By Mira RojanasakulMira Rojanasakul and Peter CoyPeter Coy

Are you about to be replaced by a robot? The question has broad implications for the U.S. economy, especially the manufacturing sector. Industries that robotize tend to increase output. But robots can have dire consequences for workers.

Two economists recently concluded that both jobs and wages fall in parts of the U.S. where more robots are installed. The March 2017 study by Daron Acemoglu of Massachusetts Institute of Technology and Pascual Restrepo of Boston University shows the commuting zones—i.e., local labor markets—where robot installations have grown the most.

Increase in robots per thousand workers from 1990–2007

Lowest

Low

Medium

High

Highest Increase in Robots

Source: Daron Acemoglu and Pascual Restrepo; U.S. Census

The upper Midwest, particularly Michigan, was ground zero for the robot explosion from 1990 to 2007. That makes sense, since the automobile industry uses more robots than any other. The other hot spots also make sense on closer inspection. In Beaumont, Texas, lots of workers are employed in the plastic, chemicals and pharmaceuticals industry, another big user of robots. Wilmington, Delaware, has a big chunk of workers in that industry and others in car manufacturing, according to Restrepo, one of the researchers.

Those increases tended to mean fewer jobs. Of course, lots of factors weigh on employment. Foreign competition, overvaluation of the dollar and rising productivity all play a big part, too. But even after taking all those other factors into account, Acemoglu and Restrepo still found that additional robots in an area reduces workers and cuts local wages. Their research included any machine that was autonomous and capable of performing multiple tasks. (So: not coffee makers.)

Relationship between industrial robot exposure and employment

Exposure to industrial robots grew relative to the employment-to-population ratio

Commuting zone, sized by population

Labels = major city in the commuting zone

Change in Employment, 1990–2007:

+10

+8

+6

Lima, OH

Findlay, OH

+4

Defiance, OH

Youngstown, OH

Lorain, OH

Racine, WI

Dayton, OH

+2

Wilmington, DE

Saginaw, MI

Cleveland, OH

Detroit, MI

0

Toledo, OH

–2

Rockford, IL

Muncie, IN

–4

South Bend, IN

Lansing, MI

Beaumont, TX

Grand Rapids, MI

Fort Wayne, IN

Overall, an increase in robots corresponded to a decline in employment within local economies

–6

Indianapolis, IN

Ludington, MI

Big Rapids, MI

–8

Traverse City, MI

0

0.5

1.0

1.5

2.0

2.5

3.0

3.5

4.0

4.5

5.0

Change in exposure to Robots from 1990–2007 ⟶

Commuting zone, sized by population

Labels = major city in the commuting zone

Change in Employment, 1990–2007:

+10

+8

Lima, OH

+6

Findlay, OH

Defiance, OH

+4

Racine, WI

Lorain, OH

Wilmington, DE

+2

Saginaw, MI

Detroit, MI

0

–2

Indianapolis, IN

–4

Muncie, IN

Rockford, IL

Lansing, MI

South Bend, IN

–6

Overall, an increase in robots corresponded to a decline in employment within local economies

Beaumont, TX

Fort Wayne, IN

–8

0

0.5

1.0

1.5

2.0

2.5

3.0

3.5

4.0

4.5

5.0

Change in exposure to robots from 1990–2007 ⟶

Commuting zone, sized by population

Labels = major city in the commuting zone

Change in Employment,

between 1990–2007⟶

+8

Wilmington, DE

Toledo, OH

+4

Cleveland, OH

Detroit, MI

0

–4

Fort Wayne, IN

Beaumont, TX

–8

Lansing, MI

0

1

2

3

4

5

Change in exposure to robots

from 1990–2007

Source: Daron Acemoglu and Pascual Restrepo

The period covered by the study ended right before the recession of 2007–09. Acemoglu and Restrepo said the recession introduced too many variables that affected employment, which would have made it more difficult to isolate the impact of robots.

So what’s happened since the recession, and what does it mean for manufacturing jobs? We gathered and plotted data from the Bureau of Labor Statistics for various manufacturing sectors to try to get a better sense. Obviously, this type of analysis doesn’t let you draw the direct conclusions Acemoglu and Restrepo were able to make about robots. But you do see some interesting trends when it comes to manufacturing employment.

