Bryam Guazhambo, Tysne Ruggio, and Raymond Wheeler
Introduction
‘The implementation of more efficient production technology has been a major force behind income inequality in recent decades. According to OECD, GDP per hour worked in the U.S. has increased from 49.8 in 1970 to 103.3 in 2010 (OECD.org). Assuming material inputs have remained relatively constant over the same period, it is evident that it is firms’ abilities to use and mix these materials more productively that drives growth. The understanding of how to utilize materials in the most efficient way is generally thought to be a product of technological change, resulting from experimentation and scientific investigation (Romer, S72).
When examining wage growth during the same period, however, it is observed to be noticeably less than productivity growth. In fact, since the early 1970s inflation-adjusted wage growth has only been increasing on average .2% per year, compared with the roughly 2% percent growth in GDP per hour worked (Shambaugh, Dunn, np). The likely cause of such different growth rates is the rise of capital-intensive production. Firms no longer need to invest as heavily into labor as they previously did, as many aspects of production have become more automated. In fact, this is apparent when analyzing labor’s share of income since the mid-1970s, as it has dropped from 65% to 57% by 2017 (Shambaugh, Dunn, np). Such a decrease is surely intentional, to decrease costs through automated production systems by investing heavily in capital and R&D relative to labor. This is critical when examining income inequality because technology advances tend to substitute low-skilled labor while complimenting high-skilled labor. Thus, the demand and wages of low-skilled workers fall, while demand for high-skilled workers is rising, resulting in an increase in wages.
The relationship between technological growth and different skill groups is evident when examining trends in the labor force. Manufacturing jobs in the U.S. are becoming increasingly hard to find, as much of the production is automated, and the remaining portion of production is offshored to lower skilled cheaper labor. In many cases, the portion of production left to U.S. workers requires high human capital, as it requires monitoring and developing high skill-biased technology. On the other hand, many service sector jobs remain unharmed by technological advances and are actually complemented by more productive equipment and processes (Autor, 1077). Overall, the influence of technological growth on labor demand has a clear and negative impact on income inequality as a whole, since firms have an incentive to invest more intensively into capital relative to labor.
Endogeneity of Technology
The decrease in labor’s share of income supports the idea of endogenous technological growth. Endogenous technological growth suggests that technological advances are fueled by profit-maximizing agents, rather than being an outside force affecting the economy (Romer, S72). For example, when analyzing companies’ financial statements, such as Microsoft, it is evident that a significant portion of their revenue is invested in R&D expenditures. In fact, Microsoft invests a little over 13% of their annual revenues into research and development on average, only being exceeded by sales and marketing expenses (16%). Such expenditures further confirm that technology is not simply an outside factor, but it instead is a way for firms to attempt to outperform competitors, even if the advantage is only short term. In regard to income inequality, endogenous growth promotes the use of high-skilled capital intensive production, while disregarding low-skilled workers. Micro-economic theory states that efficient production is the product of optimally allocating capital and labor relative to their marginal product. For example, as more efficient technology substitutes low-skilled labor, firms will increase their inputs more heavily toward capital. Consequently, firms will invest more technology and skilled workers who can develop technology, resulting in stagnant wages for low-skilled labor. As long as an investment into factor-augmented technology and skilled workers results in higher profits than production be dominated by low-skilled workers firms will invest heavily into capital and educated workers rather than low-skilled labor. Overall, the concept of endogenous growth supports the idea that skilled firms have an incentive to increase the wages of skilled workers relative to unskilled, widening the income gap.
International trade also drives endogenous growth and the effect of technology on income distribution (Romer, S98). As noted earlier, under endogenous growth firms have an incentive to keep wages stagnant for the low-skilled labor. However, in many cases companies go a step further and offshore the low-skilled portion of production, completely disregarding the domestic workers. An example would be manufacturing workers in the United States since the 1970s. Not only did automation reduce the demand for many low-skilled workers, but firms also benefited from low wages abroad, such as numerous South East Asia countries. Thus, a major portion of the U.S. workforce became unemployed, having to choose between educating themselves for a new career or taking another low-level position in a different industry. International trade theory explains countries and firms can gain from identifying and utilizing their respective comparative advantages. For example, the U.S. has a comparative advantage in many aspects of information technology as well as high-skilled labor. On the other hand, China has a comparative advantage in its production facilities and low-skilled labor. Therefore, a natural incentive occurs for U.S. firms to focus on the utilization of high-skilled labor and computer technology during the domestic portion of production, while, as noted previously, exporting low-skilled production to China.
Endogeneity of Labor and Economic Mobility
Endogenous growth does not only assume firms attempt to maximize profit through technological growth, but also that individuals attempt to meet the demands of firms through obtaining more education and skills, increasing their individual productivity. Thus, supply for high-skilled labor increases in accordance with the demand. In the manufacturing example above individuals responded to the shifts in inputs of the manufacturing industry, having the choice of becoming more educated or finding other low-skilled work. The apparent rise in skilled labor since the 1970s supports this idea, individuals respond to the changing technology and needs of employers to become a compliment to technology rather than a substitute. This was seen in the U.S. during the 1980s. Both job availability and wages fell for jobs requiring low-skilled workers, while employment and wages rose in positions requiring high-skilled labor (Autor, 1053).
