Why we shouldn't overstate the pandemic's effect on productivity growth 

A boost to productivity growth is a commonly cited macroeconomic silver lining of the COVID-19 crisis. After lockdowns and social distancing forced consumers and firms to adapt to digital channels, even for services a credible narrative of a productivity tailwind has emerged.

When such a structural tailwind is considered alongside the cyclical productivity gains associated with tight labour markets (periods in which firms struggle to find workers and speed their technological investment) it is tempting to expect much higher growth.

However, as productivity narratives proliferate it is important that policymakers, investors, and business leaders remain realistic about the macroeconomic upside. Clearly, something is stirring at the microeconomic level but extrapolations to the macroeconomy are treacherous and exuberant expectations could come at a cost.

To develop a realistic understanding of the potential we should investigate what role services play in aggregate productivity growth; what constitutes a transformative growth boost; and what unseen hurdles are slowing the growth of service sector productivity in the post-COVID economy.

Services have slowed productivity growth for decades

Advanced economies have been experiencing a productivity growth slump for many years, but it’s useful to remember that aggregates always obscure important details. A closer look at the US economy (see chart below) illustrates the dynamics. Contrary to common perception, high productivity growth has been a persistent feature in many sectors throughout the decades, as seen in the green colours in the chart. The problem is that these sectors represent a shrinking part of the economy, dominated by goods production.

Meanwhile, the service economy has relentlessly grown its share of total output, but productivity growth has stayed stubbornly low, as seen in the red colours, dragging down aggregate productivity growth.

If we hypothetically “unskew” this compositional mix-shift (i.e., assume sector weights of the 1950s but modern-day sector productivity) then aggregate productivity growth would be 60 basis points (bps) higher today – an enormous difference against the economy’s trend growth rate (which considers growth in labour, capital and productivity) of just about 2%.

This tells us not only that the service economy harbours enormous potential for productivity growth but also that any transformative upshift will have to occur here. If COVID-19 is truly a catalyst for such inflection the upside could be big. The question is how big.

What does it take to transform growth? History is a good place to start in unpacking the answer. The last time the US economy experienced a strong growth surge was during the information, communication, and technology (ICT) revolution of the 1990s which came down to three numbers: 30, 100, and 10. The step-change in productivity growth followed 30 years of sustained investment in networked computing; it then lifted US total factor productivity growth by 100bps, and that shift lasted for about 10 years.

So, if a COVID-induced growth transformation of 100bps sounds like a high bar that’s because it is.

First, consider the ICT upshift was achieved because it revolved around a broadly applicable technology that reached many corners of the economy simultaneously. But the COVID-related potential is a narrative of countless different marginal tweaks in many different sectors, firms, and business models.

It is tempting to take case studies of firms’ productivity gains, extrapolate them, and add it all up – a bottom-up approach much the way one might go about identifying cost savings in a large organization. But such micro-to-macro extrapolations are treacherous. When uncertain changes are coming from countless tweaks there will be a wide distribution of outcomes – both in magnitude and timing because the tweaks themselves vary widely in nature.

Second, don’t forget about the baseline. In the 1990s, 100bps was transformative because the baseline productivity growth rate was already 70bps and ICT lifted it to a very high 1.7%. But over the past few years, productivity growth has been very low. If productivity growth were to rebound by 100bps to around 1% that would be a welcome improvement over recent years, yet it would be a long shot from the transformative 1.7% of the 1990s.

Third, returning to our chart above, take today’s sector shares in the economy but assume that in the coming years the services economy could be lifted to the average productivity growth realized in the goods economy over the last 70 years. This would lift aggregate growth by around 80bps.

There are other reasons why large productivity boosts should be approached with caution. Microeconomic hurdles are easily overlooked in the face of the remarkable agility that consumers and firms have demonstrated during this crisis. We highlight four.

Parallelism ― when processes are duplicated. Consider online education and e-health, which are often cited as COVID-induced breakthroughs in service delivery. In both cases physical and virtual processes must now be run in parallel, adding a layer of cost for what is the same (and often worse) output. That points to lower, not higher, productivity. To unleash productivity growth, firms must find ways for additional cost to result in more valuable services or reach many more customers.

Dilution ― when low-efficiency steps are added to a service. Consider online grocery offerings that have grown in popularity during the pandemic and added the last mile of delivery to retail operations. This low-productivity activity was previously done by households and thus excluded from retail productivity figures. Thus, adding the last mile to the retail transaction will be a drag on productivity, not a boost. To offset the productivity drag, firms must now work harder at efficiency gains across their operations.

Stickiness ― when behavioural change is situational. Consider a major technological shift such as networked computing, which is a cheaper and better technology. Powerful network effects delivered self-sustaining momentum until penetration was very high. But many of the marginal and dispersed changes triggered by COVID have a "workaround" rationale and could normalize when the context does. As a simple litmus test for stickiness, we should ask ourselves if consumers are generally eager to accelerate virtualization or to reclaim their pre-COVID realities.

Adoption ― when there are barriers to absorbing and upgrading in a timely way. Even where change is compelling and the payoff clear, an economy's technological skillset is not homogenous across sectors, not to mention the differences in political, regulatory, and cultural obstacles to adoption. If online teaching were – hypothetically – superior to the classroom experience, would teachers and their unions support a structural shift to e-learning with fewer teachers?

While optimism is warranted, exuberance is not. Overstating the COVID effect on the economy could carry costs that would vary by stakeholder.

Monetary policymakers have the most to lose from unrealistic expectations. Expecting modest productivity gains, such as those driven by cyclically tight labour markets, should help them avoid the policy error of being too tight. But assuming a structural productivity boom would risk the opposite policy error. Overestimating the economy’s capacity could turn out to be the conceptual foundation for policy that is consistently too loose.

Investors also stand to lose if unrealistic expectations of productivity growth lead them to pay unjustified valuations across asset classes. Those valuations would need to reset lower when growth underperforms expectations.

On the other hand, business leaders should be exuberant because their instinct is to make the most from crisis learnings and innovate. Individually, they are leading the microeconomic renewal, which will sometimes succeed and sometimes not. Yet, collectively, their best efforts provide the credible prospect of a macroeconomic tailwind – if not a growth transformation.