Can AI be anything other than deflationary?

As AI weighs on inflationary forces, policy-makers and investors may look back at more conventional efforts to stabilise prices with nostalgia, writes Christopher Smart, managing partner of Arbroath Group.

If the principal question confronting today’s investors is the future path of inflation, second on the list must be the looming impact of artificial intelligence. It’s hard to answer the first without considering the second, as markets try to assess what to expect from the post-post-pandemic economy.

While there are many reasons to think monetary policy will need to be fundamentally tighter than it has been after four decades of deflation, AI is the one major reason to think inflation, as we know it, will soon dissipate.

AI may not deliver the ‘deflation bomb’ some have forecast, but it will serve as a powerful counterweight to the inflationary forces that are now building. No part of the economy will be untouched.

Recall, first, the familiar charts of declining inflation and interest rates since the early 1980s. Globalisation opened vast new supplies of labour, suddenly increasing productive capacity and reducing goods prices. Second, ageing citizens in rich countries have been saving for retirement as wary governments in emerging economies accumulated large reserves for the next crisis. Finally, automation has done more to hollow out manufacturing employment than any free trade arrangements with China.

Today, these currents are shifting. Rising political tensions have firms scrambling to diversify their supply lines, suggesting the cost gains from globalisation are on pause. Meanwhile, many of those who were saving furiously for retirement are now retired, ailing and in need of government support. Then there are rising international tensions that will propel defence spending nearly 5% higher annually over the next five years. There will also be extensive public and private spending with the climate transition, including investments in new sources of energy, resilient infrastructure and cleaner forms of transportation.

All else equal, central banks that have been struggling to reach 2% price growth targets may now have to lean hard in the opposite direction to keep inflationary pressures under control. Or maybe not.

In some ways, AI is like any other innovation: a tool to enhance worker productivity that has been lagging over the last decade. The standard economic argument should be that higher productivity means higher wages, more household spending, better investment returns and more capital investment. Logically, real interest rates and inflation should drift higher.

Nevertheless, there are important reasons to believe this ‘productivity boost’ is not like its predecessors. Just as robotics and automation replaced countless tasks able-bodied workers once performed, AI can now perform much of what ‘able-minded’ employees do. But if the blue-collar jobs withered away over decades as fresh capital equipment was installed and new factories were built, the hollowing out of the white-collar workforce will come with much cheaper and faster upgrades to network software.

An estimate of the number of jobs AI may eventually replace is educated guesswork at best but extensive analyses at McKinsey confirm it will be ‘lots’. Even if algorithms don’t entirely take over full-time roles as presently defined, they will rapidly peel off functions within those roles and trigger a contraction of a firm’s employee base. Employees who perform more complex and creative functions may hang on longer, but even their functions will be radically reshaped as the software improves.

Moreover, the introduction of AI to the knowledge-based sectors comes at the same time as two related, but separate transformations that will improve efficiency and reduce costs further. The internet of things connects distant operations through inexpensive sensors and mobile networks, allowing factories, tractors or airplanes to be operated and monitored at a fraction of current costs.

Meanwhile, new systems designed around distributed ledgers and blockchain will make it easier and faster to transfer value without large back offices and intermediaries.

Creation of new work will come as it always does and the benefits will begin to spread to those who get access to fresh training, alternative certifications and transitional help. But the new functions in new industries will appear much more slowly and will require creative government and private sector strategies to smooth the way.

The world’s central banks face a novel set of uncertainties. They will need to balance the inflationary forces emanating mainly from government spending against the unprecedented deflationary forces that will emerge from as fast a technological transformation as we have ever seen.