However, a significant amount of resources, as well as an adequate level of ability, is required to become high-skilled and benefit from technological change. As a result, children of low-income families may have a disadvantage, and will likely be unable to gain the human capital to become a skilled worker. Therefore, technological growth not only has a significant impact on income inequality, but also on intergenerational mobility, or the chance of a low-income child moving to a higher income level. Besides the obvious monetary costs of education, two additional resources affecting the relationship between technological growth and mobility. The first is the amount of human capital the parent has that can be passed on to the child. The second component is the level of ability the individual has (Galor, Tsiddon, 1). For example, if a high human-capital parent is heavily involved in their child’s life, then it is likely the child will have a better grasp on complex topics and processes, especially those that pertain to their parent’s industry. Also, the child’s own ability is crucial, because if they do not have the capacity to learn the techniques passed down from their parents and other forms of education they will be unlikely to succeed. However, these components are largely affected by the technological landscape at the time, and their relative impacts on intergenerational mobility may differ considerably.
The extent and severity of how these variables affect intergenerational mobility depend largely on the pace of technological growth during the period. Galor identifies two possible environments of technological growth, periods of inventions and periods of innovations. Invention periods, the more growth propelling of the two, refers to times where major technological breakthroughs are taking place, whereas innovation periods are times when previous inventions are being adapted to and becoming more accessible (Galor, Tsiddon, 2). The two variables, parental human-capital, and ability have contrasting effects in each period, as returns to ability differ.
Galor finds that innovation periods for a sector reduce the return to ability and increases rewards for the less able. The decrease in return to ability is largely a result of the accessibility granted by innovations. For example, as accessibility features are added to complex machines, high-skilled labor is no longer needed to operate it. On the other hand, less able workers become more productive, since the machine produces the same output with and without the innovation, and now the low-skilled labor is able to operate it (Galor, Tsiddon, 21). Overall, innovations have contrasting effects. The accessibility of the technology decreases mobility since ability it is not a factor, and high-skilled and low-skilled workers will be operating the same technology. In essence, opportunities for high-ability individuals are virtually the same as their low-ability counterparts (Galor, Tsiddon, 20). Additionally, innovations increase income equality for the same reason. Furthermore, as returns to ability diminish, high-skilled labor will leave the sector, decreasing the average ability in the sector. Consequently, with less high-skilled labor, technological breakthroughs will occur less often, decreasing economic growth (Galor, Tsiddon, 30).
Inventions, on the other hand, increase returns to ability, as only high-skilled labor will have the knowledge and experience to utilize it. Furthermore, breakthrough inventions increase economic mobility. Unlike innovations, workers able to utilize inventions rely heavily on ability, rather than human capital generated from their parents. Therefore, the conditions of their childhood and what their parents taught them are not nearly as important, instead, their ability to benefit from the new technology depends largely on their own skills and experience. Thus, a child from a low-income family has a larger chance of moving up in income level in an invention period rather than a time of innovation (Galor, Tsiddon, 22). Additionally, income inequality rises, as only a relatively small amount of the workforce has the capabilities to understand and operate the new technology, while the low-skilled workers operate older technology and earn stagnant wages. However, the invention period does not last long, as innovations are quickly made to make the technologies more accessible for low-skilled workers. Thus, the innovation period begins and offsets changes made in the invention period, and lowers economic growth. Overall, inventions increase returns to ability, increasing intergenerational mobility and income inequality (Galor, Tsiddon, 30).
Skill-Biased Technology
Educated workers wages have continued to increase at a faster rate than those of uneducated workers in recent decades despite the large increase in the supply of such workers. The primary explanation for this phenomenon is that computer technology is naturally skill-biased. Skill-biased technology is referred to as technology that favors skilled workers over unskilled, thus increasing demand for skilled workers relative to unskilled. David Autor, a professor at MIT, identifies several key effects of skill-biased technological change. First, low-skilled workers have experienced decreased income in the past four decades. Second, Autor discovers that the wage gap has been steadily increasing the past 20 years, even in times when returns to skill have fallen. Overall, it evident that skill-biased technological growth has a major effect on wage distribution and investment decisions.
Some of the most important and productivity-enhancing innovations are those that substitute workers for machines, such as artisans being replaced by factories and assembly lines. In recent decades, computer technology has given firms the option of developing various automated systems. For example, many low-level has been eliminated through data management systems, self-checkout technologies, and many other efficient computerized processes. In general, automation replaces the unskilled workers, while the skilled workers remain to develop new systems or are perhaps in a less substitutable job (Autor, 1116). As noted earlier, firms have recently been decreasing labor’s share of income, investing more in technology. In regard to information technology, investment has increased from 6% in 1960 to 40% in 2000 as a portion of U.S. private non-residential fixed investment (Violante, 4). The large share of investment in software and more efficient equipment not only increases production and quality but is also acts as a compliment to skilled labor. Several studies identify a correlation between the implementation of computer technology, the share of skilled labor relative to total labor, and their wage share. A possible explanation technology compliments skilled labor is that educated workers have already the experience and knowledge and adapt to further innovations with less costly training. As more technologies are adapted the further high-skilled workers’ knowledge diverges from their low-skilled counterparts until the low-skilled worker is only used for routine tasks which may soon be automated. Thus, the wages of high-skilled workers increase with their knowledge gained while low-skilled workers are stuck in a cycle of trying to catch up with technology standards and being stuck low-wage routine jobs.
