Skip to main content
Research Briefs | November 10, 2020

Understanding and Addressing the Modern Productivity Paradox

Erik Brynjolfsson, Seth Benzell, Daniel Rock


Download Research Brief

We are in the midst of a technological revolution driven by advances in artificial intelligence (AI). Machines can now accomplish many tasks that only human minds could do as recently as 10 years ago (Perrault et al., 2019), from recognizing images (Russkovsky et al., 2015) and understanding speech (Schmelzer, 2020), to generating plausible text (Brown et al., 2020) and diagnosing diseases as well as or better than human doctors (Estevaet al., 2017). These are not insignificant tasks.

More broadly, the emergence of scalable machine intelligence witha variety of applications is proving to be of first-order importance for solving many economic problems. These technologies have been amplified by an exponential growth of other digital capabilities, including worldwide digital infrastructure that now brings the mobile internet to more than 4 billion people (International Telecommunication Union, 2019). As a result, an ordinary smartphone now delivers information and services that dwarf those available to even a well-connected billionaire of the 1990s.

Yet, in spite of the emergence of these new technologies—with their enormous industrial potential—the rate of productivity growth in recent years has been disappointingly slow. According to official statistics, productivity growth averaged over 2.8 percent per year in the United States in the decade ending 2005, but since then has slowed by half. If U.S. productivity had grown at the same rate from 2005–2019 as it did from 1995–2004, U.S. GDP would have been approximately $4.2 trillion higher at the end of 2019 than it was measured to be. This modern productivity paradox is a redux of the information technology (IT) productivity paradox of the late 1980s (Brynjolfsson, 1993). That earlier divergence between the promise and practice of technology was epitomized by Nobel laureate Robert Solow’s pithy remark, “You can see the computer age everywhere but in the productivity statistics.”

Paradoxes challenge our assumptions, stimulate innovative research, and lead to improved policy recommendations. On the research side, the current productivity paradox has led to an outpouring of plausible explanations that can be classified into four categories (Brynjolfsson, Rock, and Syverson, 2018).

One possibility is that, despite the excitement of technologists and investors, today’s advances simply fall short and will never fulfill their expected economic promise. A second explanation is that the technologies are delivering, but we are failing to measure the growing output of the new economy properly, particularly the explosion of free digital goods, and that is leading to systematic and increasing shortfalls in our tallies of economic activity. A third possibility is that the technologies are privately beneficial, but the social benefits are largely dissipated in zero-sum rent-seeking. The final explanation is the one we find most compelling: new technologies take time to diffuse, to be implemented, and to reach their full economic potential. For a transformative new technology like AI, it is not enough to simply “pave the cow paths”by making existing systems better.Instead, productivity growth from new technologies depends on the invention and implementation of myriad complementary investments and adjustments. The result can be a productivity J-curve, where productivity initially falls, but then recovers as the gains from these intangible investments are harvested.

Fortunately, there are a range of policy interventions that can speed this process, not only boosting productivity, but also fostering shared prosperity. These include (1) increasing investment in R&D, directly as well as via grants and tax credits, (2) increasing human capital by reinventing education and encouraging high-skill immigration, and (3) eliminating the bottlenecks to reorganization and innovation created by burdensome noncompete rules, occupational licensing, inadequate infrastructure, unconstrained monopoly platforms, barriers to entrepreneurship, and unequal taxation of labor and capital. With the right policies, any nation can overcome the modern productivity paradox, boosting both incomes and equity.