Macro Markets and Machines_The Economic and Market Transformation Driven by AI_GWM report - Flipbook - Page 26
Macro, Markets, and Machines
November 2025
Section 4: Macroeconomics.EXE: Reprogramming Growth for
the AI Age
Walid Khalid, CFA | Director, Investment Strategy, Scotiabank Global Wealth
Juan Manuel Herrera | Senior Economist, Scotiabank Global Wealth Management
•
Productivity improvements will be the main channel through which AI affects the economy, facilitating
greater output for the same unit of labour or capital. However, economists are widely split in their estimates
for just how large this positive productivity shock may be, with uncertain impacts on GDP growth, inflation,
and monetary policy.
•
The impact of AI on aggregate productivity will show through four key channels: automation (AI replacing
workers), complementarities (AI augmenting the output of workers), deepening of automation (AI improving
the output of automated tasks), and new labour-intensive products or tasks (AI-related displacements see
workers migrate to, and boost the output of, other sectors, or the adoption of the technology creates new
economic activities).
•
Our review of existing literature on the possible impacts of AI shows that there is no clear consensus
regarding the magnitude of its productivity benefits – and, therefore, on the size of would-be GDP gains. The
IMF estimates that global productivity growth in a fast AI adoption and diffusion scenario will result in global
productivity gains of 2.4 p.p. cumulatively over 10 years, or about 0.25 p.p. per annum relative to baseline. In
comparison, the BIS considers a 1.5 p.p. boost per year, while MIT’s Daron Acemoglu places his estimates at
only around 0.1 p.p. per year.
•
AI exposure, preparedness, and access are expected to result in important cross-country differences in
would-be benefits from AI. A higher share of workers or GDP exposed to AI would mean a greater share of the
economy that could see productivity boosts (though also higher displacement risks). Stronger infrastructure
and innovation-enhancing policy foundations, or preparedness, would allow a faster and fuller harnessing of
AI’s potential. Finally, uneven access to AI equipment or IP due to geopolitical issues, for instance, could
widen the gap between leaders and laggards.
•
Existing research points to a net neutral or positive impact of AI on inflation, as productivity gains that may
result in excess supply (deflationary) are offset by income and thus aggregate demand gains that pressure
prices higher. However, inflation and monetary policy outcomes will heavily depend on whether AI will, on
net, ultimately complement or displace workers. It is too early to tell how AI will affect labour markets. Early
employment data for young workers in AI-exposed occupations show that AI is substituting labour for, at
least, a narrow subset of the workforce.
The key channel by which AI is expected to lift economic output is through improvements to productivity,
facilitating greater volumes of production for the same unit of labour or capital. The technology has the potential
to automate or hasten processes that complement – and in some cases supplement – the work of employees,
allowing workers to do more under identical time constraints, while also deepening their specialization and
productivity in other aspects of their roles that may not be automatable or that require a “human touch.”
Throughout this section, we will focus on the results and conclusions of publicly available research from
academic and multinational organizations.
As is often the case at the onset of revolutionary technologies, there are widely varying expectations for the GDPenhancing prospects of AI this early in its cycle. Estimating the possible productivity and economic impact of AI
requires a myriad of assumptions that can materially affect the final result, such as how quickly firms and workers
may integrate it into their tasks or the pace of capital investment in these technologies. In terms of overall
economic activity, it also requires important assumptions on how it complements or replaces human labour, its
impacts on wages, and labour matching frictions over the adjustment period, which all in turn affect whether it is
demand-neutral, -positive, or -negative. Estimates on the benefits of AI also vary across countries, although existing
research generally agrees that developed nations stand to benefit the most (more on this later).
Scotia Wealth Management
25