Additionally, recent technological advances allowed for the consolidation of many specialized functions, decreasing the portion of routine tasks. Instead, firms shifted their organizational structure to broad teams responsible for a diverse set of complex tasks (Violante, 6). Consequently, only the workers who are innovative, adaptable, and capable of complex operations remain. Therefore, technology advances create a naturally skill-biased organizational change, leaving routine worker’s wages stagnant (Violante, 6). Overall, technological change inherently divides labor into routine and non-routine workers. Non-routine workers continuously adapt to the newest technologies, increasing their productivity and wages, while unskilled, inexperienced workers are stuck with routine tasks, associated with stagnant wages and the threat of automation. In the coming years, automation and artificial intelligence will likely have a massive impact on the remaining routine workers. For example, customer service employees are already being replaced by voice recognition software that can interpret a caller’s message and respond accordingly. Some other routine jobs likely to be affected include warehouse workers through inventory processing systems and automated machinery, retail workers through self-checkout machines, and many other occupations that can become efficiently automated at a low cost.
To summarize, technological growth is driven by profit-maximizing firms, creating an incentive to invest heavily in high-skilled labor to develop and operate technology. Thus, low-skilled labor observes both a loss in job availability and income levels, widening the wage gap. Simultaneously, individuals act as profit-maximizing agents and obtain human capital accordingly. In fact, an increasing portion of the population is becoming college educated, while at the same time high-skilled college-educated workers wages have been increasing. Thus, it can be concluded that technology is naturally skill-biased, favoring high-skilled labor over low-skilled. This is largely due to the notion that technological advances separate labor into routine and nonroutine workers. Nonroutine workers are generally high-skilled and expected to solve complex problems or take on greater responsibility, granting them a relatively high income. On the other hand, routine workers are largely unskilled and do not require high wages, widening the wage gap between the two. Also, intergenerational mobility is largely influenced by technology, with the effects differing between periods of technological breakthroughs and innovations. Innovation periods generally decrease mobility and increase equality, since the return to ability falls and high-skill and low-skill each have similar opportunities. Contrarily, breakthrough periods increase mobility and decrease income equality, as the return to ability in the period is high, as only high-skilled labor can operate and understand the new technology.
Technological Impact on Income Inequality
Research and development has led to a significant decrease in labor’s share of income since the 1970s (7%). This has been mainly through technological advances which improves efficiency, as well as the introduction of automation. Automation has played a huge part throughout history on the job market. Automation has been used for both substituting for human labor and complementing it. For example, the textile industry in the 19th century was an employer of strictly human labor. The introduction of various machinery led tasks to be done much more efficiently. One person was then able to perform the job of many others with the help of machinery. It will be determined whether it’s feasible for technology to replace human labor as well as the implications this would have on income inequality. One conclusion determined is that computerization could be confined to low-income occupations (Frey, 2016). Also, it’s suggested that less than 5% of jobs can become fully automated (Mckinsey). There is a possibility it could force these workers to develop new skills that are irreplaceable by machinery, so perhaps these skills developed would reduce income inequality. Or, instead of it replacing human labor, what if it serves as a complement? No matter which side of the argument, it’s clear to see there will be a shift in the kind of work human labor will be doing. Using the results, it will be determined if there will be an increase or decrease in income inequality
Will technology replace human labor?
Technology has replaced human labor in the past and will almost certainly replace some human labor in the future. The extent to how much of human labor will be replaced is still up in the air, however. Some machines are made specifically to replace human labor, others are made to help with costs of production, but in the end, they are all going towards the same goal. When you consider this idea that machines will either replace human labor by speeding up the work process or simply automating it, there are some variables that you can consider. The first is the speed at which the machine can work. There will be some industries, like agriculture, where a machine will be invented that is much more efficient than a human was. It is notable that a vast majority of workers will be wiped out from that invention because of its efficiency and convenience. Industries with machinery that is efficient as the agricultural machinery human labor will be much less needed. In 1900, 41% of the population was employed in agriculture. By 2000, only 2% of the population was employed in agriculture (David 2015). 31 million people were employed in agriculture in 1900 and only 5.6 million people were employed in agriculture in 2000 despite an increase of ~200m people to the United States population. The introduction of machinery and efficiency had a clear and lasting impact on the amount of labor in that respective field. It replaced a vast majority of workers and left very few because less workers were able to do more with the assistance of technology. The excess workers were forced to flee to jobs in other industries.
