By Thomas Abrams, Head of Human Rights, Social and Governance Issues
The scale and pace of AI’s impact on employment is being hotly debated. However the results play out, investors are exposed to this transition – and they can start to mitigate the risks with early engagement.
The WEF’s Future of Jobs Report 2025 projects that AI adoption will create 78 million new jobs - but not before 92m roles are displaced globally by 2030. We have seen big impacts on the job market before: for example, when China joined the World Trade Organisation in 2001, an estimated 2.4m manufacturing jobs were displaced over roughly a decade. This time, impacts of AI job displacement may not only be much wider, but, especially with the recent acceleration of agentic AI, they may occur much more quickly.
Portfolio-level risks from job displacement
Previous waves of displacement, from manufacturing to routine service work, have been concentrated in lower and middle-income blue collar occupations.
AI is different: it can increasingly substitute for the types of cognitive work common in finance, legal, software, sciences and engineering, exposing higher-earning workers whose spending and borrowing patterns sit closer to the centre of financial system stability. Analysis of Federal Reserve data finds that the top 20% of earners account for 59% of all US consumer spending. Even if displacement is closer to the lower ends of estimates, it could trigger cascading impacts on the economy, affecting consumption, credit quality, real estate and the tax base. One study shows that even a 2% decline in white collar employment could translate to a 3-4% hit to discretionary consumer spending.
This could negatively impact consumer-facing equities, as well as the discretionary services that lower-income workers often provide. Mortgage defaults among urban homeowners can affect real estate values and mortgage-backed securities. Sovereign credit could come under pressure as tax bases narrow and demand for social spending rises in parallel.
Private credit markets are already showing strain from AI disruption – UBS estimates 25-35% of the market is exposed , with default rates potentially rising to 13% in a severe scenario. These channels reinforce one another: corporate earnings fall, credit stress deepens, and deficits widen as the tax base shrinks. Given record high asset valuations, it appears few of these dynamics are yet priced in.
Company-level risks are prevalent
AI-driven workforce displacement also creates clear risks at the company level. Companies pursuing efficiencies from headcount reductions, rather than working to identify opportunities for augmentation (where AI complements rather than replaces existing human labour), redeployment and reskilling, may benefit from near-term cost reductions and a share price boost, but risk compromising the longer-term resilience of their workforce .
As fewer younger workers enter their talent pipeline, they will eventually be left without ‘home-grown’ managers and experienced staff. Institutional knowledge, customer relationships, operational expertise and workforce trust that have taken years to build will be difficult to recover.
Companies pursuing this approach may also face heightened exposure to regulatory and reputational risk as scrutiny of AI deployment intensifies, including the prospect of retrospective measures targeting firms seen to have profited excessively from headcount reduction. By contrast, companies investing in integrating AI alongside augmentation and reskilling may be positioning themselves more durably – as well as signalling they expect to deliver more.
Few companies or investors are managing AI’s workforce risks yet
The Thomson Reuters Foundation’s AI Company Data Initiative finds that out of almost 3,000 companies, only 14% evidence policies to mitigate AI’s negative impacts on workers. Fewer than a third offer reskilling or retraining programmes, and of those that do, only 8% have programmes with quantified participation or impact metrics – the rest are largely ad-hoc, leadership-only, or unmeasured.
While investor and regulatory attention on AI’s impacts is increasing, it is still relatively concentrated on oversight of technology developers, data privacy, and cybersecurity. How risks associated with AI’s workforce impacts can be managed is still being developed.
Principles and guidance from the OECD and UNESCO provide important normative grounding, but these guidelines need translating into actionable practice. Some leading investors are building internal approaches from scratch: preventing fragmentation and ensuring effective stewardship will require building shared expectations and the collective pressure that makes engagement most effective.
The cost of continued inaction
As with the climate transition, the critical question is not whether disruption occurs but whether it is orderly or disorderly.
In an orderly transition, companies would favour augmentation, redeployment and reskilling, while governments design fiscal responses ready to scale up as displacement occurs. Proposals such as universal basic income, automatic and portable benefits, and transition funds already exist – but policy maker appetite to implement them remains limited. This scenario would reduce uncertainty for investors and support stability in consumer-facing and credit markets.
A disorderly transition might see governments resort to blunter instruments, such as fast-moving regulation, or retrospective windfall taxes on companies that are seen to have profited. Reactive legislation creates significant uncertainty and transition risk for investors.
What investors can do
To manage the transition risk, investors can engage companies and governments on two fronts: mitigating avoidable displacement through augmentation, redeployment and reskilling, and adapting to unavoidable displacement.
Company-level stewardship matters, but on its own may not be enough. If AI delivers the productivity gains many expect, aggregate demand for labour could fall regardless of how well individual companies handle the transition. This is why systemic stewardship is equally important. By engaging policy makers, investors can signal that considered, pre-emptive action – rather than reactive regulation – would help support the resilience of financial markets .
| Mitigation | Adaptation | |
|---|---|---|
|
Engage with companies to: |
|
|
|
Engage with governments to: |
|
|
Several tools can help investors begin engaging now , including the OECD’s Due Diligence Guidance for Responsible AI, which provides a framework for how companies should identify, prevent, mitigate and communicate AI-related impacts on workers, and the AICDI’s investor engagement checklist, which proposes key questions to ask corporates regarding workforce safeguards, reskilling and oversight.
What the PRI is doing
The PRI is working to help signatories move from general awareness of AI risks into structured action. This includes building on existing governance frameworks and how to manage risks from sustainability issues such as labour, human rights and climate.
We are convening signatories to share emerging practice and increase alignment on investor expectations of companies. We are also working with experts to signpost and co-develop practical stewardship tools to support signatories in engaging companies that deploy AI.
We want to hear from signatories that are interested in getting involved, whether by joining upcoming convenings, potentially working with us to develop tools, or sharing their own practice. Get in touch.
The PRI blog aims to contribute to the debate around topical responsible investment issues. It should not be construed as advice, nor relied upon. The blog is written by PRI staff members and occasionally guest contributors. Blog authors write in their individual capacity – posts do not necessarily represent a PRI view. The inclusion of examples or case studies does not constitute an endorsement by PRI Association or PRI signatories.
