The Pakistan Economic Survey 2025–26 puts employment at 77.2 million people in 2024–25, compared with 67.25 million in 2020–21. The figures look like a straightforward comparison. They are not. Pakistan changed how it defines employment between the two surveys. Apply the earlier definition to the 2024–25 data and the estimate is 79.7 million. Both recent figures are correct, but they answer different questions.
The Pakistan Bureau of Statistics explains the change in its 2024–25 Labour Force Survey, which reports these counts for people aged ten and over. Under the earlier definition, the estimate included about 2.48 million people who produced agricultural goods for their own households. Under the newer international standard, employment measures work for pay or profit, and production for a household’s own use is measured separately. Someone growing food primarily to feed the family still works. The change is where that work appears in the statistics. It does not show that 2.48 million people lost jobs.
That distinction matters well beyond the release of a survey. If an official wants to know how employment changed since 2020–21, the defensible comparison is 67.25 million then and about 79.7 million in 2024–25, using the earlier definition for both years. Even that comparison shows a change in the number of people classified as employed, not a count of jobs created by a particular policy. If the question is how many people fell within the newer definition in 2024–25, the answer is 77.2 million. There is no need to choose one number for both purposes.
The unemployment rate is an equally useful test. The Planning Ministry’s release reports 6.3 per cent in 2020–21 and 6.9 per cent in 2024–25 under the old definition. For the same 2024–25 survey, the rate is 7.1 per cent under the new definition. The bureau reports about 5.9 million unemployed people in 2024–25 under either measure. The rate is higher under the new definition because the labour force used to calculate it is smaller, not because the survey found more unemployed people under that definition.
The effect is especially worth explaining when reporting on agriculture. A smaller share of employment classified as agricultural under the new standard tells us something about the composition of paid or profit-oriented work. By itself, it cannot tell us that families stopped growing food for themselves or that workers moved to industry or services. To make either claim, we need evidence on changes over time under the same definition.
Readers should not have to reconstruct these distinctions from a methodological note. The bureau can lead its next release with the latest figure and explain, in the next sentence, which earlier figure it can be compared with. It can do the same for unemployment. A short note next to the first use of each number should make the survey period and definition clear. The Finance Ministry can carry that explanation into the Economic Survey rather than leaving readers to join figures drawn from different measures.
Statistical agencies should keep the old figures accessible while they build a consistent series using the new measure. That does not mean preserving two competing headlines indefinitely. It means making the transition clear in the places where people actually use the figures: the executive summary, the public release and the downloadable data. If an older estimate cannot be reconstructed on the new basis, say so. A gap is less misleading than a trend that treats two different definitions as one.
The same discipline matters when evaluating a jobs programme. A training provider can report how many people attended. A department can report how many later entered paid work, provided it checks rather than assumes the transition. The national survey can show what happened in the labour market. None of those measures, on its own, shows that a programme produced the national change.
Separating work for household use also gives Pakistan an opportunity to make that work more visible. Provincial departments planning agricultural services need to know who produces food for their households, regardless of whether that work generates income. A team assessing a programme that helps people enter paid employment, however, needs a different measure. A short profile of own-use production, disaggregated by sex and province where the sample permits, can serve the first purpose without distorting the second. Neither measure should be used as a shortcut for estimating household income or needs.
Editors can help now. When a headline says employment rose, it should say which years are being compared and use the same definition for both. When a report cites the newer employment figure, it should explain what happened to the own-use category. And when an unemployment rate moves, check whether the number of people without work changed or whether the calculation changed. Pakistan’s survey already gives readers the evidence to make those distinctions. The next step is to put that explanation where the figures are first reported.