AI-Revolution, weniger Jobs für Junge: Was Taxibranche und Buchhandel zeigen
While a challenging macro backdrop (high interest rates, tariff / political uncertainty) could disproportionately hurt fresh entrants to the labour market as firms reduce hiring, the academics found that even within firms, employment was skewed towards less AI-exposed areas for younger people.
Since then, there’s been further anecdotes of pressure e.g. PWC announcing a cut in its graduate intake from 1’500 to 1’300, partly driven by AI.
As investors, we need to understand how quickly AI will be adopted and its impact on productivity. A key uncertainty is whether we need a painful macro recession first to drive ‘creative destruction’. Difficult times focus the minds of companies and consumers to cut costs and try new, cheaper ways of doing things.
Or can firms deliver innovation and technological change during more benign macro backdrops? A micro industry in secular decline should see an ageing workforce as new hires dry up. To test this, I’ve explored black cabs and bookstores, two sectors renowned for technological disruption.
I found a rightward-shift in the age-profile of licenced black-cab drivers as Uber* expanded through London. This is consistent with young people no longer taking ‘the knowledge’ in the late 2010s as satnavs made it obsolete.
I also explored bookstores. They were early victims of e-commerce – their output peaked in 2004. Recall Amazon* started off selling books online before disrupting the market with its kindle e-reader in 2007. The chart below shows how their secular decline in employment was led by a slump in younger workers.
We make the following observations:
In two sectors renowned for technological disruption (taxis and online bookstores), we saw the biggest problems for the youngest workers. This is consistent with ‘no hiring’ of young workers exposed to AI automation being a proxy for technological change, as per the ‘canaries in the coalmine’ hypothesis of the Stanford paper.
Second, the data become more erratic as we move to smaller industrial samples. This is why we appreciated the use of the 3.5 million employee ADP dataset by Brynjolfsson, Chandar and Chen. We used the lagging ACS survey of 3.5 million households.
It will be more difficult to find statistically significant results in research that uses the more timely but, alas, much smaller 60’000 monthly household CPS survey. Even the unemployment rate for recent graduates (cited by Powell at the latest FOMC press conference) is volatile, let alone employment by industry.
So the struggles of today’s youth in sectors most exposed to AI automation according to the recent Stanford paper could potentially be a signal of technological disruption and stronger productivity growth.
Quelle: L&G | gregorcom
Bildquellen: Monkey Business Images/shutterstock | L&G | gregorcom
