Could the Rise of AI Mean Falling Prices?

There’s been much discussion of AI’s deflationary potential, but this issue must be viewed in the broader context of other megatrends influencing a new investment regime.

There is a long history of automation leading to deflation, or at least disinflation—it was one of the key economic narratives before the COVID-19 pandemic. Does the rollout of AI offer a possible new impetus for deflation? If it does, it would have massive implications for asset allocation as well as imply that nominal fixed-income assets are attractively priced after the recent rise in yields.

However, we think this issue needs to be seen in the broader context of other megatrends that are influencing the investment regime.

We’ve made the case in recent notes, including our black book, that we were already in a new strategic investment regime even before AI exploded into the public consciousness. Specifically, we mean a new regime versus the dominant paradigm that has been in place since the 1980s. The confluence of deglobalization, a shrinking working-age population in most key economies and the need to pay for the energy transition all imply a default setting of higher inflation.

A Potential Counter to Inflationary Secular Trends

What forces could work against this trend? AI is the strongest contender. In meetings with CIOs and asset allocators over the past six months, AI has been brought up by many investors struggling to assess how, as an economic force, it will interact with these other trends. Two potential narratives are key here, viewed through the narrow lens of economic and investment concerns: AI’s roles as a potential deflationary source and as a force for higher growth through better productivity.

Assessing the potential impact on prices requires untangling the near-term cyclical inflation wave from longer-term inflationary forces likely to act on prices over the next five to 10 years. The cyclical wave took much longer to tame than central banks expected and was essentially a hangover from supply constraints born in the pandemic and simultaneous extra demand from fiscal support. Russia’s invasion of Ukraine added another inflationary impetus. AI is a force that acts over much longer time horizons, so it’s appropriate to compare it to the other mega forces we noted earlier.

It’s incredibly hard to quantify the scale of AI’s impact at this early stage of its rollout, especially as the role of automation in the disinflationary environment of recent decades is hotly debated. There is an unavoidable joint-hypothesis problem: How much of the disinflation in recent decades was from automation and how much was from the extra supply of labor from globalization and favorable demographics along with policy choices and independent central banks?

Even if AI’s impact is hard to quantify at this stage, through what mechanism could it impact prices? As with the internet, it’s about giving powerful tools to a wide range of economic agents at low cost, enabling them to reduce the cost of producing many services.

The Labor Dimension of AI’s Productivity Inroads

A specific impact on wages is a key part of any narrative of AI and deflation. As with many previous technology waves, AI in the near term creates a risk of displacing workers, reducing labor’s bargaining power. What is notable and different from the recent past is that the professions most at risk from AI have very low rates of unionization (Display). Other things equal, this implies that workers may have less bargaining power in the face of disruption.

AI raises profound questions on what the future of employment (and indeed retirement) could look like. We’ve made the point that the resulting impact is probably more a political and social choice than something determined by technology alone, despite what techno-determinists would argue. If AI really does eradicate jobs in a more significant way than previous technology waves, it ultimately raises the probability of a universal basic income being enacted, which would suggest a radically different relationship between policy choices and inflation.

Unpacking Potential Impacts on Inflation Baskets

One route to attempt to quantify AI’s impact on inflation is to consider which parts of the headline inflation index are most likely to be impacted.

Housing is by far the largest category of the US Consumer Price Index basket (Display), accounting for nearly 45% of the total. Housing construction, highly regulated and labor-intensive, had one of the lowest rates of productivity growth over the past 30 years. AI and automation can improve efficiency and reduce construction time, but we don’t see a major AI-powered productivity breakthrough. Continued below-average productivity in the CPI’s largest component should dampen AI’s disinflationary impact.

Currently, AI holds a lot of promise for healthcare applications, such as speeding up drug discovery and breakthroughs in protein folding. However, the rapid population aging will create a very strong demand for social-care workers in the coming decades. While AI can lower costs and improve outcomes in certain areas of healthcare, the needs of a rapidly aging population will still likely drive above-average growth of costs in this area.

Further factory automation and AI-assisted product design will help contain costs for new cars and trucks, the key component of the transportation category. However, this will be offset by the need to invest in strengthening supply chains and re-shoring, as well as wage demands from an increasingly vocal and highly unionized labor force. Developments in precision agriculture, predictive analytics and automation of restaurant services will help limit price rises for the food and beverage category, but climate change will create more volatility for input prices from more frequent supply squeezes.

Hard to Forecast Today…and Likely to Be Unevenly Distributed

To sum this up, we can find potential disinflationary effects of AI-driven productivity gains in nearly all sectors of the economy. However, the benefits are not only extremely hard to forecast today but also likely to be unevenly distributed across industries. And large, highly regulated parts of the economy, such as construction, transport and medical care will face structural impediments that are inflationary. So, in our view, a key economic point about AI is that it will be even more important to consider the cross-sectional effects in addition to the aggregate economic outcome.

We’ve made the point that the potential positive and negative aspects of AI will likely be priced in different ways. The potential for raising productivity and the possible input into dampening inflation will likely be priced faster than the potentially negative implications, which have more to do with second-order social and political implications.

What Does All This Mean for Investors?

We still think investors should be prepared for a new regime than the one prevailing since the mid-1980s, which featured higher inflation and lower real growth than the norm. AI does provide a force that could mitigate some of these other macro forces, but there are likely limits to the extent that AI can lower overall inflation. Thus, the bottom line is that we think inflation and real assets are still key long-term focal areas of strategic asset-allocation decisions.