Less wages, fewer jobs? MIT economist: We are being replaced by robots

At present, robots are widely used in various industries, and even in some fields, the degree of automation of robots is no less than that of humans. However, when humans enjoy the convenience of work and life brought by robots and automation, they do not know that they are being replaced by robots. Last year, Amazon CEO Jeff Bezos publicly stated that robots will replace all warehouse positions in the world within 10 years. Today, how much work has robots and automation really replaced us?

As early as October last year, a 17-year study (1990-2007) covering 19 industries in the United States and involving 722 regions gave precise data. The study was co-authored by MIT economist Dalron Azimoru and Boston University Assistant Professor of Economics Pascual Restrepo.

Pascual Restrepo, Assistant Professor of Economics, Boston University

They conducted a comprehensive analysis of detailed robot deployment data and demographic, employment, business, and salary data from agencies such as the US Census Bureau, Economic Analysis Bureau, and Bureau of Labor Statistics. They found that adding one robot to every 1,000 workers reduced the US employment-to-population ratio by about 0.2%. This means that within the United States, for each additional robot in the manufacturing industry, an average of 3.3 workers will be replaced; moreover, the increased use of robots has also reduced workers’ wages by about 0.4% (adjusted for inflation).

In addition, they believe that the impact of robots on employment varies by industry and region, and may be one of the important reasons for increasing income inequality. “We found that robots have a very serious negative impact on workers’ employment. In the more affected areas, the real wages of workers are falling because robots are very good at competing with them.”

They analyzed the impact of robots on employment in 722 commuting areas (mainly metropolitan areas) in the United States, and found that the frequency of robot use varies considerably geographically. The most affected area in the United States is where the automotive industry is located, and Detroit, Michigan, is the biggest victim. “In the United States, different industries have different characteristics in different places,” said Azimoglu. “The most obvious area for robot substitution problems is Detroit.”

Paris-Claire Algui, Professor of Economics at the University of Saclay

David Otor, an economist at the Massachusetts Institute of Technology

In the course of conducting this study, they also spent a considerable amount of time to study whether the employment trends of workers in robot-intensive areas may be caused by other factors such as trade policy, but they did not find a complicated empirical effect. Therefore, they believe that robots do have a direct impact on income inequality. The manufacturing jobs replaced by robots come from some workers who do not have other better employment options. There is a direct link between the use of robots and automation and the decline in the income of blue-collar workers. Azimoru said, “This involves major distribution issues. When robots are deployed to manufacturing plants, pressure is on low-skilled workers, especially medium-skilled workers. Among the technological factors that have increased employment inequality over the past 30 years, Automation accounts for a larger part.” But they also found that in commuting areas, each robot replaced up to 6.6 jobs locally. However, due to the reduction of commodity costs and other reasons, the addition of robots in manufacturing has benefited workers in other industries and other regions.

The same is true in France
Coincidentally, the same answer was obtained in another study by the two scholars of Azimoru, Restrepo and Claire Algui, a professor of economics at the University of Paris-Saclay, namely the rapid realization of automated manufacturing Enterprises can flourish, but the overall employment of the industry will decline.

From 2010 to 2015, the industries with the least investment in robots in France included paper and printing, textile and clothing manufacturing, home appliance manufacturers, furniture manufacturers and mining companies; while pharmaceutical companies, chemical and plastic manufacturers, food and beverages Manufacturers, metal and machinery manufacturers, and automobile manufacturers are the manufacturing industries that add the most robots to their production lines. In fact, from 2010 to 2015, the use of robots in manufacturing increased by 20%, resulting in a 3.2% decline in industry-wide employment. However, during this period of time, companies using robots have increased their employees’ working hours by 10.9% and their wages have increased slightly.

They surveyed 55.39 million French manufacturing companies, 598 of which purchased robots between 2010 and 2015. The study used data provided by the French Ministry of Industry, customer data from French robot suppliers, customs data on imported robots, company-level sales, employment and wages and other financial data. They found that companies that used robots earlier would eventually hire more workers, and unemployment pressure was shifted to companies that did not use robots or used them later. As their investment in technology promoted more growth and more market share, they added more employees. “We know that companies use robots to reduce costs. It is very likely that companies that used robots earlier will expand faster than other companies.” Azimolu said. In contrast, the labor share of companies that have not increased robots has not changed, and for each 10% increase in the use of robots by competitors, the number of employees in these companies will fall by 2.5%. Obviously, companies that do not invest in technology are losing out to their competitors.

Although the employment of companies using robots will increase, robots in general have made more workers in the manufacturing industry unemployed. A study from MIT economist David Otor and other scholars also confirmed this conclusion. At the same time, it also pointed out that the decline in the overall labor share of the United States was driven by the earnings of “superstar companies”, and these companies found ways to reduce labor shares and gain market power. Although these “superstar companies” may hire more employees, and even as the company grows, the salary will be relatively high, but overall, the labor share of their industry is declining.

Although economists provide many possible explanations for the general decline in labor share, such as technology, taxation policies, changes in the labor market system, etc., Azimru believes that technology, especially automation, is the primary factor. “Most of the economic literature on technology, globalization, and labor market mechanisms are now discussing how to explain the decline in labor share, many of which are quite interesting assumptions, but in France, only companies that use robots have reduced their The share of labor, which is all the reason for the decline in the labor share of French manufacturing. This really emphasizes that automation, especially robots, is a key part of understanding what is happening.” Azimru said.

Robots and automation exacerbate inequality
Another study by Azimeru and Restrepo believes that the impact of automation on the labor market and income inequality is more severe than the results of previous studies. They set 1987 as a key turning point in this process, because in this year, jobs lost due to automation are no longer replaced by an equal number of similar jobs. Azimuru said, “Automation is critical to understanding inequality.”

Research shows that from the 1960s to the 1980s, many new employment opportunities brought about by technology have benefited low-skilled workers. But starting in the 1980s, especially in the 1990s and early 2000s, low-skilled workers faced a double blow. They were hurt by unemployment. The upcoming new tasks came more slowly and only benefited high-skilled workers. In short, those lost factory positions or telephone answering jobs will not return. Azimuru believes that automation is “different from general technologically-biased technological change” because it can replace jobs without bringing much productivity to the economy. For example, the self-checkout system in supermarkets or pharmacies reduces labor costs, but it does not improve work efficiency.

Where automation is relatively high, workers with lower skills not only do not benefit, they are even economically passive. More specifically, the annual growth rate of productivity has been estimated to be about 2% since 1963, and the wages of low-skilled workers have increased by about 1% per year. However, the actual wages of low-skilled workers have actually been declining since the 1970s. Enterprises will use fewer workers, but they will not bring so many benefits to other workers. The company has no reason to hire more workers or pay more wages to other workers. But Azimolu also said: “Don’t despair, don’t be frustrated. Not that technology is harmful to employees, but that the choice of technology development direction is crucial.”