday the Pakistan Bureau of Statistics publishes a number, and every Friday it is reported as though it were inflation. It is not. It is the Sensitive Price Indicator, and it measures something narrower and more useful than the headline treatment suggests.

For the week ended 20 August 2026, the SPI rose 0.49 per cent week-on-week, taking the annual rate to 9.66 per cent. Two weeks earlier it had risen 0.28 per cent. The week before that it fell 0.91 per cent. Those swings are not the economy changing direction three times in a month. They are what happens when you measure 51 items at weekly frequency.

What is in the basket

The SPI tracks 51 essential items. Prices are collected from 50 markets across 17 cities, with a base year of 2015–16 set at 100. The index sat at 358.60 points in early August, meaning the basket costs roughly three and a half times what it did a decade ago.

The basket is deliberately narrow and deliberately weighted towards things households buy constantly: wheat flour, rice, pulses, cooking oil, sugar, tea, milk, eggs, chicken, beef, mutton, onions, tomatoes, potatoes, bananas — alongside petrol, diesel, LPG, firewood, washing soap, cigarettes and a small number of clothing and footwear lines.

That composition explains the volatility. Vegetables move violently with weather and supply. Fuel now moves daily. When onions rise 14 per cent in a week, a 51-item index notices in a way a 356-item index does not.

The five indices nobody reports

This is the part worth building an article around, because it is in every PBS release and almost never in the coverage.

PBS does not publish one SPI. It publishes six: a combined figure and five separate indices for different household income bands — up to Rs17,732 a month, Rs17,733–22,888, Rs22,889–29,517, Rs29,518–44,175, and above Rs44,175.

They diverge, consistently and in a predictable direction. In the week ended 6 August, the combined index rose 0.28 per cent — but the lowest income group saw 0.43 per cent while the highest saw 0.20 per cent. In the week ended 30 July, when the index fell 0.91 per cent, the lowest group fell 1.19 per cent and the highest 0.68 per cent.

The reason is that food and fuel are a far larger share of a poor household’s spending. The poorest group feels roughly twice the movement of the richest, in both directions. When commentary says prices rose by such-and-such this week, it is quoting the combined figure, which describes nobody in particular and understates what happened to the households least able to absorb it.

Reading a release properly

Three habits make the weekly number far more informative.

First, look at the item spread, not just the index. In the week ended 20 August, 23 items rose — representing about 45 per cent of the basket by weight — while 10 fell and 18 were unchanged. A small index move driven by two violent items is a different situation from a small move with broad participation.

Second, separate the weekly noise from the annual trend. Week-on-week movement is mostly vegetables and fuel. The year-on-year figures are where structural pressure shows: in the week ended 20 August, onions were up 132.5 per cent against a year earlier, tomatoes 93.2 per cent, wheat flour 62.3 per cent, LPG 51.7 per cent, high-speed diesel 33.1 per cent and petrol 27.7 per cent.

Third, watch for offsetting moves. Petrol rose 3.77 per cent in that same week while diesel fell 5.15 per cent and eggs fell 5.78 per cent. Anyone reporting only the increases produced a materially misleading picture of the week.

SPI is not CPI

The two are routinely conflated and measure different things.

SPI compared with CPI

Frequency: SPI uses Weekly; CPI uses Monthly.

Items: SPI uses 51 essentials; CPI uses 356 items urban, 244 rural.

Coverage: SPI uses 50 markets in 17 cities — urban only; CPI uses 68 urban markets in 35 cities plus 27 rural centres.

Purpose: SPI uses Short-term barometer of essential-goods prices; CPI uses The official measure of consumer inflation.

Behaviour: SPI uses Volatile, food and fuel heavy, runs hot; CPI uses Smoother, broader, the number policy responds to.

The State Bank sets policy against CPI, not SPI. The SPI is an early-warning instrument — valuable precisely because it is narrow, because it arrives four times a month, and because it captures what a household notices at the shop before it reaches an official inflation print. Rural Pakistan is not in it at all, which is a real limitation and worth stating whenever the number is used.