Book Navigation Links
A Look At The Skill-Biased Technological Change Paradigm And Why It Fails To Adequately Explain Labor Market Shifts | Varun Palnati
Varun Palnati is an Economics major and Labor Studies Minor at UMass Lowell. He is very interested in labor economics, the effects of automation, and econometrics, and hopes to pursue a graduate degree, preferably a PhD. He teaches chess in his spare time.
**First Place Award for Research**
Abstract: The main topic this paper addresses is the paradigm of Skill-Biased Technological Change (SBTC), which describes a labor market theory concerning low and high skilled workers’ situations in the job market after new technology is introduced into the market, and how the theory holds up in the real world. This is important because automation and the effects of it are a topic that has many people concerned about the impact it could have on their jobs. By looking at how the job market for high-skilled workers has evolved over time, I hope to draw solid conclusions about how automation affects highly skilled workers, and whether the effect corresponds to the SBTC paradigm, which states that highly skilled workers should see benefits from automation. The sources I used to examine this paradigm were a paper by Autor et al. describing the SBTC model and what the expected results should be, and two papers by Lazonick et al. and Schmitt et al. which describe the ways that highly skilled workers have been affected by the introduction of new technology into the marketplace. Overall, by looking at the results of the two papers that detail how the labor market has affected highly skilled workers, a conclusion can be drawn that the SBTC paradigm does not seem to be improving the lives of high skilled workers and that they are seeing little to no benefit from automation. These results are both contrary to previous introductions of new technology into the marketplace and to mainstream economic theory, which states that highly skilled workers usually benefit from the introduction of new technology since they are generally complements to that technology, unlike low skilled workers, who tend to be substitutes.
Recently, many people have been concerned that automation will take their jobs, or will make their current jobs obsolete. Also, the disappearance of the middle-class and other forms of job loss have caused the center of the job market to drop out, leaving a void in the job market. Many people believe that technological shifts, such as the computer and digital technology, which is a form of automation, have been responsible for this change in the jobs available for people. Many economists have been analyzing this idea, and they have come up with a catchall term to explain this, called skill-biased technological change (SBTC). This idea, which states that shifts in technology increase the production capacity of higher skilled workers more and therefore the job market demands more of them, seems reasonable in theory, but upon closer analysis of labor market trends, does not seem to explain the overall shifts in the job market very well. Therefore, the aim of this paper is to address the issues with the concept of skill-biased technological change by looking at some of the common evidence for and against the idea.
One view in favor of SBTC is provided by David Autor, in his paper on the shape of employment growth and Polanyi’s Paradox.1 In his paper, he describes how earnings changes are related to technological change and automation by discussing Polanyi’s Paradox, which causes the simultaneous growth of high-wage, high-education jobs and low-education, low-wage jobs. He describes Polanyi’s paradox as the constraint on substituting computers for workers. If a worker is not performing a task that has clearly defined rules but instead requires tacit knowledge, computers cannot substitute for that person’s work since the rules of the task are not defined clearly enough for computers to replicate it. Tacit knowledge is defined as knowledge that is understood implicitly but cannot yet be replicated by current coding technology.2 Some examples are identifying specific species of birds and writing a persuasive paragraph - things that can be understood implicitly but difficult to describe and quantify in a way that computers can do it.
Computers can still affect tasks that require tacit knowledge, however, by complementing them, making them easier to perform or improving the overall output of those tasks indirectly. So in cases where computers are able to complement tasks, workers can actually benefit from their presence, since they make the jobs performed by the workers easier and more efficient,3 such as jobs like accounting and construction work.4 Basically, computers benefit workers who supply tasks that they cannot reproduce, which leads to the conclusion that computers can both help and hurt workers. If they are substitutes for the work that workers perform, they will displace them, but if they complement the work workers perform, then they are beneficial to the job market overall, as productivity growth raises the value of the tasks that only workers can perform. Therefore, the demand and hence the wages paid to the workers who have the skills to perform such tasks will rise.
