The 2026 growth forecast was upgraded on AI capital expenditure. In the same quarter, resident employment growth slowed to 2,200 and retrenchments rose to 4,620. The labour market is not contracting — the composition of hiring is what changed.
On 11 August, the Ministry of Trade and Industry upgraded its 2026 GDP growth forecast to “4.5 to 5.5%”, from “2.0 to 4.0%”. The ministry cited the better-than-expected performance of the economy in the first half, and an improved outlook for the rest of the year. Its stated reason was the acceleration in global AI-related capital expenditure.[1] The economy grew 5.9% on a year-on-year basis in the second quarter, easing from 6.3% in the first. The first half ran at 6.1%.[1]
Six weeks later, the Ministry of Manpower published the labour market report for the same quarter.[2] Total employment rose by 11,400. That is the nineteenth consecutive quarter of growth, and a larger increase than either of the two preceding quarters. Unemployment stayed low in June: 1.9% overall and 2.9% for residents.[2]
On the headline figures, neither jobless growth nor a labour market slump is visible. The ministry’s own conclusion is more specific, and it is where this memo starts. The labour market “remained resilient”, but “conditions became less favourable for some resident workers, as resident employment growth moderated, retrenchments increased and six-month re-entry outcomes weakened”.[2]
The aggregate conceals a split. Of the 11,400 net additions in the second quarter, 9,200 were non-resident, driven by Work Permit Holders in construction and manufacturing. Resident employment grew by 2,200, down from 5,400 in the first quarter.[2]
Source: Ministry of Manpower, Labour Market Report, second quarter 2026.
Resident employment is still growing. This is a moderation, not a decline. The non-resident increase is concentrated in sectors with long-standing structural labour demand, not in the sectors where AI adoption is highest.[2][3] What the split changes is who the increment reaches. For the quarter, most of the net workers added to the Singapore economy were not residents. The quarter’s growth was attributed to AI-related capital expenditure rather than to domestic hiring.[1][2]
Resident employment is the channel connecting GDP growth to household income. That is why this series matters more than the total. A quarter in which it slows while the total accelerates is a compositional signal — and it is the number the ministry itself flagged.[2]
Retrenchments rose to 4,620 in the second quarter, or 2.0 retrenched per 1,000 employees, from 3,830, or 1.6 per 1,000, in the first. The increase was driven by business reorganisation and restructuring in outward-oriented sectors: manufacturing, information and communications, and financial services.[2]
The consequence appears in re-entry rather than in unemployment. The share of retrenched residents who found work within six months fell from 60.7% to 54.9%. The twelve-month re-entry rate stayed broadly stable, from 69.4% to 69.8%.[2] Read together, the two rates indicate that retrenched workers are taking longer to be re-employed, not leaving the labour force: the pipeline is slower, not broken.
The market also remains tight by vacancy. There were 1.48 vacancies per unemployed person in June, similar to 1.46 in March, even as job vacancies eased from 73,300 to 68,600. The fall was mainly in PMET vacancies in financial services and information and communications. Entry-level PMET vacancies remained sizeable, accounting for 45.3% of all job vacancies.[2]
The most direct evidence on automation’s workforce effects is not a forecast but the ministry’s inaugural report on AI adoption among firms, released on 30 April.[3] Its findings complicate the displacement story in both directions.
The report’s own conclusion is that “there is no indication of significant job displacement at this point”, with the caveat that adoption is uneven and that smaller firms risk falling further behind.[3] The sectors shedding workers in the third quarter overlap with the sectors adopting fastest. That is a correlation across two datasets published five months apart, not a causal finding. No published Singapore series attributes retrenchments to AI substitution. The reason the ministry records remains business reorganisation and restructuring.[2]
The argument that AI breaks the historical pattern rests on a specific mechanism, and it is worth stating precisely because it is the part that would change policy if it held.
Earlier general-purpose technologies expanded employment through a reinstatement effect. Automating a task freed workers for newly invented tasks. That is why employment kept rising through computing and electrification.
Professor Will Cong of Nanyang Business School argues that AI agents differ in kind. He calls them “capital-manufactured labour”, able to absorb the new tasks as readily as the old ones. A firm can then add productive capacity by deploying more agents rather than hiring more people. The expansion appears in the national accounts as software and compute investment, not as wages.[4]
Associate Professor Walter Theseira of the Singapore University of Social Sciences puts the distributional version of that argument. The concern, he says, is not jobless growth in the aggregate. It is growth in which employment still rises while the middle of the job distribution thins, and in which capital investment becomes self-sustaining, “no longer needing complementary labour”.[4] Cited IMF research points the same way. It suggests AI could raise wealth inequality through higher returns to capital rather than through wages.[4]
Three considerations argue against reading this quarter as the arrival of AI-driven displacement.
The labour market is not slack. Unemployment at 1.9%, 1.48 vacancies per unemployed person, and nineteen consecutive quarters of employment growth are not the profile of an economy growing without its workforce. Jobless growth describes output rising while employment does not. That is not this economy.[2]
The resident slowdown has a rival explanation that has nothing to do with AI. Singapore has run a structurally tight labour market for a decade against an ageing population and a falling fertility rate. Low resident employment growth in a quarter can reflect the size of the resident labour pool rather than substitution. The growth upgrade itself rests on capital expenditure, not on domestic hiring.[1]
Firms report augmentation. 6.2% reported reduced headcount after adopting AI, against 18.9% redesigning roles and 13.9% creating AI-related jobs. Were automation substituting for labour at scale, that distribution would look different.[3]
The bearish case worth taking seriously is therefore narrower than “jobless growth”. Adoption is early, at 3.8% core integration. Displacement, if it comes, would appear first in the sectors already restructuring. It would show in re-entry times and in the composition of hiring before it showed in the unemployment rate.[2][3]
The government’s position is explicit and predates the quarter. The Prime Minister stated in February 2026 that Singapore “will not have jobless growth”, and Parliament unanimously backed a motion affirming that position in May.[4]
Responding to queries, the Manpower and Trade and Industry ministries pointed to improved hiring sentiment: the share of firms intending to hire in the next three months rose from 43.9% in June to 48.7% in July. Both ministries acknowledged that job searches have taken longer and that openings have dipped.[2][4]
The instruments are already funded:
Wage expectations are the softer signal. The share of firms intending to raise wages eased from 29.3% in June to 27.9% in July. Both hiring and wage expectations remain below their February 2026 levels.[2]
This analysis is based on publicly available data as of 2026-09-26. For related coverage, see The Capital Share and The Back Office Restructuring.