For starters, U.S. industry still likes robots. Orders for new robots slumped during and just after the recession. Since then, they’ve shot back up.

U.S. industrial robot orders (units)

Recession

Source: Robotic Industries Association

Automation doesn’t end with robots, however. BLS data shows that “equipment”—machinery, furniture, vehicles and computer software, along with some other components—is increasingly important in production processes compared to labor. Within manufacturing industries, which have historically been the leading edge of robotics usage in production, there were increases across the board.

Change in equipment intensity, 1987–2014

Increase in the use of capital vs. labor in production

  • +668% Computer and Electronic Products
  • +500% Petroleum and Coal Products
  • +356% Furniture and Related Products
  • +327% Apparel and Leather and Applied Products
  • +249% Transportation Equipment
  • +199% Machinery
  • +183% Printing and Related Support Activities
  • +119% Chemical Products
  • +115% Food and Beverage and Tobacco Products
  • +113% Nonmetallic Mineral Products
  • +105% Miscellaneous
  • + 79% Primary Metal Products
  • + 67% Textile Mills and Textile Product Mills
  • + 62% Fabricated Metal Products
  • + 54% Plastics and Rubber Products
  • + 49% Electrical Equipment, Appliances and Components
  • + 9% Paper Products
  • + 4% Wood Products
  • Source: Bureau of Labor Statistics, Office of Productivity and Technology

    That seems to coincide with increased productivity in many sectors. Often, when there’s an increase in equipment use, there’s also an increase in output. And output is growing faster than employment, which means that output per worker is rising.

    But there are huge disparities between sectors. In some, like the apparel sector, output has fallen along with employment, even as the industry has become more automated. Some of the sectors with the biggest job losses were ones that didn't automate intensively, like textiles and paper products. That goes to show that keeping out robots won't necessarily protect your job.

    Change in employment and output in U.S. manufacturing industries

    For 3-digit NAICS manufacturing industries, percent change since 1987

    Equipment Intensity

    Employment

    Output

    Computer and Electronic Products

    Petroleum and Coal Products

    Furniture and Related Products

    Leather and Applied Products

    Apparel and Applied Products

    Transportation Equipment

    700%

    700%

    600%

    600%

    500%

    500%

    The first industrial robots were developed for the auto industry, which still accounts for over half of U.S. robot orders.

    400%

    400%

    300%

    300%

    PC boom starts around this time.

    200%

    Apparel made in the U.S. went from about 50% in 1990 to 3% in 2014.

    100%

    100%

    0

    0

    -100%

    -100%

    Machinery

    Printing and Related Support Activities

    Chemical Products

    Food Products

    Beverages and Tobacco

    Nonmetallic Mineral Products

    300%

    300%

    Had an average annual growth rate of 20.5% in robot orders since 2011.

    200%

    200%

    100%

    100%

    0

    0

    -100%

    -100%

    Textile Mills

    Textile Product Mills

    Miscellaneous

    Primary Metal Products

    Fabricated Metal Products

    Plastic and Rubber Products

    300%

    300%

    The metals industry invested $164 million in robot shipments in 2016.

    200%

    200%

    100%

    100%

    0

    0

    -100%

    -100%

    Wood Products

    Electrical Equip., Appliances and Components

    Paper Products

    300%

    300%

    200%

    200%

    100%

    100%

    0

    0

    -100%

    -100%

    Computer and Electronic Products

    Petroleum and Coal Products

    Furniture and Related Products

    Leather and Applied Products

    Apparel and Applied Products

    700%

    700%

    600%

    600%

    500%

    500%

    400%

    400%

    300%

    300%

    PC boom starts around this time.

    200%

    200%

    Apparel made in the U.S. went from about 50% in 1990 to 3% in 2014.

    100%

    100%

    0

    0

    -100%

    -100%

    Transportation Equipment

    Machinery

    Printing and Related Support Activities

    Chemical Products

    Food Products

    The first industrial robots were developed for the auto industry, which still accounts for over half of U.S. robot orders.

    300%

    300%

    Had an average annual growth rate of 20.5% in robot orders since 2011.

