The jobs forecast almost nobody quotes in full: 170m up, 92m down

The three numbers are one sentence, and headlines take a third of it. The report states creation, displacement and the net in a single passage. Any one of them quoted alone is arithmetically true and directionally misleading, which is why the same report gets cited as proof that automation destroys jobs and as proof that it creates them.

Churn and growth answer different questions, and 22% is not a fourth opinion. Adding creation and destruction gives the share of today's jobs that changes in some way, which is a measure of disruption. Netting them gives the change in how many jobs exist. A workforce can be violently reshuffled and still grow, and this forecast says it will be both.

This is a survey of employers, not an observation of a labour market. The figures are what companies expect to do over five years, collected before the period began. That makes them a statement of intent from people who may be wrong about their own hiring, and it is a different class of evidence from the payroll counts and occupational projections these numbers are often set beside.

Cite the pair and the net together: 170 million created, 92 million displaced, 78 million net, per the WEF Future of Jobs Report 2025. Dropping the third number is where most misuse starts.

Say it is an employer survey covering 2025 to 2030. That tells a reader it is a forecast of intent rather than a measurement, which is the caveat that matters most here.

Keep churn separate from growth. Use 22% when your point is how much the labour market reshuffles and 7% when your point is whether it grows. They are not competing estimates of the same thing.

Do not set these figures against Frey and Osborne's 47% or the OECD's 14% as if all three answered one question. Those score how automatable jobs are; this counts jobs employers expect to add and cut.

Quoting 92 million displaced without the 170 million created. It is half of a sentence presented as a finding.

Reading 22% as a fourth forecast. It is creation and destruction summed, a churn measure, not an alternative to the 7% net.

Treating the numbers as a measurement of automation. Technology is one of five macrotrends in the report, alongside the green transition, geoeconomic fragmentation, economic uncertainty and demographic shifts. The totals are not attributable to automation alone.

Presenting a survey of employer expectations as an outcome. Nothing here has been observed yet; the period it describes had barely started when the report was published.

Every figure on this page was located in the WEF report PDF itself rather than in coverage of it, which matters for this particular statistic because the retelling is where the third number goes missing.

We quote the report's own arithmetic and do not recompute it. Creation, displacement, the net and the churn share are all stated in its text, so nothing here is our subtraction.

Confidence is High across the table because each figure is printed plainly in the source and is current for its vintage. High describes how safely you can quote the number as what the report says, not how likely the forecast is to come true.

We deliberately do not convert these into an annual rate. The report gives a five-year total, and dividing by five would invent a smoothness the survey does not claim.