Domestic Affairs Jae Seung Lee

Minimum Wage in the Era of Automation

Will higher minimum wage lead to less employment? How will new technology affect low-wage workers?

By: Jae Seung Lee

Will higher minimum wage lead to less employment? How will new technology affect low-wage workers?


This March, the Maryland General Assembly passed a bill of phasing in a minimum wage increase to $15 an hour by 2025. California, Illinois, Massachusetts, New Jersey, New York and Washington D.C. enacted increases in statutory wage to $15 in upcoming years. Wage floor hikes are gaining momentum in the Congress as well. In January, Democrats introduced a bill, the Raise the Wage Act, that would increase the federal minimum wage to $15 per hour by 2024. It will likely face opposition in the Senate, where Republicans hold the majority. States in which minimum wages are either below or equal to federal minimum wage, $7.25 an hour, will have to adapt this new federal standard. While federal subcontractors have been paid at least $10.35 per hour since 2018, federal minimum wage has not increased since 2009.


The first minimum wage law in the United States was passed in Massachusetts in 1912. Several other states also passed the minimum wage provisions that cover a fraction of demographics such as female and teenage workers. In 1938, the Fair Labor Standards Act enforced federal minimum wage for the first time in history. Institutionalists viewed this as an efficient method to diminish poverty by guaranteeing minimum income.

However, George J. Stigler, the Chicago economist, argued that the optimum minimum wage can work only when labor demand and supply satisfy certain conditions, which he believed to be immeasurable with tools of his time. He claimed that the optimum wage can vary with “occupation and [the quality of worker],” that it varies among firms, and that it also varies with time, “often rapidly.”  He pointed out that “hourly rates are effective only for those who [are employed]” and that hourly earnings do not translate to annual earnings due to seasonality, overtime and shifts. Until the early 1990s, the general consensus on the issue had been that minimum wage decreases employment and earnings.


Since Card and Krueger (1994), economists have been divided on the employment effect of minimum wage. By comparing employment growth at fast food restaurants in New Jersey and Pennsylvania, David Card and Alan Krueger (condolences to Professor Krueger’s family and colleagues) found that a rise in the minimum wage has no negative effect on employment. Their data is based on two waves of phone-call surveys on fast-food franchise restaurants: employers that hire low wage workers the most and also comply with the minimum wage regulations. Their paper’s result is based on difference in difference (“DID”) approach with NJ, where the minimum wage rose from $4.25 to $5.05, as a treatment and PA, where the minimum wage was constant, as a control. NJ and PA were chosen due to its adjacency, which minimize the heterogeneity between the two states other than the uprating. For about a year-long sample period, stores in NJ hired 0.59 person/store more than before while stores in PA hired 2.16 person/store less. This striking result re-kindled the debate on minimum wage.

David Neumark and William Wascher used the payroll data instead to analyze the same phenomenon. Replicating the same DID approach Card and Krueger used, the authors found “NJ minimum-wage increase led to a 3.9% to 4.0% decrease in fast-food employment in NJ relative to the PA control group” and attribute the other paper’s error to “substantially more variability” of unreliable phone call surveys. As a response, Card and Krueger published another paper using data from the Bureau of Labor Statistics and concluded there is no significant effect. Other literatures including Powell (2016) that construct synthetic control group and Clemens and Wither (2014) that uses triple difference models also vary in results as they use different sources of data, compare dissimilar regions or demographics and set diverse control groups.

Overall, the past literature has been limited to empirical analysis of minimum wage. Most of the papers are concentrated on estimating general employment effect of minimum wage. There has been lack of detailed discussions on causal relationship between minimum wage and dependent variables and in-depth analysis of structural transformation. Lordan and Neumark (2017) is a very significant paper as  it is the first to look into the share of employment and change in labor market’s structure in response to minimum wage in context of automation. The paper offers a great insight to policy makers in the age of automation and artificial intelligence, in other words, rapid conversion of labor to capital.


Technological advance and increased mechanization indeed have changed the U.S. job market. The United States has lost 6 million manufacturing jobs in stark contrast to increase in manufacturing output between 1990 and 2014. Over the same time period, healthcare and social assistance have added roughly 9 million jobs. Increase in productivity due to such change and decreasing cost of capital relative to cost of labor contribute to different compositions of capital and labor within firms. However, minimum wage legislations, by raising the cost of labor artificially, may distort the original relationship between labor and capital. Many unskilled workers with automatable jobs are vulnerable to such sudden changes while some skilled workers benefit from this phase of development.

