The rise of the machines - Are Science Fiction movies becoming true?

The trends in robotization are unreversable by now. The International Federation of Robotics (IFR) shows that the number of installed robots has grown by 85 percent - only in the last 5 years (2014-2019). Asia is playing an increasing role in this dynamic. It accounts for 2/3 of the newly installed robots in 2019. It is not surprising that the effects of robotics have caught the eye of the public debate. How are they influencing our labor markets? Are they really grabbing our jobs, and if yes, which ones? And what are their effects on inequality? Let’s take a tour around the world, guided by the findings of economic scholars in the field.

Schumpeter’s Theory of Creative Destruction

Innovation and technological change has accompanied us for long, and robots are just one of the recent technological advances we are facing nowadays. Already in 1942 Schumpeter recognized the destructive but inevitable force of technological change. What economists refer to as the Schumpeter’s theory of “creative destruction” (for a detailed explanation see this report by MIT) refers to a process in which new production units constantly replace outdated ones. This is what Schumpeter refers to as the essence of what drives and moves growth.

So what about robots? How are they acting out their potential of “creative destruction”? The pioneer paper by Graetz and Michaels (2018) looks at robotization in different industries and 17 different countries from 1993 to 2007. The researchers find that, overall, an increase in robot adoption leads to an increase in overall productivity as well as wages and a decrease in prices. Still, it has negative employment effects on the low-skilled. This, on the other hand, could then lead to an increase in inequality betweeen skill groups. Robotics as a potential driver of inequality is also comfirmed by recent work of Moll et al. (2021). They find a positive effect of robot adoption on inequality due to income benefits for the high-skilled as well as increases in returns to capital (or wealth) for its owners.

I, Robot - And Evidence from the US

robots_us

Recent evidence from the US shows that robots have negative effects on wages and employment Acemoglu and Restrepo (2020). The researchers at MIT and Boston University explore the labor market effects of an increased exposure to robots at the commuting-zone level. This is based on the fact that, according to the authors, robots and workers compete on the labor market and that robots potentially displace some of the human workforce (a phenomenon they call the displacement effect). The authors find sizable effects, namely that an increase of 1 robot per 1,000 workers from 1993 to 2007 recuded wages by nearly 1 percent and the employment-to-population ratio by nearly 0.39 percentage points, relative to a commuting-zone without robots. This means that one new robot decreases employment by 6 workers. But robotization does not only have negative effects. It leads to productivity increases and decreases in prices of those goods produced by robots (Acemoglu and Restrepo (2020)). When taking account of these positive side effects, the overall effect of robotization is still negative but by much less. In this case, one new robot decreases employment by 3.3 workers.

So what about inequality? Acemoglu and Restrepo (2020) find that certain groups are more affected by robots than others. Specifically, effects are more marked for the manufacturing sector and those industries which concentrate a larger number of robots. Additionally, the effects are more pronounced for workers conducting routine manual jobs as well as blue-collar workers. The researchers also look at the effects by education group and find negative impacts on all workers, independently of their educational status. Still, when looking at the wage distribution, they show that those earning less than the average are the ones most affected by the wage decrease. The fact that robot adoption does not result in positive impacts for the high-skilled means that robots are not complementary to their tasks and increase their productivity (as is the case for other forms of technological change, as for example in the IT sector).

Ex-Machina - And Evidence from Europe

Europe. Let’s move away from the US and have a look at Europe. Similar to the US, researchers have become more and more interested in studying the effects of robots on local labor markets. In 2018, an analysis by Chiacchio et al. (2018) finds that, when looking at a total of 6 countries in the EU, one additional robot decreases the employment-to-population ratio by 0.16 to 0.20 percentage points. The young as well as middle-educated are the ones most affected by these negative employment effects, and the authors find no evidence on wage growth. The overall effects are significant but half of those observed in the US. This could be driven by different labor market policies and welfare systems. But they are not the only ones who have studied the impact of robotization in the EU. Evidence from different countries come to different conclusions, and I am going to summarize two of them for you in the following.

robots_germany

Germany. In Germany, the evidence points towards insignificant effects of robot adoption on local labor markets (Dauth et al. (2021)). While robotization has a negative impact on employment outcomes in the manufacturing sector, these job losses are compensated by job growth in the service sector. Within firms affected by robot technology, incumbent workers start to adapt their tasks and instead of leaving firms take over new tasks, which are more complex and of higher quality. In addition to that, those in jobs, which are complementary to the ones in which workers are replaced by robots (such as managers or technical scientists), benefit from robot adoption due to increased productivity. The researchers find negative effects on young workers who just join the labor market. But as the young adapt their educational choices and increasingly go to colleagues and universities, the impact is minimal. The overall effects on wages are insignificant but ambigous. While those staying with their employers, but switching to higher-quality tasks, face relatively large gains in earnings, those forced to switch firms face losses in their wages.

