World Economy

Productivity Slowdown Even More Puzzling

The sluggish pace of productivity growth would have dire consequences for future gains in living standards
The high resolution MRI, CT, and sonogram images underpin advances in medical diagnosis.The high resolution MRI, CT, and sonogram images underpin advances in medical diagnosis.

The growth in labor productivity – real output per hour worked–in the US slowed markedly around 2004. During the previous ten years, labor productivity in the business sector had risen at an annual average pace of more than 3%.

Productivity received a boost from innovations in the production of digital information and communications technologies and the spreading use of ICT across all sectors of the economy. Despite the apparent continuation of the digital revolution, productivity growth slowed to about 2% a year during 2004-2010 and then dropped to a paltry pace near just 0.5% during 2010-2016. Outside the US, labor productivity growth also declined in other advanced market economies, economists David Byrne and Dan Sichel wrote for

If sustained, this sluggish pace would have dire consequences for future gains in living standards. Over long spans of time, increases in labor productivity are the fuel that boosts living standards and the main reason that material well-being is significantly higher today than 50 or 100 years ago. At a 2% pace, labor productivity would double every 35 years, while at a 0.5% pace, that doubling would take almost 140 years.

No Accord Reached

Economists have debated the reasons for the slowdown, though no consensus has emerged. Explanations generally fall into three categories. It is attributed to either supply-side accounts pointing to factors limiting the growth rate of potential output; demand-side stories highlighting inadequate demand or interactions between weak demand and sluggish growth of potential output; or, finally, measurement explanations arguing that the tools of economic measurement have not kept up with the digital revolution and that the slowdown is a figment of poor measurement.

"We suspect that this debate will not be resolved anytime soon. That being said, we want to refocus the discussion about measurement of the digital economy. While we agree that available evidence on mismeasurement does not provide an explanation of the productivity slowdown, we also believe that biases in measures of ICT are significant and that this mismeasurement has important implications for economic growth in the future.

"In particular, we argue that innovation is much more rapid than would be inferred from official measures and that these on-going gains in the digital economy make the productivity slowdown even more puzzling. At the same time, we believe that this continued technical advance could provide the basis for a future pickup in productivity growth," the economists say.

Mismeasuring the Digital Economy

The short answer is yes. A key challenge to measuring changes in real GDP over time is correctly deflating nominal GDP to account for changes in prices. Getting price changes right for ICT production and investment is particularly difficult. And, as highlighted in Byrne et al. (2017), official measures of prices point to very slow rates of decline in the prices of high-tech products.

But, a growing literature indicates that prices of ICT are actually falling much faster. For example, Byrne et al. (forthcoming) developed a new price index for microprocessors used in desktop computers. Their index fell at an average annual rate of 42% during 2009-13, while the most comparable official price index–the producer price index–reports an average decline of 6% during the same period.

In addition to ICT production, the digital economy encompasses an increasingly wide range of economic activity. The list is long and includes the myriad applications delivered by internet service providers, the energy exploration that relies on world class supercomputers to process seismic data, the surging e-commerce sector cannibalizing shopping malls, the Hollywood movies produced with computer-generated imagery, the high resolution MRI, CT, and sonogram images that underpin advances in medical diagnosis, and many others.

Developing price indexes that capture quality change for these industries is challenging. Moreover, the deep investment in ICT in all of the industries mentioned suggests that the official prices for their output, which in most cases show little change, may be missing something important.

Prices of many ICT products are still almost completely unstudied, including devices used in the medical, aerospace, and defense sectors and the industrial robots that have been the subject of much recent media attention.


Add new comment

Read our comment policy before posting your viewpoints