- Macro Strategist
Luxembourg, Institutional
ChangeThe views expressed are those of the author at the time of writing. Other teams may hold different views and make different investment decisions. The value of your investment may become worth more or less than at the time of original investment. While any third-party data used is considered reliable, its accuracy is not guaranteed. For professional, institutional, or accredited investors only.
In my view, advances in artificial intelligence (AI) could alter the macro environment meaningfully by raising both trend growth and expectations of long-term real interest rates, but by how much and when is far less clear. Below I set out an initial framework to help answer those questions within the context of deglobalisation and demographic change.
How could AI transform the macro backdrop?
Generative AI technologies offer the promise of human-like output, with high usability thanks to natural language and a wide range of potential applications.
If successfully implemented, AI-driven automation could boost productivity growth by improving efficiency and freeing up resources for more productive tasks. Such an outcome would be welcome in a world where one of the most prevalent market themes over the past decade has been the fear of secular stagnation, where trend growth and R* (the real interest rate when economies operate at full potential) so low that interest rates cannot fall far enough to stimulate investment.
How big could the macro impact be?
The lack of data makes it incredibly hard to predict the likely macro impact of AI. Some academic studies (Figure 1) have attempted to do so based on a bottom-up breakdown of automation potential by sector and the speed of adoption of past technological advances. Understandably, the range of estimates about the potential boost to productivity are large and depend on assumptions about the level of task automation, associated structural worker displacement and timeline of adoption.
In summary, these studies estimate that productivity growth could be raised by anywhere between 0.5% and 7%. If we take the average across the seven studies, the estimated boost to productivity is as much as 2.5%, while potential trend growth — inferred from the share of labour in production — could increase between 0.1% and 2%, with an average of 1%. Those estimates are very large.
Figure 1
What does this mean for bond yields?
Since the global financial crisis (GFC), we’ve seen a steep decline in real bond yields, mostly because of two forces:
- A big squeeze in risk premia, which dropped from an average range of +100 to +200 basis points (bps) before 2008 to -100 to 0 bps for most the ensuing period.
- A sharp fall in estimates of R* from 200 bps before 2008 to between -100 and +100 bps afterwards.
R* over time is highly correlated with the trend in productivity and real GDP growth (Figure 2), so it’s not surprising that the drop in estimates of R* coincided with the steep decrease in productivity post-GFC. Investors believed that R* in developed economies had fallen to zero because of the faltering supply and demand picture driven by factors such as a lack of innovation, ageing populations, falling education standards and unequal income distribution. As a result, central banks struggled to get interest rates down far enough to stimulate spending and investment, which prompted discussions about removing the lower bound on rates and, in some instances, rate cuts into negative territory.
Figure 2
If AI could reverse this downward trend in productivity, the level of interest rates that would have been enough to slow the economy over the past 15 years may no longer be sufficient. Higher rates may go hand in hand with increased valuations of risks assets as these rate rises mostly reflect higher estimates for long-term return on capital. Indeed, there is tentative evidence that markets may already have started to price some expectation that AI will raise productivity and trend growth.
What is the counternarrative?
We still have little clarity on the speed of transition, the degree of automation possible and the proportion of jobs that are likely to be displaced. In my view, the two most critical questions that still need answering are:
- Where is the investment?
Higher productivity growth is usually correlated with higher capital expenditure (capex). Yet, since the global economy reopened after the COVID pandemic-related lockdowns, global capex has been disappointing in aggregate, particularly outside of the US, with global capex intentions even starting to trend down recently. - How will governments respond?
AI is likely to bring significant job displacement — a Goldman Sachs report estimated that 300 million full-time jobs could be displaced worldwide over a 10-year period. 1 Historically, innovation has always created new sectors and new jobs. For instance, a significant proportion of the jobs that exist today didn’t exist 50 years ago. However, the accompanying uncertainty represents a major challenge for governments. The correct response from a macro point of view is to focus government policy primarily on reskilling workers to allow them to adapt rapidly and shift to more productive outputs. But given the potential for a significant political backlash, the government response may focus mostly on improving safety nets and reducing the associated cost of unemployment, which would reduce the positive impact on R* and potentially lead to much higher inflation.
What about the wider context?
Another important consideration is that AI is not happening in isolation. In some ways, the incentive to adopt these new technologies is higher because several other structural forces, most notably demographic changes and deglobalisation in high-income countries, are raising the cost of labour and lowering productivity.
- Demographics — AI could help offset the negative implications of the expected drop in the pool of available workers in high-income countries, but the extent to which it will do so is, as yet, unclear.
- Deglobalisation — Our research suggests that we are likely to see deglobalisation intensify, which could work in the opposite direction to AI by lowering productivity. Alternatively, the incentive to adopt AI may increase because deglobalisation has made labour more expensive. In turn, successful adoption of AI may encourage countries to accelerate the process of deglobalisation even further as it lessens the need for access to foreign workers and supply chains.. At this stage, it is impossible to tell which trend will prevail.
Bottom line
- We think that AI clearly has the potential to alter the macro landscape and counteract the secular stagnation theme, and we see tentative evidence that markets have started to anticipate this potential recently.
- While markets need to anticipate and price the future, it does seem early given the lack of investment to date and the lack of clarity on governments’ response to potential job displacement, which could entail more inflation on the journey to AI implementation.
- Moreover, the speed of implementation, scope of applications and level of usage are important factors to consider. The potential of AI should also be assessed in the context of deglobalisation and demographics as both are likely to constrain trend growth and long-term real rates.
- On balance, I think investors should prepare for meaningful adjustments in rates and asset prices as market participants seek to gauge the extent to which AI can lift productivity and trend growth.
1S. Briggs and D. Kodnani, “The potentially large effects of artificial intelligence on economic growth”, Goldman Sachs, March 2023.
Equity Market Outlook
In our Equity Market Outlook, we offer a range of fundamental, factor, and sector insights.
By
Andrew Heiskell
Nicolas Wylenzek