AI Faces Worse Macro Backdrop Than the 1990s IT Revolution

Summary

• IT capex shows the internet revolution has been ongoing since the 1990s.
• Yet the macro data shows no obvious impact on productivity.
• This could reflect measurement issues and a falling labour income share that has made for sluggish demand growth and inflation.
• Investors have still benefitted from the IT revolution through high profits and valuations.
• The AI revolution could be less beneficial to investors due to de-globalization and greater workers’ market power.

Market Implications

• AI-linked investment adds upside risk to demand and therefore raises the risk of overheating and an end to disinflation. This is an end-2024/2025 risk.
• It does not change my expectations of a first cut at the June FOMC.

The Permanent IT Revolution

The IT revolution is often dated to the 1990s, but in reality it never stopped. In the US, internet and computer use surged in the 1990s thanks to technological innovations such as the PC, mobile phone and World Wide Web (Charts 1-3; Appendix Table 2).

The explosion of IT capex paused in the early 2000s as businesses had overinvested and the US entered a recession (Chart 2). But IT investment growth soon resumed and proved immune to the GFC and pandemic.

As a result, the IT intensity of the US private sector has kept increasing. It accelerated during the pandemic, in line with the worker shortage (Chart 3).

Unlike with businesses, household spending on IT goods and services only took off in the 2000s. In line with businesses, it has kept increasing and surged during the pandemic (Chart 4).

The IT revolution was enabled by an extraordinary increase in the productivity of the computer industry, most of which was passed to final users through lower costs (Chart 5).

More recently, productivity growth has slowed. The price of computers has stabilized since the early 2010s and that of chips since the late 2010s. Moore’s Law, which had predicted a doubling of transistors per square inch of chip every 18 months, no longer applies.

However, the cost of software, which is five times as large as hardware investment, has continued to fall.

Yet all this impressive investment has had only limited impact on measured productivity.

High Capex, But No Productivity Increase

Table 1 shows key macro and market indicators. It shows that, as discussed above, IT and overall capex have increased steadily since the 1970s. However, there is no commensurate acceleration in productivity: a pickup in productivity in the 2000s reflects mainly the negative impact of the GFC on employment.

What the data shows, however, is a faster increase in profits than GDP from the 1990s. This may well reflect that the IT revolution has been mainly labour augmenting: that is, it has reduced the demand for labour and therefore labour bargaining power and income share.

The data also shows a steady decrease in inflation, which reflects several factors outside the IT revolution. These include Fed policy, globalization (that has been partly enabled by the IT revolution), and sluggish recoveries from recessions after the 1990s.

The lack of productivity ‘bang’ for the IT ‘buck’ has been long debated. One possible explanation is that productivity and GDP are underestimated.

Another possible explanation could be Keynes’ paradox of thrift: when households want to save more, they decrease demand for businesses’ products and therefore lower growth. Because the propensity to save out of profits is greater than out of wages, by increasing the profit share, the IT revolution could have increased ex-ante savings and made for sluggish demand growth, income and inflation. Sluggish demand growth in turn makes for sluggish productivity growth.

In any event, investors have benefitted from the IT revolution through higher profits and stronger valuations: low inflation has been accompanied by falling real rates that have supported higher PEs.

AI Less Favourable Macro backdrop

How different is AI from the ongoing IT revolution? At some level, very different. The AI revolution is still in its early stages with no telling which new businesses will emerge. We could be at a stage equivalent to the early 1990s when no one would have predicted that a technology as transformational as e-commerce would become dominant.

Also, the AI revolution, like the IT revolution, is likely to be labour augmenting. However, the tasks that AI replaces will likely be more complex than those IT replaced. For instance, many businesses now rely on automated, web-based billing and payments that were previously performed manually. The AI revolution could go one step further and lead to computer-generated business proposals.

Meanwhile, the macro backdrop is different from that of the IT revolution. For instance, globalization is in retreat. By contrast, the globalization of the 1990s and 2000s allowed the IT revolution to play a major role in the lengthening of global supply chains and increasing profits. This time, the scope for greater profitability could be more limited.

In addition, there could be more resistance now against a decrease in labour’s income share (Chart 6). The success of populism suggests workers’ income share could have already reached its political limits. Furthermore, the workers likely to be impacted by AI will have higher skills and income and therefore more political clout than the workers displaced by the IT revolution. Workers could therefore resist the adoption of AI or demand a greater share of the productivity gains.

This could see the AI revolution accompanied by stronger growth and inflation, but possibly lower profits than the IT revolution. For investors, the results could be mixed – possibly slower earnings growth and lower valuations as higher real rates are needed to keep inflation in check.

Market Consequences

It is unclear how the AI revolution is impacting the economy because there is scant data on its adoption. Most likely, we are still at the early stages of the propagation of this new technology. So at this stage, AI is adding upside risk to already high and rising capex. As such, it adds to the risks of overheating and an end to disinflation, which I see as an end-2024/2025 risk.

In the longer run, as discussed above, the impact on growth and inflation could turn out more positive than that of the IT revolution.

Appendix: Table 2, Key IT and AI Benchmarks

.

Dominique Dwor-Frecaut is a macro strategist based in Southern California. She has worked on EM and DMs at hedge funds, on the sell side, the NY Fed , the IMF and the World Bank. She publishes the blog Macro Sis that discusses the drivers of macro returns.

.

(The commentary contained in the above article does not constitute an offer or a solicitation, or a recommendation to implement or liquidate an investment or to carry out any other transaction. It should not be used as a basis for any investment decision or other decision. Any investment decision should be based on appropriate professional advice specific to your needs.)