Although Bureau of Labor Statistics reports that nonfarm business sector labor productivity grew at a 3 percent annual rate in the third quarter, this healthy growth has been more the exception than the rule in recent years. From 1995 to 2004, average annual labor productivity growth was 2.9 percent. Since 2005 the average annual growth rate has fallen to 1.3 percent. During this same period, technologies such as artificial intelligence or related applications have seen significant breakthroughs.
How can productivity growth and other aggregate statistics remain lackluster across many developed countries during a period of such technological progress? In a recent working paper from the National Bureau of Economic Research, Erik Brynjolfsson and Daniel Rock of MIT, with Chad Syverson of the Booth School of Business, do not see a contradiction between “forward-looking technological optimism and backward-looking disappointment.” They show four possible reasons why we might even expect this clash of expectations and results.
The false hope explanation is that the optimism generated by the development of new technologies is unfounded, and the new discoveries will ultimately fail to deliver real benefits. The authors point to the developments in driverless cars as one example of a recent development that could increase productivity by reducing the time spent commuting or allowing people to work during the trip.
The authors also claim that artificial intelligence is a general-purpose technology (GPT), meaning it meets the following three criteria: “pervasive, able to be improved upon over time, and be able to spawn complementary innovations.” The underlying technology necessary to make autonomous vehicles work, such as image recognition or video interpretation, already meet all three of the criterion. Past examples of GPTs, such as the steam engine or electricity, had a profound effect on living standards and significantly increased productivity.
As I’ve written previously the mismeasurement of productivity growth cannot explain the recent dynamic, as research has found no systemic increase in mismeasurement that coincides with the slowdown.
Another possibility is that the benefits of new technologies are concentrated among relatively few industries or firms, which limits their effect on the economy as a whole. In this telling, the concentrated distribution of the benefits would also encourage wasteful “gold rush” style activities from actors involved. A new breakthrough induces firms to focus more on becoming one of the beneficiaries and blocking access for others than further developing the technology. The authors acknowledge some preliminary evidence of certain aspects of this argument, such as the increase in market concentration in select industries and some signs of a divergence in productivity levels between firms.
However, the aggregate effects of industry concentration and the theory of gold rush activities are not yet well understood, and the diffusion of artificial intelligence could reverse recent trends.
The authors’ preferred explanation springs from their focus on the time lag between new technologies and their implementation. The rate of technological diffusion can be particularly slow for general purpose technologies, such as artificial intelligence, that have the most transformative effects.
These lags come from two sources: the need for the stock of new technology to reach a large enough point to affect aggregate statistics such as economic or productivity growth, and the need for complementary investments necessary to make use of the new technology. Both of these mechanisms take time, and help explain the lag between evidence of technological breakthroughs and those developments having a discernible effect on aggregate statistics or living standards.
The authors compare the recent trend to an earlier period with the diffusion and development of portable power technologies, what they term the combination of electrification and the internal combustion engine.
Labor Productivity Growth Driven by General Purpose Technologies
Source: Brynjolfsson, Rock, and Syverson, “Artificial Intelligence and the Modern Productivity Paradox: A Clash of Expectations and Statistics,” NBER Working Paper 24001, November 2017.
The authors derive two lessons from the comparison. First, accelerated productivity growth can follow a sluggish period. Second, productivity growth generated by general purpose technologies, whether artificial intelligence or portable power technologies, can arrive in multiple waves.
The recent trend of lower productivity growth has negatively affected the U.S. economy. If productivity growth had not slowed after 2004, by a conservative estimate GDP would be about $3 trillion higher per year. Despite the relatively lackluster performance in many aggregate statistics, the rate of technological developments does not appear to have stalled after that inflection point.
The authors reconcile this seeming contradiction and suggest that we are in the interstitial period where those technological breakthroughs have not yet pervaded all aspects of the economy. At this point, the investments both in the technologies themselves and complementary areas have not yet reached the point necessary to fully harness their potential.
Measuring the developments in artificial intelligence, and understanding their implications for different aspects of day-to-day life, is a challenging task. For this reason a consortium of researchers at MIT and Harvard, including Erik Brynjolfsson, launched an AI Index to compile and track developments in artificial intelligence. The index will make it easier to see what is happening within the general purpose technology, and to disseminate this information to the scientific community, policymakers, and the public.
If firms and entrepreneurs continue investing and developing in artificial intelligence, and their efforts are not stifled by regulation, these technologies could deliver substantial benefits in the coming years.
Charles Hughes is a policy analyst at the Manhattan Institute. Follow him on Twitter @CharlesHHughes.
Interested in real economic insights? Want to stay ahead of the competition? Each weekday morning, E21 delivers a short email that includes E21 exclusive commentaries and the latest market news and updates from Washington. Sign up for the E21 Morning Ebrief.