The Economist January 22nd 2022 Finance & economics 69Updateinprogress
W
hen thepandemic first struck, unemployment soared. Not
since the Depression had American joblessness surpassed
14%, as it did in April 2020. But fears of a prolonged period of high
unemployment did not come to pass. According to the latest avail
able data, for November, the unemployment rate for theoecdclub
of mostly rich countries was only marginally higher than it was
before the pandemic. By now it may even have drawn level. The
rich world’s labourmarket bounceback is the latest phenomenon
provoking economists to look again at a foundational question in
the discipline: whether robots help or harm workers.
The gloomy narrative, which says that an invasion of jobkill
ing robots is just around the corner, has for decades had an ex
traordinary hold on the popular imagination. Warning people of a
jobless future has, ironically enough, created plenty of employ
ment for ambitious public intellectuals looking for a book deal or
a speaking opportunity. Shortly before the pandemic, though, oth
er researchers were starting to question the received wisdom. The
world was supposedly in the middle of an artificialintelligence
and machinelearning revolution, but by 2019 employment rates
across advanced economies had risen to alltime highs. Japan and
South Korea, where robot use was among the highest of all, hap
pened to have the lowest rates of unemployment.
Many thought that the pandemic would at last prove the doom
mongers right. In mid2020 a highly cited paper published by
America’s National Bureau of Economic Research argued that co
vid19 “may accelerate the automation of jobs”, and another assert
ed that it was “reinforcing both the trend towards automation and
its effects”. A paper published by the imf wondered whether the
jobs lost during the pandemic would “come back”. Part of the logic
was that since robots don’t fall ill, bosses would turn to them in
stead of to people—as seemed to have happened in some previous
pandemics. Others noted that bursts of automation tend to occur
during recessions.
Two years on, though, the evidence for automationinduced
unemployment is scant, even as global investment spending is
surging. The rich world faces a shortage of workers—by our reck
oning there are a record 30m unfilled vacancies across the oecd—
which is hard to reconcile with the idea that people are no longer
necessary. Wage growth for lowskilled workers, whose occupa
tions are generally thought to be more vulnerable to replacement
by robots, is unusually fast. There is still little evidence from
America that “routine” jobs, thought to be easier to automate, are
shrinking relative to other sorts of jobs.
Considering that so many doubts about the “robots kill jobs”
narrative have arisen, it is not surprising that a different thesis is
emerging. In a recent paper Philippe Aghion, Céline Antonin, Si
mon Bunel and Xavier Jaravel, economists at a range of French and
British institutions, put forward a “new view” of robots, saying
that “the direct effect of automation may be to increase employ
ment at the firm level, not to reduce it.” This opinion, heretical as
it may sound, does have a solid microeconomic foundation. Auto
mation might help a firm become more profitable and thus ex
pand, leading to a hiring spree. Technology might also allow firms
to move into new areas, or to focus on products and services that
are more labourintensive.
A growing body of research backs up the argument. Daisuke
Adachi of Yale University and colleagues look at Japanese manu
facturing between 1978 and 2017. They find that an increase of one
robot unit per 1,000 workers boosts firms’ employment by 2.2%.
Another study, by Joonas Tuhkuri of the Massachusetts Institute
of Technology (mit) and colleagues, looks at Finnish firms and
concludes that their adoption of advanced technologies led to in
creases in hiring. Unpublished work by Michael Webb of Stanford
University and Daniel Chandler of the London School of Econom
ics examines machine tools in British industry and finds that
automation had “a strong positive association with firm survival,
and that greater initial automation was associated with increases
in employment”.
Noneconomists can be forgiven for rolling their eyes at the
profession’s apparent aboutface. But things are not as simple as
saying that economists had got it wrong before. For a start, statis
tical methods have improved since the publication of the founda
tional papers in robonomics, such as one by Carl Benedikt Frey
and Michael Osborne of Oxford University in 2013, which was
widely interpreted as saying that 47% of American employment
was at risk of automation. The methodology used by Mr Adachi
and his coauthors is particularly clever. One problem is untan
gling causality: firms on a hiring spree may also happen to buy ro
bots, rather than the other way round. But the paper shows that
firms buy robots when their prices fall. This helps establish a caus
al chain from cheaper robots, to more automation, to more jobs. The onrushing wave...of research
A second qualification is that the “new view” does not establish
that automation is “good”. So far, it has had little to say about job
quality and wages. But a forthcoming book by David Autor, David
Mindell and Elisabeth Reynolds of mitfinds that even if robots do
not create widespread joblessness, they may have helped create an
environment where the rewards are “skewed towards the top”.
Others argue that automation reduces job quality.
Mr Aghion and his colleagues add that even if automation
boosts employment at the level of the firm or industry, the effect
across the economy as a whole is less clear. In theory robotadopt
ing companies could be so successful that they drive competitors
out of business, reducing the total number of available jobs. Such
questions leave researchers with plenty more toinvestigate. But
what seems clear at this stage is that the era of sweeping, gloomy
narratives about automation is well and truly over.nFree exchange
Economists are revising their views on robots and jobs