It’s been shown in some cases that when technology helps humans become efficient enough, that the same amount of human labor will not be needed in that industry (David 2015). In order to reduce the price of something in order to sell more, efficiency is the key component. Building more for less is the end goal. If the human labor is cheaper than the cost of machinery as well as costs of running the machine, humans will be used. However, if there comes a machine that is efficient enough to work at a similar pace for less cost that machine will be used instead. For example, automated telephone lines. It’s more than likely that you’ll speak to one when you call just about any big business. There will be several prompts and it will guide you to the answer or subject you’re seeking simply by pressing numbers. This can be seen with several websites as well, who are working on bots that detect phrases and attempt to answer your question before transferring you over to a human. It’s true that the AI can not always answer the question you need to ask, but it does help eliminate the need to talk to multiple people. The AI can simply transfer you to a worker in the part of the company that you intended to speak to. To go along with that, the AI is improving at a rapid rate. Google for example, introduced an AI bot named Duplex at a developer conference that can make phone calls for someone (googleblog). It’s able to talk on the phone without sounding like an automated robot. It’s programmed to be able to have “complex conversation” and set up appointments. Between now and 2030 the demand for office workers is expected to fall by 20% due to automation (McKinsey 2017).
On average, age and skill levels are negatively correlated with technical automation which implies a skill bias. Even though there is a skill bias, there is almost no occupation that is completely immune to automation. “About 60% of all occupations have at least 30% of constituent activities that could be automated” (McKinsey 5). Some examples of these occupations include occupations that spend time collecting and processing data, store clerks, travel agents, as well as web developers (Manyika et al 1). Even CEO’s were estimated to have about 25% of their work automated. Some of the jobs that can be automated “… include analyzing reports and data to inform decisions, reviewing status reports, preparing task assignments” (Mckinsey 8). About 47% of total U.S. employment is at “high risk” of becoming computerized (Frey et al 1). Another source however says that less than 5% of these jobs can become fully automated and at least some human labor would be required (McKinsey 5). Some examples of occupations that are have less than a 5% chance to become automated include psychiatrists and legislators. Even if only 5% can become fully automated, this has negative implications regarding human labor. It refers to the idea that, like the example of agriculture given earlier, the number of jobs in a certain field can be diminished with the help of technology. Technological advances will bring efficiency and therefore reduce the number of workers needed. You can say that efficiency is a good thing, and those workers can be placed elsewhere. But one of the occupations that is at the greatest risk, service occupations, also happened to be the place where job growth has been the greatest in the United States (Frey et al 3). There are several potential consequences to the labor market that could come from technological advances in this industry.
One great example of technology almost completely replacing human labor is in grocery stores. Amazon Go is a checkout free grocery store. In this store there are some cameras and sensors that track what each customer takes from the shelves. Instead of having to wait in line to get cashed out, the customer simply leaves the store and are billed to the credit card that is on file (Dastin). This eliminates the need of multiple workers including cashiers. The store should only use 3 to 10 people which is a huge reduction in labor compared to some other grocery stores. On average Amazon would like to use 6 people per shift. There will be one worker which is validating ID’s in the section selling alcohol or tobacco, another will be stocking shelves and then there will be a guest services manager. By creating a store that is largely automated, Amazon can cut back on labor costs and remove a large part of human interaction. Going into 2021, Amazon plans to open as many as 3000 cashier-less stores (Soper 1). With Amazon being a huge leader in regards to innovation, this movement could have a huge impact in the direction shopping goes. The costs saved by the technology used in these Amazon Go stores can make it a major brick and mortar retail competitor. If you combine this with its online presence, competition is going to be forced to innovate and potentially adapt this or similar cost cutting technology.
Another example comes from China, where a factory has replaced 90% of its human labor with some sort of technology. This factory, named Changying Precision Technology Company is almost entirely run by robots. Since switching over to robots instead of human labor, the company has seen fewer defects as well as a higher rate of production (Forrest 1). This company primarily creates parts for cell phones. Among its technology includes automated production lines and automated transportation. The workers in this factory, similar to the Amazon stores, are there for the jobs automation can not do as well as maintenance. Some workers monitor the production lines and others monitor a computer control system. By cutting down from 650 employees to just 60, the factory actually increased its production by 162.5% (Forrest 6). In the case of factory production, there’s almost no job that a human can do that a machine or automated system can’t do more efficiently in regards to production as well as costs.
Replacing 90% of human labor is a huge feat, but Tesla has its eyes set on an even larger number. Elon Musk has promised fully autonomous robo-taxis by the year 2020. Ford has set its sight on a similar goal, aiming for full autonomous driving by 2021. If this feat is completed, then companies with drivers such as Uber, Lyft, as well as other taxi services will be 100% replaced by technology. Not only will Tesla take over a large portion of the market and wipe out thousands of jobs, it will be doing the humans jobs much safer! There has been one accident for every 3.34 million miles driven in autopilot in a Tesla, which makes it around 7x safer than the average human driver in a Tesla (Tesla Team).
Technological advancements have been influencing human labor for years and will continue to happen in the future. However, sometimes the advancement doesn’t lead to a direct substitute. After ATMs were invented, number of bank tellers per bank decreased by more than 33%. Technology was nearly a perfect substitute for this position. Although the number of tellers per bank decreased substantially, the number of branches per bank increased by 40% (Autor 6-7). This situation produces an alternate scenario, where instead of substituting human labor it acted as a compliment. This points to the idea that instead of substituting human labor, it could end up complimenting instead.
Will technology complement human labor?