Lordan and Neumark’s empirical analysis draws on monthly pooled Current Population Survey (“CPS”) data of low-skilled individuals with high school diploma equivalent or less from 1980 to 2015. The authors used a Routine Task Intensity (“RTI”) proxy from Autor and Dorn (2013) and Autor et al. (2015) to measure how much of “the tasks within an occupation are automatable”. Based on repeatability, codifiability and other characteristics, the authors determined how plausible substitution with technology is for each occupation. For example, two main sources of routine intensive tasks are blue-collar manufacturing and codifiable administrative support jobs.

With the proxy, the authors estimated the share of employment with area fixed effects, which are state effect dummy variables interacted with urban or non-urban area effects, and year fixed effect. The authors also measured differences in effects by age, race and sex. In an aggregate level of analysis, the paper measured effect on share of employment by different demographics and industries. A caveat here is that agriculture and mining industries are excluded due to low representation in many states. Using the data from Annual Social and Economic Supplement (“ASEC”) of the CPS, the authors were able to capture transitions to other jobs or non-employment with narrow occupation codes and broad industry codes.


A potential problem with these estimation analyses is “whether [the] wage variation is correlated with shocks to low-skill labor markets”. If endogenous policy affects the shocks, then the identification of causal effect may not be right. However, as the paper’s results are estimations of effects on a subgroup of low-skilled workers, it is very unlikely to have an endogenous policy chosen specifically for the group. The individual level analysis is even safer since it is controlled for “yearly shocks to states, and [the] urban and nonurban separately”. On the other hand, such restrictions may control for shocks correlated with minimum wages as well.

The authors found a 10 percent increase in minimum wage causes 0.31 percentage point decrease in the share of low-skilled workers’ automatable jobs. The effects are particularly large in manufacturing, transport and service sectors and significant. Manufacturing jobs’ elasticity in the shares is -0.18 while average elasticity of all industries is -0.10. Degree of drop in shares is also very strong among older groups, especially in manufacturing. Younger generations are less sensitive in general and more elastic in-service sectors.

Low-skilled labor market with low entry barrier can be either reduced in size or inundated with excess supply of workers from other sectors. According to Mckinsey research, as of 2015, 64% of the 749 billion working hours spent on manufacturing-related activities globally were automatable with current level of technology. The automation potential chart confirms the paper’s estimate of low-skilled workers’ job security after minimum wage uprating: Manufacturing (60%) and Transportation and warehousing (60%).


In the paper’s estimate, employment retention rate also declines largely in manufacturing, and it is significant in services. There is not much correlation between two industries with significant results, yet the retention rate varies significantly by demographics. Especially in manufacturing, older and younger workers’ implied elasticities of the probability of becoming unemployed are -0.28 and -0.17 each. This highlights a vulnerable group, older workers, that has been often ignored in previous literature.

Older workers are vulnerable to employment shock because there is less room for improvement or attainment of human capital compared to younger workers. Returns to investment in human capital decrease over time, and it eventually becomes lower than the opportunity cost of funds; unemployed senior worker is unlikely to find another job. As a worker gets older, it becomes more costly to invest in skills, and his or her human capital diminishes over time. When a firm looks to save costs by reducing labor, low-skilled elder workers are likely to be chosen due to these reasons. According to Coile, Levine and McKnight (2014), workers nearing retirement age are generally more likely to have health problems, too. Potential employers’ reluctance to hire them extend period of unemployment and end of company sponsored healthcare coverage increase health risk of people age 55 or older.


In terms of job-switching, the paper’s results indicate significant adverse effect of minimum wage with elasticity of the probability of losing or switching, -0.15. Effects on all the industries except retail are significantly negative. Higher minimum wage seems to push low-skilled workers at automatable job to move to service and retail as well. This implies reduction of those jobs in other markets and excess supply in service and retail labor market.

The negative effects have grown over the years in general which reflects increased influence of automation due to easier accessibility or perhaps higher levels of minimum wage. Relatively larger effects on oldest workers attribute to lower probability of retention in vulnerable jobs and unwillingness to switch to safer occupations. Higher skilled workers, those work at places with high share of low-skilled automatable jobs, are more likely to benefit from the automation. Significantly positive results for youngest and middle age groups but oldest groups, again, imply that jobs that utilize the automation require skills that are more difficult for older workers to obtain.


