The Productivity Paradox: Now’s Not the Time to Give Up On ICT

A continuing puzzle lies at the heart of global economic performance this decade. That of a significant deceleration (and across a broad range of countries) in productivity growth. Unfortunately short-term predictions on its near-term recovery also remain moribund. The Conference Board’s 2019 Productivity Brief for example, notes that “Our latest estimates extend the downward trend in global labor productivity growth. The results also indicate that the long awaited productivity effects from digital transformation are still too small to see in a lasting improvement at the macroeconomic level”.

Fig 1. Source: Economist Intelligence Unit Country Data; National Statistical Offices; Huawei Calculations

How can this be when it feels like we live in such a highly technological age? With such profound recent improvements in information communication technologies such as artificial intelligence, IoT and 5G?

In fact, a continuing theme of economic discussions about information communications technologies (ICT) has been the puzzle of its impact on the statistics we use to measure economic growth. This was first raised during the latter half of the 1980s and came to be called the ‘Solow Paradox’. In 1987, renowned economist Robert Solow remarked that “you can see the computer age everywhere but in the productivity statistics”.

The Solow Paradox was temporarily resolved in the mid-to-late 1990s by a sharp acceleration in productivity growth, particularly in the United States. This was commonly referred to as the ‘Information Technology Revolution’. But as Fig 1 illustrates, there has been a sharp deceleration in growth across many countries since the mid-2000s. Indeed, for the United Kingdom (Fig 2) the stagnation in output per worker means that the average employee produces almost 20% less than would have been expected if the 1980-2007 trend had continued. This ‘lost productivity gap’ goes a long way to explaining stagnating wage growth for many countries across this decade too.

What’s really going on? Has the Solow Paradox returned with a vengeance?

The productivity slowdown has been particularly marked in some countries:

Fig 2. Source: Economist Intelligence Unit Country Data; Huawei Calculations

Much academic research has been undertaken in recent years into this issue. Economists now appear to have a better understanding of this macro-retrenchment. Originally one of the more popular touted causes was one of basic economic mismeasurement. Specifically the idea that a large share of the consumer surplus generated by Internet-related goods and services was not being captured effectively in the national GDP accounts (mostly because many internet services are free but for other reasons too). However, rigorous estimates of Internet-related consumer surplus by economists struggle to account for the gap implied by the productivity slowdown. And this problem has always been with us – think of the original consumer surplus created by white goods such as the washing machine or vacuum cleaner. They freed-up many weekly hours of domestic drudgery but are relatively cheap!

The ‘Solow Paradox’ period can now be understood as a gestation period during which both ICT capital and a range of complementary assets (labour skills, new organizational structures, new regulations) were being accumulated so as to fully exploit the underlying technology. Hence investment levels preceded productivity boosts, which eventually appeared in the national accounts data in many countries from 1995.

Increasingly, major industrial revolutions should be viewed as multi-decadal waves with periods of peaks and troughs. The ICT revolution, which arguably began in the late 1970s, is still evolving.

The most likely explanation of the current slump in productivity growth lies in the notion of another broad implementation lag. And its length holds great promise for the future. Our current era is plausibly a time when firms are still working out strategies for utilizing the full potential of artificial intelligence (AI) and robotics, as well as other complimentary general purpose technologies (GPTs) such as the cloud, big data analytics and 5G. Like other general purpose technologies, the full impact will not be felt until a wave of complimentary factors are developed and implemented. The required adjustment costs, organizational changes, and new skills can be considered as an increasing stock of intangible capital. And there is increasing optimism that this stock of intangible capital – which by its nature goes mostly unrecorded – has been growing rapidly this decade. Much business activity (especially by leading firms) is being redirected into these types of activities which are not yet easily measurable as economic outputs.

Rather, a more important challenge for policy-makers is likely to be how the economy of the near future absorbs a set of potential ‘super innovations’ that will help power this resurgence in trend total factor productivity growth.

Two industrial revolutions compared side-by-side:

Fig 3. Source: Artificial Intelligence and the Modern Productivity Paradox: A Clash of Expectations and Statistics, Erik Brynjolfsson Daniel Rock, Chad Syverson (updated from Syverson (2013))

Brynjolfsson et al (2017) directly compare the electricity and ICT revolutions (Fig 3) and find that the timing of gestation and delivery match up closely to the ICT era, although the electricity revolution happened slightly quicker. Indeed, the late 1910s surge in productivity growth may have answered some economist’s query as to why one sees new-fangled electric motors and internal combustion engines everywhere in 1910, but not in the productivity statistics!

It is sometimes foolish to make predictions about the economy. But it looks an increasingly sound bet that a productivity revival may be just around the corner, if we are careful to build a suitable and global enabling environment for it right now.

Leave a Reply