    200%

    200%

    100%

    100%

    0

    0

    -100%

    -100%

    Beverages and Tobacco

    Nonmetallic Mineral Products

    Miscellaneous

    Primary Metal Products

    Textile Mills

    300%

    300%

    The metals industry invested $164 million in robot shipments in 2016.

    200%

    200%

    100%

    100%

    0

    0

    -100%

    -100%

    Textile Product Mills

    Fabricated Metal Products

    Plastic and Rubber Products

    300%

    300%

    200%

    200%

    100%

    100%

    0

    0

    -100%

    -100%

    Electrical Equip., Appliances and Components

    Paper Products

    Wood Products

    300%

    300%

    200%

    200%

    100%

    100%

    0

    0

    -100%

    -100%

    Computer and Electronic Products

    Petroleum and Coal Products

    Furniture and Related Products

    700%

    700%

    600%

    600%

    500%

    500%

    400%

    400%

    300%

    300%

    PC boom starts around this time.

    200%

    100%

    100%

    0

    0

    -100%

    -100%

    Leather and Applied Products

    Apparel and Applied Products

    Transportation Equipment

    700%

    700%

    600%

    600%

    500%

    500%

    The first industrial robots were developed for the auto industry, which still accounts for over half of U.S. robot orders.

    400%

    400%

    300%

    300%

    200%

    200%

    Apparel made in the U.S. went from about 50% in 1990 to 3% in 2014.

    100%

    100%

    0

    0

    -100%

    -100%

    Machinery

    Printing and Related Support Activities

    Chemical Products

    300%

    300%

    Had an average annual growth rate of 20.5% in robot orders since 2011.

    200%

    200%

    100%

    100%

    0

    0

    -100%

    -100%

    Food Products

    Beverages and Tobacco

    Nonmetallic Mineral Products

    300%

    300%

    200%

    200%

    100%

    100%

    0

    0

    -100%

    -100%

    Miscellaneous

    Primary Metal Products

    Textile Mills

    300%

    300%

    The metals industry invested $164 million in robot shipments in 2016.

    200%

    200%

    100%

    100%

    0

    0

    -100%

    -100%

    Textile Product Mills

    Plastic and Rubber Products

    Fabricated Metal Products

    300%

    300%

    200%

    200%

    100%

    100%

    0

    0

    -100%

    -100%

    Wood Products

    Electrical Equip., Appliances and Components

    Paper Products

    300%

    300%

    200%

    200%

    100%

    100%

    0

    0

    -100%

    -100%

    Computer and Electronic Products

    Petroleum and Coal Products

    700%

    600%

    500%

    400%

    300%

    PC boom starts around this time.

    200%

    100%

    0

    -100%

    Leather and Applied Products

    Furniture and Related Products

    700%

    600%

    500%

    400%

    300%

    200%

    100%

    0

    -100%

    Apparel and Applied Products

    Transportation Equipment

    700%

    Apparel made in the U.S. went from about 50% in 1990 to 3% in 2014.

    600%

    500%

    The first industrial robots were developed for the auto industry, which still accounts for over half of U.S. robot orders.

    400%

    300%

    200%

    100%

    0

    -100%

    Machinery

    Printing and Related Support Activities

    300%

    200%

    100%

    0

    -100%

    Chemical Products

    Food Products

    300%

    Had an average annual growth rate of 20.5% in robot orders since 2011.

    200%

    100%

    0

    -100%

    Beverages and Tobacco

    Nonmetallic Mineral Products

    300%

    200%

    100%

    0

    -100%

    Miscellaneous

    Primary Metal Products

    300%

    The metals industry invested $164 million in robot shipments in 2016.

    200%

    100%

    0

    -100%

    Textile Mills

    Textile Product Mills

    300%

    200%

    100%

    0

    -100%

    Plastic and Rubber Products

    Fabricated Metal Products

    300%

    200%

    100%

    0

    -100%

    Electrical Equip., Appliances and Components

    Paper Products

    300%

    200%

    100%

    0

    -100%

    Wood Products

    300%

    200%

    100%

    0

    -100%

    Source: Bureau of Labor Statistics, Office of Productivity and Technology, Robotics Industries Association

    Bottom line: Robots do replace workers. On the other hand, some industries that don't automate end up losing workers anyway, because their costs are too high and their customers go elsewhere. For workers, robots are only part of the problem.