In a study of around 1500 companies, it was shown that firms achieve the most significant performance improvements when humans and machines work together (Wilson, Daugherty 2018). When humans and machines work together, there are many benefits. There are three ways in which a smart machine can help humans expand their abilities, which include amplifying cognitive strengths, interacting with customers/employees to allow the human to focus on higher-level tasks and embodying human skills to extend physical capabilities (Wilson, Daugherty 2018). Artificial intelligence can amplify cognitive strengths by giving access to a large database of information in mere seconds. By interacting with customers and employees, as mentioned in the google example, it can lead to more efficient means of communication. There are some cases where this is not true, however AI can be relied on to answer a majority of the frequently asked questions and send the customer to a customer service representative if the machine can’t answer the question. In order to embody human skills, there are some cases where there are wearable robotic devices that can provide extra strength or can even automate heavy lifting (Wilson, Daugherty 2018). Some examples of benefits directly gained from artificial intelligence complimenting human labor include leadership, teamwork, creativity, social skills and speed (Wilson, Daugherty 2018). There are some things that cannot be easily replaced by robots such as social skills, and others that are easily replaced such as analyzing data. When you combine robots and humans, you can take the best qualities and put them together. This enables the firm to be more efficient than ones that use just humans, or just robots by having each of them specifically focus on what they are better at. This points to the idea that instead of outright replacing human labor, it should compliment it or at least have a combination of the two.
Unilever is an example of a company that is currently implementing a system of both AI and human labor working together. In their hiring process, there are three stages. The first stage is to play a series of games in which you are evaluated in several things, like your tolerance to risk as well as reaction time. The AI generates a score based upon several factors including how fast the questions are answered and the way they’re answered. If the first stage goes well, the next stage is a video interview. Predetermined questions are given, and a video recording is sent in of the applicant answering the questions (which is given a certain time restriction). The AI has the ability to analyze deeper than just the answer to the questions. It picks up on things such as the tone of voice and body language when answering the questions. The third stage is an invite to an interview by a human, who will make the final decision. The system is relatively new however due to the popularity of smartphones, Unilever was able to double its number of applicants by switching to this hiring method. (Wilson, Daugherty 2018). The hiring process became much more efficient by allocating the first two rounds of hiring to AI. If they were to use the same hiring process with humans proctoring the first and second round, it would take quite a few employees to interview 30,000 people. Instead, this labor is now freed up to work on other tasks better suited to human labor.
Two industries that became automized, textiles and steel were examined (Bessen, James E. 1). A specific pattern that is referred to as the “inverted U” shows that technology which increases productivity can not be directly correlated to job loss. The result is, “If demand increases sufficiently in response to productivity-improving technology, then employment will grow; otherwise, it will fall.”(Bessen, James E. 6) This is proven with similar results found in another case, where the introduction of ATMs reduced the number of bank tellers per bank by 33% (Autor 6). However, because there was a 40% increase in number of branches per bank the number of bank tellers increased from approximately 500,000 to 550,000 (Autor 6). In this case, it directly cut the number of jobs available but acted as a stimulus by increasing the demand for bank tellers by decreasing costs. The job of a bank teller shifted from primarily handling money, into various tasks in “relationship banking” (Autor 7). This shows that the technology forced these bank tellers into adapting to different jobs that weren’t susceptible to automation, like customer service. Although there are still bank tellers today who handle money, that number has decreased drastically since the introduction of ATMs. This example shows that although a near perfect substitute was developed, it increased demand for bank tellers to the point where employment grew.
Given these results, it’s clear to see a shift in the labor that is demanded will occur. There will be instances where technology will substitute human labor and there will be cases where technology will compliment human labor. Sometimes, even both can occur at once. As mentioned, the introduction of ATMs reduced the number of bank tellers by about 33%. This is a direct replacement of workers. However, once the demand increased due to the introduction of the ATM, the overall number of workers increased. This paved a transition path for bank tellers to focus mainly on customer service and the technology could focus on handling the money (for the most part). At first, human labor was replaced but soon after the technology complimented the human labor because of the demand shift. This demand shift plays a determining role into income inequality, as well.
Automations effect on Income Inequality
Automation has played a role in how income inequality has gone up. “… new technologies are by their nature complementary to skills, so there has always been some skill-biased technical change, and the recent past witnessed rapid introduction of new technologies, leading to an acceleration in skill-bias.” (Daron 1055). When there are more skilled workers, then there will be a larger market for skill complementary technologies (Daron 1082). For example, “The wages of college graduates and of other skilled workers relative to unskilled labor increased dramatically over the past fifteen years. To many economists and commentators, this is a direct consequence of the complementarity between skill and new technologies.” (Daron 1082). Therefore, automation will be correlated to an increasing skill bias. “An exogenous increase in the ratio of skilled workers or a reduction in the cost of acquiring skills could increase wage inequality. The likely path is first a decline, and then a large increase in the skill premium” (Daron 1083). This is exactly what happened in the 1960’s and 1970’s in the United States, where college premiums decreased and then increased to even higher prices than before. Given an increase in the supply of college graduates, research and development that was made to be beneficial to these graduates should increase (Daron 1083). This increasing skill bias is directly correlated to income inequality, which ties together automation and income inequality.