What does that mean for inequality? Well, there are definitely losers and winners in the story of robotization. While those who manage to stay benefit from robot adoption, those who leave their firms experience losses. So, inequality could rise among those responsible of the same tasks, but not among different skill-groups or along the wage distribution. Additionally, labour market institutions could be important to protect those at risk of the negative effects of robot adoption. Dauth et al. (2020) find that the displacement effect of robots is larger in labor markets with lower labor market protection.

robots_france

France. A recent paper conducting analysis using firm-level data from France comes to very different results than the ones found for the US. In contrast to Acemglu and Restrepo (2020) the authors of What Are the Labor and Product Market Effects of Automation? New Evidence from France find that automation leads to an increased labor demand, higher profits, lower consumer prices and higher sales. It therefore has an overall positive effect. Employment effects are even positive for unskilled industrial workers. This could be an indicator for the productivity effect (an increase in productivity driven by robotization) outweighting the displacement effect (robots substituting for workers) (see Aghion et al (2020)). Consequently, in a globalized world, in which robotization is unstoppable by now, it might be damaging to introduce protective mechanisms aiming at securing employment against robots. Firms placed in countries without those mechanisms might outperform those placed in protected economies through the induced productivity gains. Additionally, in the case of France, there are no significant effects of robot adoption on inequalitiy.

Wall-E - And Evidence from Asia

robots_japan

Japan. Asia has become a driving force behind the adoption of robotics worldwide. So how do the effects observed in Asian markets differ from the ones spotted in the US and EU? Research by Eggleston et al. (2021) looks at the effects of service sector robotics in Japan, using establishment-level data. To encounter some of the challenges induced by Japan’s aging population, the country has encouraged the implementation of service robots in nursing homes. These robots have led to an increase in the number of nurses in caring homes. Still, these new positions are marked by flexible and irregular employment contracts. Additionally, the researchers observe a wage decrease for regular nurses. This could be driven by an increase of part-time work (the reduction of hours worked).

Indonesia. For Indonesia, the effects seem to be positive as robotization leads to productivity gains at the firm-level (Presidente (2021)). Robot adoption leads to an increase in employment and the service but also manufacturing industry. The author finds positive effects and positive spill-over effects of robot adoption even for those firms not enganging in robotization. Still, the positive effect is not equally distributed. It leads to a skill wage premia and negative effects of employment in value added.

Transformers - What about the developing world?

robots_dev

Labor markets in developing countries have very different characteristics than the ones in the developed world. A large share of the labor force is informal, low-skilled and conduct different tasks. Additionally, developing countries might not have the same capacity to absorb new technologies (Maloney and Molina (2019)). It is therefore a valid question to ask what robotization does to developing countries. Do the results found for advanced industry nations hold? Work by the World Bank (2019) shows that this might not be the case. While robotization does result in displacement effects, it goes hand in hand with off-shoring (the practice of basing some of a company’s processes or services overseas, so as to take advantage of lower costs) in FDI destination countries, creating new jobs. This off-shoring actually takes place in the same manufacturing sector subject to most of the displacement induced by robotization. This is different for what is observed in developed countries and speaks for a complementary induced through off-shoring. Still, there are negative effects for some developing countries, as Brazil and Mexico (Maloney and Molina (2019)).

So bringing it all together - Are robots taking over?

When looking at the evidence presented here on the effects of robot adoption on different countries, it is undismissible that robotization is impacting labor markets. The sign of the effects seems to depend on a variety of factors, which are still not fully understood. An underlying driver could be the technological stage of the economy, the strength of labor market institutions and welfare systems, or the occupation as well as industry structure of countries facing robotization.

One thing is clear though - all papers have shown that there are winners and losers when confronted with robotization. This might lead to increases in inequality, not just among countries, but also within countries. Some have found increased inequalities among skill groups as well as age groups, or workers performing different types of tasks. Others have shown that it is mostly the rich and capital owners which benefit most. But robotization can also be a potential for development, when monitored carefully. When the productivity increases outweigh the displacement effects, robots can push countries up on their development path. This could fuel growth, job creation and competition, leading to overall positive effects.