European Central Bank’s research suggests that the most popular adjustment channels in response to minimum wage are cuts in non-labor costs, product price rise, and improvements in productivity. To save labor costs, firms tend to reduce hiring rather than laying off directly. However, direct layoff is not the sole effect of minimum wage on employment. Increase in productivity of capital can cause adjustment of labor and capital thereby causing negative employment effect. For instance, fast food franchises, which are major employers of minimum wage workers, have been installing kiosks for higher efficiency. McDonalds is planning to install them to most of its 14,000 locations by 2020. They will likely replace cashiers and reduce training costs of entry employees. As restaurant business is labor heavy, the firm is more elastic in response to labor cost change.

Likewise, retail industry is going through a huge transition of automation. Similar to Amazon Go, which will enable automatic bills without waiting in lines for store personnels, Walmart’s Project Kepler will offer cashier-less service at stores, give product recommendations and facilitate purchases through messaging. Amazon’s recent move to raise its standard wage to $15 an hour and its lobbying effort for federal minimum wage raise will likely pressure Walmart, Target and Costco to match their wage to retain and attract employees. In contrast to other firms that have thousands of physical outlets with in-store workers, the e-commerce giant takes much less damage in raising their wage. The more automated a company is, the less likely it will be hurt by higher labor cost, or higher minimum wage. Simultaneously, firms which had higher share of labor will look to decrease it through innovation.


According to Autor and Salomons (2018), “labor’s share of value added was steady or rising in the 1970s” and “fell steeply in the 2000s.” Based on data of 28 industries of 18 OECD countries since 1970, labor-share displacing effects of productivity growth have become more pronounced over time, especially in 2000s. Estimates of the paper confirm that “automation [has become] in recent decades less labor-augmenting and more labor-displacing.” Two other industry-level measures of automation and technological changes, patenting flows and adoption of industrial robotics, also confirm the same result. Thus, big corporations’ effort to automate their business in response to higher labor cost is displacing labor even more today.

Innovation is inevitable. Firms develop new technology and apply it to boost their profits. During this process, skilled workers may find more opportunities in newly created markets while medium or low-skilled workers may be displaced from original jobs through automation. Minimum wage is one of main public policies intended to guarantee these workers higher standard of living, but it does not necessarily protect them from labor displacement. Perhaps, its uprating accelerates conversion to capital. Business climate is changing faster than ever, and its trend of increasing share of capital to produce leaves low-skilled or older workers vulnerable. As a new era of innovation dawns, society now calls for urgent and innovative approaches to fix the current labor market.


  1. EPA (Mike Nelson)
  2. Reuters (Lucas Jackson)
  3. US Government (Department of Labor)
  4. Card and Krueger (1994)
  5. Industry Today
  6. US Bureau of Labor Statistics; Mckinsey Global Institute Analysis
  7. Getty Images (Paul Taggart)
  8. Getty Images (Leon Neal)
  9. Wall Street Journal (Kevin Hagen)
  10. Lordan and Neumark (2017)

Work Cited:

Autor, D., & Salomons, A. (2018). Is Automation Labor-Displacing? Productivity Growth, Employment, and the Labor Share. Brookings Papers. doi:10.3386/w24871

Bader, H. (2019, March 04). Minimum Wage Hikes Are Wiping Out Jobs. Retrieved from

Bodnár, K., Fadejeva, L., Iordache, S., Malk, L., Paskaleva, D., Pesliakaitė, J., . . . Wyszyński, R. (2018). How do firms adjust to rises in the minimum wage? Survey evidence from Central and Eastern Europe. IZA Journal of Labor Policy,7(1). doi:10.1186/s40173-018-0104-x

Card, D., & Krueger, A. (1994). Minimum Wages and Employment: A Case Study of the Fast Food Industry in New Jersey and Pennsylvania. American Economic Review,84(4), 772-93.

Ingraham, C. (2018, January 11). What does a $15 minimum wage do to the economy? Economists are starting to find out. Retrieved from

Lordan, G., & Neumark, D. (2017). People Versus Machines: The Impact of Minimum Wages on Automatable Jobs. National Bureau of Economic Research.

Neumark, D., & Wascher, W. (2000). Minimum Wages and Employment: A Case Study of the Fast-Food Industry in New Jersey and Pennsylvania: Comment. American Economic Review,90(5), 1362-1396.

What happens when older workers experience unemployment? : Monthly Labor Review. (2014, October 01). Retrieved from

Wood, P. (2019, March 21). Maryland lawmakers give final OK to increase minimum wage to $15 an hour. Retrieved from

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s