Another result varies depending on the economy involved. For example, in advanced economies occupations that require only a secondary education or less see a net decline from automation while occupations that require college degree see an increase in occupational growth (Manyika et al 2). When the increase in demand for occupations requiring a degree, the result is that income inequality will increase. Labor supply and demand leads to the idea that wages should stagnate or fall when labor demand declines. In emerging economies however, there will be a higher labor demand for all occupations involved (Manyika et al 2). Therefore, depending on the economy involved, income inequality may not be a direct result of automation.
There are several things to take into consideration for this next study by Hong and Shell. This is a controlled experiment where R&D costs, labor market adjustments or employment flow to industries. The 90-50 ratio is the ratio of income from 90th to 50th percentile, 50-10 ratio is 50th percentile to the 10th percentile, and the Gini Coefficient is a measure of income inequality (Hong, Shell 1). If a job has a 60% chance of being automated, then there would be 3 different results that could possibly happen with a 60% chance. The three hypotheticals are that for the first, income =0. The second will have income be reduced to minimum wage and the third is that income would be reduced by 20%. If all these factors are held constant, it is shown that no matter which scenario happens given a 60% chance of automation taking over an occupation income inequality will increase (Hong, Shell 3).
In conclusion, technology will both replace human labor and compliment it depending on the occupation. There are some occupations that are more likely to be automated than others such as data processing and others that will be complimented such as the hiring process at Unilever. It should be noted that if the costs to automate are greater than the cost of human labor, then human labor will be used instead. However, given the efficiency of production as well as the idea of economies of scale, numerous examples such as Amazon’s stores as well as Tesla’s robo-taxi fleet shows that there is likely to be several scenarios where the benefits will outweigh the cost. This leads to the conclusion that income inequality will grow with the implementation of automation. Using the results from Hong and Shell, the Gini coefficient or level of income inequality will rise in all scenarios given automation replaces human labor. Under the assumption that Tesla, Amazon and other companies will be able take advantage of economies of scale during their production, the cost of research and development will not outweigh the benefits. Labor market adjustments and employment flow to industries however still cannot be accounted for, so more research should be done in order to obtain a more accurate result.
The Role of Computers in the Labor Market
The role of computers can influence the relative labor demand in many ways. Computers, as well as other microprocessor-based technologies, have facilitated the automation of the process of production in the previous decades. Computer-based technologies have seen to provide many advantages, such as providing a bigger return, when using a greater amount of information to create more products and services that fit the specific needs of its customers.
The sector that is impacted the most by computers would be white-collar work (Bresnahan, 1997). The impact that computers have had on white-collar work is by taking jobs of organizing, routinizing and regularizing that people would do more intuitively but not as organized as computers would do, so substitution of computers for human labor seems to have a smaller impact on managerial and professional jobs and have a bigger impact on clerical and production jobs. computer-based production has shown high levels of service as well as new service products, although the labor-demand impact shows to be at the firm level since many computer business systems create the production process for many service industries (Bresnahan, 1997). A study at firm levels found that maximizing information technology can be connected with the rise of employment of more educated workers, as well as a bigger investment in training workers(Bresnahan, Brynjolfsson, & Hitt, 2000). Many economists believe that the role of computers is of a greater impact, the role of computers has not been used only to increase the demand of its product, but many believe that computers created a technological change that has altered the organization of work, which also affects the demand for workers with various skills (Autor, Katz, & Krueger). Computer technology has been extensively used since the 1950s and has not slowed down since. For instance, in previous decades we’ve seen microprocessors being used in manufacturing machinery since the 1970s. The increase in demand for personal computers leads to the creation of many products, such as the Apple 2 in 1977 and the IBM portable computer in 1981, these products allowed for rapid computerization in the decades that followed. With these products entering the market, historical trends suggest that computerization affected the relative skill demands in the service sector before the impact of the computer on manufacturing could be felt.
In order to compute how big the effect of computerization had on the relative demand for skills. A study was done by (Autor, Katz and Krueger) which analyzes computerization by looking at the spread of computer technology by examining the percentage of workers who directly use a computer keyboard. The framework does seem to have a negative aspect to it, as it doesn’t take account workers who use devices that have embedded microprocessors that are not operated by a keyboard. The framework uses data from the Current Population Surveys (CPS) that include data from October 1984, 1989 and 1993 with sample sizes of 61,704, 62,748, and 59,852. What the framework showed that the number of workers who used computers from 1984 to 1993 had increased 2.4 percent of the workforce per year. The results from the framework also provided an insight into which groups were affected the most, for example, more females used computers when compared to men, they have seen an increase of 21.6 percent of women using computers in 1984 to having 53.2 percent of women using computers in 1993. Although men also saw a rise in the usage of computers at work, seeing an increase of 20% starting in 1984 until 1993. Although the biggest difference in computer utilization can be seen in education and occupation. In education workers with a college or higher education saw an increase of 28.1 percent from the year 1984 to 1993. While workers with less than a high school education only seen a rise from 5.1 percent in 1984 to 10,5 percent in 1993, so there was an increase but a minimal increase. Regarding occupation, blue-collar workers did not see a much increase in using computers at work, there was only an increase of 10 percent between the years of 1984 and 1993. Meanwhile, white-collar workers saw a much bigger increase, being 39.7 percent in 1984 to 67.6 percent in 1993 (Autor, Katz, & Krueger).
