Guess what data tells us about the magnitude of the Uberisation of jobs?

Accelerating fears, decelerating productivity

Not one day goes by without hearing about predictions of massive job losses due to "Uber isation". However, statisticians find it hard to read the effect of said Uberisation outside of a very narrow set of sectors (car transportation, rental of apartments, etc.). Furthermore micro-economic studies tell a quite different story for sectors outside this set. In manufacturing [1], for example, the way production adapts to new technologies is very slow and incremental. Productivity gains may indeed have many causes, such as additive manufacturing (3D printing"), increased digitization of processes made possible by the price drop of sensors and of information processing, or a simply better organization of the manufacturing chain. And each year, industrial productivity increases slowly, even in the absence of any technological/business model disruption. Actually if something new is happening it is rather a deceleration in productivity gains and productivity investments rather than an acceleration!

Not the same impact for interaction, production and design jobs

The issue is therefore not to brandish the digital threat some gurus and consultants like to evoke, but rather to understand the nature of such threat for each job category. In many cases, the threat is more likely to come from a competitor - who may be quicker than you in using digital in order to increase performance, or to propose innovative services instead of an Internet giant or a start-up popping up out of the blue. There is real risk of investing too much to counter a theoretical Uberisation, and there is also real risk in not investing enough to reinforce your own productivity and quality of service. These poor investment choices, fueled by hazardous extrapolations and narratives, bring inspiring optimism to investors and fears to businesses: this is precisely the definition a bubble [2].

In transportation, Uber reduces "interaction" jobs (handling the relation between a cab driver and its customer), not "production" jobs: there are not less people driving a car. The design jobs (management, design, artists) are still far from being disrupted: at best, artificial intelligence provides them with tools eliminating a small and repetitive part of their job. In the case of Uber, interaction and production jobs are complement, not substitutes: the lower the cost and the complexity of the former, the more plentiful the latter. Cab companies owners, call centers operators, and taxi licenses owners can indeed worry about Uberisation. But drivers shouldnt as Uberisation leads to a net increase in the number of driver jobs, and this is a net gain for society as a whole after taking into account customers surplus in the tally. Please also note that were still very far from the driverless cabs plane autopilots have been existing for years, and they still need a human supervision.

In order to estimate the overall impact of Uberisation on employment, we used employment data broken down across 88 industries (retail, metallurgy, construction ...) and different functions (production, interaction and design) in France. In our model, "production" and "design" activities are affected by productivity gains (use of machines, robots for production, artificial intelligence) but they dont disappear entirely. We show that interaction jobs may however be replaced entirely by the Uberisation process. Our analysis was completed using industry experts assessments and takes into account the dynamics of demand. For example, the risk of a decline in employment in healthcare services is reduced because the demand for such services increases faster than their productivity.

Schumpeter beats Uber

Our model estimates that, in the medium term, 14 % of the total employment could be displaced by digital. This can of course just be an order of magnitude, but it is consistent with other studies (OECDs[4] estimate is 9%). Some jobs are indeed made redundant by digital, and they are the ones the most visible the "interaction" jobs (sales staff, call centers...). But millions of "production" jobs (construction, garbage collection,...) are hardly affected. Industrial jobs do indeed see annual productivity gains, but the size of such gains has been decreasing, not increasing, in the last decade.

Structure of the jobs by industry

Our model also reveals some winning sectors. They are the ones complementary to interaction jobs: the simpler and the cheaper the interactions, the more customers will consume services sold through theses interactions. Farmers are already experiencing a maximum level of competitive pressure from purchasing groups. An Uber for fruit and vegetables (which would be delivered directly to customers) could hardly reduce their margins even more. On the other hand, it could allow customers to raise the level of quality and increase the farmers revenues. As mentioned above, a decrease in interaction costs will lead to an increase in production demands efficiency gains are not lost, but reallocated somewhere. Finally, the economy is not inert: in France, 15% of all jobs are destroyed every year and the same percentage are created because of evolutions related to technology (more renewable energy jobs and less jobs related to coal), competitiveness (exchange rate variations, know-how development...) or consumer tastes (more e-cigarette and less mobile phone shops). The impact of Uberisation (14% of employment over several years) is not much compared to Schumpeterisation (15% of employment each year)!

There is no doubt that digitization will reduce the number of jobs in some sectors and increase it elsewhere. With each technological wave, there are risks for some jobs categories, but also opportunities for jobs and value creation. It is better to start with factual and realistic numbers in order to better deal with such risks and grasp such opportunities.

Vincent Champain, senior executive in a multinational company and president of the Long Term Observatory think tank (http://longterme.org), and Frederic Benque, investment partner at NextWorld. Many thanks to Joyce Bessis for her help on the English version .

[1] https://hbr.org/sponsored/2015/06/big-data-and-big-oil-ges-systems-and-sensors-drive-efficiencies-for-bp

[2] See Robert Schiller, Rising Anxiety That Stocks Are Overpriced , New York Times,2015.

[3] Employment Study 2013, INSEE

[4] OECD, Automatisation et travail indpendant dans une conomie numrique .