This overall trend, however, has reduced the number of middle-skill jobs overall, as Autor has found. He states that the trends of quick employment growth in both low and high skill jobs have seen the bottom drop out of the middle of the job market, causing a sharp decrease overall in the availability of formerly middle-class jobs.5 This is clear evidence of a wider shift towards the overall growth of jobs that computers complement, which seems to be jobs on either end of the scale. This is causing a shrinking of the availability of middle-class jobs and therefore lowering the number of middle-class people in the country overall. This development is summarized as the “job polarization hypothesis,” which suggests that as a result of SBTC, the overall job growth in the economy has grown at either end of the spectrum, causing a simultaneous increase in high-skill, high income jobs and low-skill, low-income jobs.6 Skill-biased technological change postulates that there will be a decrease in the demand for jobs that require routine skills that computers can do, depressing the wages and numbers of those jobs, while it increases the demand for sophisticated skills that enable workers to perform tasks that complement computer technology. When growth in the demand for computer-era skill outpaces the growth in the supply of college-educated people with such skills, the wages of these college-educated members of the labor force will rise.7
There are, however, some major issues with skill-biased technological change, as evinced in a paper written by Lazonick et al. where they compare the trends predicted by SBTC to what actually happens for STEM workers, the workers who should ideally benefit the most from the gradual shift towards computers, since the members of the STEM fields work in some of the most high-skilled jobs around, and there is a shortage of them overall.8 Therefore, if STEM workers are able to find consistent employment at high wages, SBTC has successfully created a framework for measuring job generation in the real world; if they cannot, it is a strike against that framework. They found that the current careers of STEM workers are characterized by less employment security, shorter job tenure, and declining returns to STEM education than SBTC would predict.9
Skill-biased technological change considers education to be the primary means of skill development, in essence implying that skills are determined outside the context of employment. Education level determines whether or not the labor market outcomes of a particular group will be improved. SBTC ignores the possibility of on the job training being relevant to greater wage increases. As Lazonick et al. say, most scientists learn to provide value over time by working on the job and receiving training.10 They require sustained employment in learning environments after the classroom to obtain the higher wages that SBTC would predict they would. This directly contradicts SBTC since scientists and other STEM people do not generate value for their employers directly from their education, which SBTC uses as an important indicator of which jobs will receive increases. It also contradicts Autor’s idea of job polarization, since this shows that one of the most important groups of high-skilled workers has not received the benefits he envisions they would under SBTC, and also shows that while there may be growth in overall high-skill, high-wage jobs, STEM workers have not received increasing benefits, contrary to what SBTC would have predicted for them.
Also, a report by the Economic Policy Institute (EPI) found that there has been no real wage polarization between the lower end of the distribution and the middle.11 In addition, they found that changes in occupations do not explain much, if any, of the discrepancy in wages. They found that the share of wage variation explained by occupational differences has actually declined in the 2000s after increasing very slowly through the 1980s and 1990s.12 Since SBTC predicts an overall growth in jobs depending on how those jobs are affected by computers along with wage increases for those jobs, to have occupational differences not explain as much of wage variation as it did before computers became widespread is directly contradictory to the theory. Also, Autor’s job polarization hypothesis relies on the differences created by computers affecting occupations. To have the explanatory power of wage differentials decline when comparing them suggests that occupations are not as good a predictor of wages as SBTC says they should be. In addition, the EPI paper found no wage polarization between the middle of the income distribution and service wages.13 Since Autor intends to explain the growth in the lower end of the distribution by looking at the overall increase in wages and employment in service jobs, the fnding that there was no real difference between the middle of the distribution and service wages blows a hole in the idea that SBTC is contributing to job polarization. Overall, the lack of the predicted wage changes on both ends of the spectrum, when considering scientists and service jobs, clearly shows that STBC and job polarization, while appealing, simply do not ft the data.
Another conceptual framework that provides more problems for the SBTC paradigm are the ideas of the new economy business model (NEBM) and the old economy business model (OEBM) as postulated by Lazonick et al. The old business model consisted of a career with one company for the entire length of one’s working life, complete with promotion opportunities up the chain and a retirement package when the worker retired. This fell out of favor when new tech companies offered employees greater up-front pay in the form of stock options in exchange for a loss in benefits and job tenure, which became the NEBM.14 This ended up being a trade that many tech employees were happy to make, so companies that had been successful with the OEBM, such as Intel and Hewlett-Packard, transitioned over to the NEBM to cut overall costs. However, this ended up affecting new workers quite adversely. First, this transition ended up reducing the ability of scientists and other high-skilled jobs to receive the training they needed to make a successful transition from college to the workplace. Before, they could stay at one company and learn what they needed to there since the company could safely train them without worrying that they were going to leave. However, under the NEBM, companies are disincentivized to do that because their employees that they spent time training could leave for another company that will give them more upfront pay under the NEBM. This contributes to keeping wages overall of high-skilled laborers lower than they should be since they are not receiving the training necessary to increase their wages and take advantage of the introduction of computers and other cutting-edge technologies. This directly repudiates the idea of SBTC since these high-skilled workers have not seen any benefits from an increase in technology since they lack the skills to benefit. Having an education is not enough for them to take advantage of the benefits that computers provide to their job; they need to receive on the job training in the form of group learning that comes from sustained career employment to learn how to integrate these computers and new technologies into their jobs. Also, interestingly, they fnd that workers’ expected earnings decrease over time. They found that the adoption of the NEBM placed the careers of high-tech workers in jeopardy when they reached 40 or 50, when one would expect them to be at their most productive under OEBM.15 This shows that over time, since the skills they have learned are becoming more obsolete, that high-skilled workers are actually experiencing a decrease in their expected earnings over time, which contradicts what SBTC would expect since it equates education and more time in the workforce with potential wage increases. Also, the fact that employees tend to get paid in stock also incentivizes companies to focus on using their profits to pump up their stock prices rather than using it to invest in their employees, as Lazonick et al. show. Pfizer and Merck, two of the biggest of the pharmaceutical companies, have spent 66% and 42% of their net income on stock buybacks, and another 60% and 58% on dividends.16 This reinvestment of their profits into jacking up their stock prices shows that overall, companies are no longer trying to create jobs and increase output with their profits. As a result of the NEBM, which incentivizes employees who hold large amounts of stock to work on jacking the prices up, job growth in these sectors has been minimal. This minimal growth in the STEM sector directly contradicts what would be expected from SBTC and the job polarization hypothesis, which would expect large growth in high-skilled jobs, which has not happened. Overall, the tendency of companies to shift from the OEBM to the NEBM has shifted their focus from training their workers to boosting their stock prices, which means that high-skilled jobs are not seeing the payoffs that STBC and the job polarization hypothesis would expect.
