Financial markets are increasingly enthusiastic about AI. Many investors see it as a transformative technology that could reshape the modern economy, lift productivity and perhaps even permanently boost long-term economic growth. AI is often described as a “general purpose technology” – comparable to electricity or the Internet – because it has the potential to transform large parts of the economy simultaneously rather than improving only a narrow sector or production process.
AI will almost certainly change how economies function. It is hard to deny that incomes will rise as workers adopt the technology and become more productive. But whether AI can permanently raise the rate of income growth is a much harder question to answer.
In the short term, growth is likely to rise. One reason is mechanical: Investment spending is itself part of GDP. As firms spend heavily on computing infrastructure, software and employee (re)training, measured output will increase. Growth will also be boosted as AI spreads gradually through the economy, boosting the productivity of workers and firms that embrace it. As we see incremental rises in productivity towards a higher level, productivity growth itself will temporarily become stronger.
But once that adjustment process is complete, will the long-run growth rate remain higher? That is far less clear. And yet this is the more consequential question for investors and policymakers.
Potential impacts on monetary policy and markets
A crucial distinction must therefore be made between the level of income and the rate of income growth. A one-off increase in productivity can permanently raise the level of output and living standards without permanently increasing the long-run growth rate. If productivity rises by 20%, society becomes permanently richer. But once that adjustment has occurred, growth may simply return to its previous rate.
If AI were to permanently increase long-run productivity growth, the effects on financial markets and monetary policy would be profound. Faster expected growth would increase the equilibrium real interest rate and, over time, the policy rates set by central banks.
The logic behind this is simple: Consider someone who receives a promotion and expects to earn a much higher salary in the future. Faced with better long-term prospects, that person may decide to buy a larger house or apartment, financing the purchase with future earnings.
The same principle applies to the economy as a whole: If households and firms expect future incomes to rise faster, they will want to borrow and invest more today. But future output has not yet been generated. Real interest rates therefore rise to restrain borrowing and keep demand in line with available resources.
Expectations of stronger future growth can therefore push up real interest rates long before the economy itself visibly changes.
If investors came to believe that AI would permanently boost long-term growth, the effects on financial markets could be dramatic. Stronger expected future earnings would support equity valuations, while higher equilibrium real interest rates would tend to put upward pressure on long-term bond yields.
Historical experience and the need for caution
History offers many examples of transformative technologies that have fundamentally changed the way economies functioned. They include: Steam power, railways, electricity, automobiles, aviation, antibiotics, computers, semiconductors, the Internet and mobile communications. Each produced enormous gains in productivity and living standards.
In many cases, contemporaries believed that the economy had entered a fundamentally new era. The scale of the technological changes often appeared so large that permanently faster growth seemed almost inevitable. But the long-run growth rate itself appears remarkably stable.
To examine this question, we used historical data on real GDP per capita from 1800 to 2022 for a group of countries including Argentina, Australia, Belgium, Brazil, Canada, Chile, Denmark, Finland, France, Germany, Italy, Norway, Spain, Sweden, Switzerland, the UK and the US. We then estimated the underlying long-term growth trend using a standard statistical technique.
Real GDP per capita is not a perfect measure of productivity growth but it provides a useful proxy for long-term changes in living standards and productive capacity across countries and over time. And crucially, it disregards growth coming from an increasing population and labour force.
The results show substantial fluctuations in trend growth over time, including strong post-war growth and a smaller acceleration during the information technology boom around the year 2000. But despite repeated waves of transformative innovation, there is little evidence of a sustained upward trend in long-run growth rates.
Periods of unusually rapid growth do occur, most notably after wars and deep recessions, when economies recover lost output and rebuild capital stocks. The period after 1945 is a clear example of this. However, these episodes tend to be temporary. Once reconstruction is complete and economies return to normal, growth rates generally fall back again. The results are also very similar if the sample is restricted to advanced economies alone.
This historical experience therefore suggests caution is needed when assessing the long-term implications of AI. The technology may well produce a large one-off increase in productivity and living standards but history offers less support for the view that it will permanently raise the underlying rate of economic growth.