With the increase in information technology, what percentage can we say that information technology has had on job polarization? One contributor to jobs polarization would be routine tasks replacing technological change. As the progress in technology has seen tremendous steps, it has also made information technology relatively cheaper than what it was before. Economist William Nordhaus predicts that the real cost of performing a standardized set of computational tasks using information technology has accumulated a decline of at least a trillion-fold in the cost of computing, as computational tasks using information technology has seen a decrease of about one-half annually in the last sixty years (Autor, 2010). As the price of information technology rapidly decreasing, it has lead to an environment where it creates enormous economic incentives for firms to replace information technology with expensive labor in workplace Tasks, while also creating advantages for workers whose skills rapidly become more productive as the price of information technology decreases.
For instance, from 1940 to 1970 the U.S. saw a rapid increase in automation and technology where physically demanding, repetitive, and dangerous work was decreasing, ushered out by incredible productive gains in agriculture (Autor, 2015), with the growth of technology many corporations, along with the increase in health care services and higher education created many jobs for skilled professionals, while also supporting many clerical and administrative jobs. although the rapid increase in technology seem to be altogether positive, after 1970s many of the occupations that had been complemented by technology had slowed down, with jobs at the top of the skill ladder such as professional and managerial occupations growing even more between 1980 and 2010 when compared with the time period between the 1940 and 1970 (Autor, 2015).
While jobs at the top of the skill ladder continued to increase, many blue-collar jobs began to decrease rapidly as many blue-collar occupations are routine task intensive, which happens to be the easiest tasks replaced by computers. As the labor market saw an increase in information technology, which can be said about the role of computers in job polarization. Computers, thanks to programmers who fully understand the steps required to perform a task, follow procedures and simulates every step precisely, therefore jobs such as processing a firm’s payroll, would be simulated by computers rather than humans. As information technology becomes cheaper, many computers and robots have substituted many workers in explicit, codifiable tasks. These tasks are known as “routine tasks” can be easily automated and have led to a big decline in occupations such as clerical and administrative support. For instance, routine intensive tasks include bookkeeping and factory jobs, these jobs can be computerized as we know the rules of these tasks such as copying and calculating, so with computerization, spreadsheets replace bookkeepers and robots replace factory workers.
Although computerization substituted many routine-intensive tasks, there are many tasks that computers or machines cannot substitute but complement different occupations. Such tasks, which are the most vulnerable to automation, also require skills that we only understand tacitly. There are two tasks that have proven to be extremely difficult to computerize, these are known as “abstract” and “manual” tasks. (Aurtor, Levy and Murnane) Abstract tasks which require creativity, intuition, and problem-solving capabilities that are found in “characteristics of professional, technical, and managerial occupations” (Aurtor, Levy and Murnane) This set of tasks is commonly found at the high-end tail of labor supply, where highly educated workers benefit from positive effects brought on by computers. At the other end of the labor supply, we find manual tasks, which are tasks that require “situational adaptability, visual and language recognition, and in-person interaction” (Aurtor, Levy and Murnane) we can find these occupations in food preparation and serving jobs, where workers are not highly skilled but physically adapt to perform many tasks that require to be performed on site, hence these manual tasks cannot be outsourced because they have a large supply of labor. Since occupation that are intensive in abstract and manual tasks are commonly found at opposite ends of the labor supply, computerization of routine tasks may lead to the growth of both ends of the labor supply, where there’s growth in high education, high-wage jobs, as well as growth in low-education, low-wage jobs, both at the expense of middle-wage, middle education jobs (Autor, 2015) therefore, contributing to the presence of employment polarization, although not all the blame should go to information technology, as there are many more factors that contribute to employment polarization.
Finally, does occupational polarization lead to wage polarization? There are three factors that that point towards no, those are “complementary, demand elasticity and labor supply” (Autor, 2015). For instance, the effect of computerization on wages in abstract tasks jobs are complemented by information technology and computerization. Tasks that include managerial, professional and technical occupations all need the constant help of many evolving complementary jobs, such as financial analysis and medical knowledge. These complementary tasks all help the highly educated worker to reduce cost as well as increasing the information available to them. Regarding demand elasticity, only an inelastic demand for the output of abstract tasks can harm wage gains, but evidence suggests that technology has increased the output of jobs, while demand for their services has kept pace (Autor, 2015). Regarding labor supply, since many abstract intensive jobs require higher education, many U.S. males have responded poorly to the rising educational premium. Since there was only an increase of 12 percent of men working with less than ten years of experience. Therefore, many abstract tasks are greatly affected by information technology where it represents a complementary relationship between routine and abstract tasks.
Although abstract tasks have shown not to produce wage polarization, what about manual tasks? Manual tasks act differently from abstract tasks as many manual tasks do not require the complement of technology, only on limited opportunities can we see a manual job being involved with complementary or substitution. As aggregate evidence shows that the final demand for manual tasks-incentive work that focuses on services, relatively prices inelastic, which suggests that gains from productivity in manual tasks that decreases the price per unit of service provided will not increase the costs on their outputs. Regarding labor supply in manual tasks, labor is elastic, mostly due to the low requirements and low education(Autor, 2015). In conclusion, manual task-intensive that are complemented by technology is very limited, if not any.
