Hong Kong’s Graduate Vacancies Have Fallen From 80,000 to 31,000 in Three Years

Hong Kong's unemployment rate is 3.7%. Its graduate job market has lost roughly 60% of its openings since 2022.

Figures from Hong Kong's Joint Institution Job Information System show the number of full-time vacancies suitable for university graduates fell from around 80,000 in 2022 to around 31,000 in 2025. Within that, one sector stands out: information and communications vacancies listed on university platforms have declined 76.7% since 2021 — the field graduates were told to enter.

That contraction is happening in an economy that grew 5.1% in the first half of this year and where the headline unemployment rate has been improving. The two facts are not in conflict. They describe different parts of the labour market, and only one of them appears in the monthly statistics.

Adoption is high, and its shape is unusual

Hong Kong workers have taken up AI tools faster than almost anywhere.

A study by HKUST Business School conducted between March and April 2026, drawing 3,722 valid responses and described as one of the largest recent surveys on the subject in the city, found more than 70% of working professionals frequently use AI tools in their jobs — a rate significantly above the global average. Many respondents nonetheless remained uncertain about how to reskill and stay competitive.

McKinsey's February 2026 survey found a similar headline figure alongside a revealing distribution: around 70% of white-collar workers had adopted AI tools, 88% reported productivity improvements and more than 90% engaged daily — but fewer than 25% deployed AI across complete workflows, and only 14% of executives were frequent users, four to five times lower than the rest of their organisations.

That last figure matters for understanding where the pressure lands. If usage is concentrated among junior and mid-level staff performing discrete tasks rather than among leadership redesigning processes, the near-term effect is task substitution at the bottom of the organisation rather than transformation across it.

Employers are hiring for revenue, not for training

The corporate side of this is visible in KPMG's Hong Kong Employment Outlook 2026, based on a survey of 281 executives and professionals conducted in January.

It found 24% of organisations widely deploying AI in 2026, three times the 2025 figure, with understanding and applying AI now ranked the most crucial employee skill. On headcount, 22% of respondents expected decreases — a significant rise on previous years — while only 19% expected increases, the lowest in recent years and a figure that has continued to fall.

The composition of that pessimism is the notable part. For the first time since 2020, C-suite respondents were more likely than the broader workforce to anticipate headcount reductions, at 29%. The people with the clearest view of deployment plans were the most cautious.

Recruiters describe the same pattern from the market side. Employers are prioritising immediate productivity, making the traditional graduate training pathway more difficult to sustain, with entry-level and generalist roles being absorbed by automation while competition intensifies for specialists. Robert Walters found employers identifying administration and business support, IT and digital transformation, and accounting and finance as the functions most at risk — which describes a substantial share of what graduate schemes have historically recruited into.

Not a uniform contraction

The market is bifurcating rather than shrinking evenly, and the divergence is wide.

ManpowerGroup's third-quarter survey placed Hong Kong's net employment outlook at -9%, among the weakest globally. But finance and insurance posted a positive employment outlook of 33%, with particular demand in fintech and compliance roles.

PwC's 2026 Global AI Jobs Barometer found AI job postings in Hong Kong rose 50% year on year from 2024 to 2025, lifting their share of total postings from 3.6% to 4.7% — still below the 2023 peak, but indicating renewed demand for AI skills. The same research argues companies making the largest AI-driven productivity gains have been raising wages and headcount faster than firms least exposed to the technology.

So the picture is not mass displacement. It is a labour market splitting into roles that command premiums and roles being absorbed, with the entry-level rungs of the ladder thinning fastest.

Underneath it sits capability that keeps advancing, visible in the finances of the firms producing it — OpenAI's annualised revenue has roughly doubled to top $40 billion, driven substantially by AI products aimed at exactly the tasks entry-level staff have traditionally performed.

What to watch

The graduate intake figures for this year are the cleanest available signal. If vacancies stabilise near 31,000, the adjustment may be complete; if they continue falling, the pipeline problem compounds, because a cohort that does not get trained does not become the mid-level workforce five years out.

The second is whether reskilling provision closes. Across surveys, Hong Kong workers report lower access to organisational learning opportunities than their global peers while expressing higher concern about AI's impact on their roles.

The third is the monthly labour statistics themselves. A 3.7% unemployment rate is consistent with a market where displaced and never-hired workers exit the labour force rather than register as unemployed — which is precisely what happened in the first quarter, when the rate held steady while total employment fell.