This is my weekly column on Forbes.fr
It is available in French here
Here is the English version
The issue of robots and employment emerged recently during the French election campaign, with presidential candidate Benoît Hamon discussing two aspects of this matter at great length. The first aspect is job shortages caused by robotization of the economy: the result of this shift is to make basic universal income a necessity to offset the impact of these shortages on French citizens’ income. The second area involves taxing robots in order to finance the social model as well as education. I will look into these issues in detail in three successive columns. The first will deal with the traditional relationship between innovation and employment, the second will focus on the introduction of artificial intelligence and the third will discuss the issue of taxing robots and explore the key question of who owns these machines. 
The focal question here is employment and the effects of a more robot-based workforce on jobs. The taxation of robots is also key, as the various interpretations of this policy can have diametrically opposing effects.
Questions on the relationship between technological innovation and jobs are far from new. Right from the start of the industrial revolution, technological innovation was the cause of extreme conflict, leading to vast and violent demonstrations. The Luddite movement in the English textile and weaving industry (1811-1813) and the Lyon silk weavers or “canuts” group in France (1831-1834) both serve to show the extent of resistance to the introduction of innovative manufacturing processes.
In 1980, French economist and sociologist Alfred Sauvy wrote a fascinating book on the issue “La Machine et le Chômage” (Dunod 1980, “Machines and Unemployment”, publication in French). Using figures to back up his theory, he showed that throughout history, innovation and employment have not been mutually exclusive.
Innovation profoundly alters the structure of the labor market due to the high productivity gains it generates, thereby changing the innovative sector on a permanent basis as the number of jobs it provides decreases while production remains the same. However, the ensuing price decline and resulting stronger purchasing power, as well as the profits unlocked by innovation, lead to higher demand and the development of other sectors. New sectors can be born out of the impact of these advances, but fresh momentum can also be breathed into older existing sectors, which enjoy a revival as a result of higher overall demand generated by increased purchasing power and greater profits.
So we can see that the labor markets reaps both direct and indirect benefits, and this momentum leads to net new jobs. This is broadly speaking what we have witnessed in the past. Historically, innovation has not been a negative factor for jobs in the medium term. However, in the short term, the situation is obviously more complex due to resistance and inflexibility.
This is why Alfred Sauvy stated that the labor market must be able to adapt to suit this new environment. This also requires each individual to adapt his or her skills and talents to this changing context, and so education must be at the very center of this process to ensure that each worker can constantly adjust to this new and shifting climate.
We can also look at the matter from a slightly different angle, which helps shed light on the question of the decline in industrial jobs in conjunction with innovation. There is often confusion between innovation and globalization in this respect, as many attribute the decline in industrial jobs to the role of globalization. Research in this field shows that technological progress in production processes has had a much greater impact on the decline in industrial jobs than globalization.
To understand this logic, Paul Krugman wrote a short article in 1997 on the impact of innovation on employment. He summed up the key issues perfectly.
His approach is simple but not simplistic.
He describes an economy that produces only two things: hot dogs and the buns to put them in. The economy employs 120 million workers, with each of the two industries employing 60 million workers. It takes two person days to produce either a hot dog or a bun, which means that 30 million hot dogs and 30 million buns are produced each day.
He goes on to suppose that a hot dog can now be made in one day instead of two: 60 million hot dogs but still only 30 million buns will be produced each day. Labor is reallocated between the two sectors and this balances out with 40 million workers making hot dogs and 80 million making buns. So 40 million hot dogs with buns can be made every day.
However, this long-term balance on the labor market can involve severe short-term disparities.  As Krugman suggests, failing to take a comprehensive view of the situation leads to an analysis that hot dog production has increased by 33% while employment has fallen by 33%, and we immediately jump to a negative conclusion on the impact of innovation on employment.
But in fact, there was no reduction in overall jobs but rather a reallocation of labor. Productivity gains in a sector can reduce jobs in that sector but for the economy as a whole this reduction is by no means an insignificant matter. Krugman’s simple example illustrates this point perfectly.
This example is even more striking if we substitute manufactures for hot dogs and services for buns.
For decades, we have seen that progress in the industrial sector has led to an increase in activity and a reduction in the number of jobs. Productivity gains have been redistributed into services, where jobs are increasing the most due to lower productivity gains in this sector. More services are required to meet higher demand, thereby employing more workers due to low productivity growth.
This explains the clear expansion in jobs in the services sector, and this process best illustrates the decline in industrial jobs. The impact of this trend is much greater than the effects of globalization.
In our current times, the worry is that the nature of innovation has changed with the resolution of Polanyi’s paradox. This paradox states that “we can know more than we can tell” and this has long been a barrier to the sophistication of innovation: if a human being cannot fully and precisely describe the various stages in a process, then a programmer cannot program all these aspects into a machine.
The difference now is that Polanyi’s paradox is no longer a paradox. Machines can learn even when humans have not taught them. With this seismic technological shift, it looks like nothing can ever be the same again.
To be followed …