Overall, the role of computers and information technology has been a factor on job polarization, as we’ve seen technology substitutes and complements certain tasks, with creating jobs for both high-wage jobs and low-wage jobs at the expense of the middle-wage jobs. Although, technology is a factor in job polarization, the effect it has is not of big magnitude in contributing on income inequality. Regarding wage polarization, technological change does not play as a factor that contributes towards wage polarization.Technological advancements have contributed to income inequality, but the weight of that contribution is not of big magnitude,
Works Cited
Acemoglu, Daron, and David Autor. “Skills, Tasks and Technologies: Implications for Employment and Earnings.” Handbook of Labor Economics, vol. 4, no. b, 2011, pp. 1043-1171.
Acemoglu, Daron. Quarterly Journal of Economics, vol. 113, no. 2, 1998, doi:10.1162/qjec.1998.113.issue-2.
Autor, David H. “Why Are There Still So Many Jobs? The History and Future of Workplace Automation.” Journal of Economic Perspectives, vol. 29, no. 3, 2015, pp. 3–30., doi:10.1257/jep.29.3.3.
Autor, David, et al. “Computing Inequality: Have Computers Changed the Labor Market?” 1997, doi:10.3386/w5956.
Bessen, and James E. “Automation and Jobs: When Technology Boosts Employment.” By Jikun Huang, Ruifa Hu, Scott Rozelle, Fangbin Qiao, Carl E. Pray :: SSRN, 18 Mar. 2017, papers.ssrn.com/sol3/papers.cfm?abstract_id=2935003.
Bresnahan, Timothy F. “Computerization and Wage Dispersion: An Analytical Reinterpretation.” SSRN Electronic Journal, 1997, doi:10.2139/ssrn.68992.
Brynjolfsson, Erik, and Lorin M Hitt. “Beyond Computation: Information Technology, Organizational Transformation and Business Performance.” Journal of Economic Perspectives, vol. 14, no. 4, 2000, p. 48., doi:10.1257/jep.14.4.23.
Dastin, Jeffrey. “Amazon’s Automated Grocery Store of the Future Opens Monday.” Reuters, Thomson Reuters, 22 Jan. 2018, www.reuters.com/article/us-amazon-com-store/amazons-automated-grocery-store-of-the-future-opens-monday-idUSKBN1FA0RL.
Daugherty, H. James WilsonPaul R. “How Humans and AI Are Working Together in 1,500 Companies.” Harvard Business Review, 4 Apr. 2019, hbr.org/2018/07/collaborative-intelligence-humans-and-ai-are-joining-forces.
Forrest, Conner. “Chinese Factory Replaces 90% of Humans with Robots, Production Soars.” TechRepublic, www.techrepublic.com/article/chinese-factory-replaces-90-of-humans-with-robots-production-soars/.
Frey, Carl Benedikt, and Michael A. Osborne. “The Future of Employment: How Susceptible Are Jobs to Computerisation?” NeuroImage, Academic Press, 29 Sept. 2016, www.sciencedirect.com/science/article/pii/S0040162516302244#bb0075.
Galor, Oded, and Daniel Tsiddon. “Technological Progress, Mobility, and Growth.” American Economic Review, 1 Oct. 1996, pp. 1-35.
“GDP per Hour Worked.” OECD, 2018, data.oecd.org/lprdty/gdp-per-hour-worked.htm. Accessed 10 Apr. 2019.
“Google Duplex: An AI System for Accomplishing Real-World Tasks Over the Phone.” Google AI Blog, 8 May 2018, ai.googleblog.com/2018/05/duplex-ai-system-for-natural-conversation.html.
Hong, Sungki, and Hannah G. Shell. “The Impact of Automation on Inequality.” Economic Research – Federal Reserve Bank of St. Louis, research.stlouisfed.org/publications/economic-synopses/2018/11/19/the-impact-of-automation-on-inequality.
Manyika, James, et al. “A Future That Works: Automation, Employment, And Productivity.” Mckinsey & Company, Mckinsey Global Institute, Jan. 2017, www.mckinsey.com/~/media/mckinsey/featured insights/Digital Disruption/Harnessing automation for a future that works/MGI-A-future-that-works-Executive-summary.ashx.
Manyika, James, et al. “Jobs Lost, Jobs Gained: What the Future of Work Will Mean for Jobs, Skills, and Wages.” McKinsey & Company, www.mckinsey.com/featured-insights/future-of-work/jobs-lost-jobs-gained-what-the-future-of-work-will-mean-for-jobs-skills-and-wages.
Romer, Paul. “Endogenous Technological Change.” NBER Working Papers, no. 3210, Dec. 1989.
Shambaugh, Jay, and Ryan Nunn. “Why Wages Aren’t Growing in America.” Harvard Business Review, 24 Oct. 2017.
Soper, Spencer. Bloomberg.com, Bloomberg, www.bloomberg.com/news/articles/2018-09-19/amazon-is-said-to-plan-up-to-3-000-cashierless-stores-by-2021.
Violante, Giovanni L. “Skill-Biased Technological Change.” The New Palgrave Dictionary of Economics. Originally published in The New Palgrave Dictionary of Economics, no. 2, Jan. 2016.