There are more empirical issues with SBTC, such as the idea that technological change may not even be the primary driver of changes in wages. A paper by the EPI has stated three points that SBTC and the job polarization hypothesis fail to answer adequately. One important point is the failure of education wage differentials to adequately explain the growth of wage inequality, which mainly happened among workers with similar training and experience.17 Since rising wage inequality has happened among workers of similar training and experience, education cannot explain it, which directly contradicts SBTC, which claims that education is one of the primary reasons for the growth in wage differences. As well, SBTC fails to adequately explain the massive pay rise among the top 1%.18 This rise is actually the primary change in the distribution of earnings and should be the focus of any explanation of change in the distribution. The rise in pay is significantly greater than that expected by SBTC and the job polarization hypothesis, which expects pay to rise and more jobs to be created in technology-using occupations. However, not as many additional jobs have been created as would have been expected, and the pay increases to the jobs that are already in place at the top of companies have been significantly larger than SBTC predicts. Third, the observed education wage gaps could be due to something other than technological change, such as changes in unionization, globalization, or in industry regulation.19 Since this has been observed to be the case, SBTC, which relies exclusively on education gaps to explain the polarization of job and wage growth as a whole, falls flat. If this change is not due to technological advancement, then SBTC loses any sort of foundation it once had.
So now that some alternate views of the job market and an increase in wage inequality have been proposed, how best to fix them? A committee meeting under President Johnson in the 1960s proposed a guaranteed minimum income and free two-year education for displaced workers, which seem like reasonable places to start.20 This will allow workers who have been displaced to derive relevant training, since they can no longer acquire that at companies, and reenter the workforce under the NEBM, which rewards workers with newer, more relevant skills. Also, a guaranteed minimum income will allow these workers to receive this training without having to dip into their savings while they are out of work, which will make workers more likely to take up the offer of reeducation. These measures may help workers who have been displaced by the NEBM obtain relevant skills and reenter the job market.
BIBLIOGRAPHY
Autor, David H. “Polanyi’s Paradox and the Shape of Employment Growth.” Federal Reserve Bank of St. Louis: Economic Policy Proceedings, Reevaluating Labor Market Dynamics (2015): 129–177.
Lazonick, William and Moss, Philip and Salzman, Hal and Tulum, Öner, Skill Development and Sustainable Prosperity: Cumulative and Collective Careers versus Skill-Biased Technical Change,” Working Paper Series, Institute for New Economic Thinking No. 15 (December 8, 2014).
1 Autor, David H. “Polanyi’s Paradox and the Shape of Employment Growth.” Federal Reserve Bank of St. Louis: Economic Policy Proceedings, Reevaluating Labor Market Dynamics (2015): 129–177.
2 Ibid
3 Ibid
4 Ibid
5 Ibid
6 Ibid
7 Ibid
8 Lazonick et al. “Skill Development and Sustainable Prosperity: Cumulative and Collective Careers versus Skill-Biased Technical Change,” Working Paper Series, Institute for New Economic Thinking No. 15 (December 8, 2014).
9 Ibid
10 Ibid
11 As quoted in Autor, “Polanyi’s Paradox.”
12 Ibid
13 Ibid
14 Lazonick et al., “Skill Development.”
15 Ibid
16 Ibid
17 As quoted in Autor, “Polanyi’s Paradox.”
18 Ibid
19 Ibid
20 Autor, “Polanyi’s Paradox.”