sean-pollock-203658-unsplash.jpg

US Economic engagement

With present speed of technological progress and advancement there is the real possibility that the job market will experience disruptions and destabilization. While the results may not be on apocalyptic levels, it is important to track changes and see how these changes precipitate into the job market in order to create useful policies for the future. Currently we will examine the effects of "narrow" AI—defined as AI that performs one narrow task—since it has near-term business impact unlike general artificial intelligence. 

Today, it may be challenging to predict exactly which jobs will be most immediately affected by AI-driven automation. Because AI is not a single technology, but rather a collection of technologies that are applied to specific tasks, the effects of AI will be felt unevenly through the economy.
— EXECUTIVE OFFICE OF THE PRESIDENT, 2016

Overview:

Every job, even the most advanced one, consists of tasks that have a potential to be automated. For some jobs, automating most basic tasks enables a worker to spend her time more efficiently, put energy and time into more challenging problems, and as a result contributes to the general productivity. However other workers, whose job consist of only the tasks with high automation potential, will be pushed out since their tasks can be now executed in a faster and cheaper manner through new technology. Moreover, increased marginal returns to capital and hence substituting capital for labor can have significant consequences during the early transitional stages. Historical analysis of previous technological revolutions can be informative in understanding both of these mechanisms.

 
robin-sommer-
 

Skilling and De-skilling Revolutions:

The rich literature on technological revolutions provides contradicting evidence regarding the impact on employment and standard of living. How can this contradiction be explained? First, let us consider what is a technological revolution. According to Caselli (1999) it is the implementation of new types of machines that will have wide-ranging consequences. This difference in consequences is explained by the fact that workers will differ in how easily they can acquire skills necessary to operate the new generation of more productive machines. Hence, the extent to which Schumpeterian “creative destruction” will destruct labor market depends on the type of revolution. If the new skills can be acquired more cheaply than older skills, workers face de-skilling revolution. If there is a higher cost associated the with the pre-existing machines, the revolution will be skill biased.

As Frey and Osborne postulate in their very influential paper, the initial part of the Industrial Revolution focusing on assembly line improvements worked in exactly the former way. To enable low cost production on a massive scale, the whole process of assembling previous done by one or two people was deconstructed into smaller parts executed by many more workers. The gains from using the most advanced technology increased significantly, and the number of users using the most advanced technology rose complementing the unskilled labor. As a result, despite middle-skill artisans losing their jobs, researchers provide data that real wages over the period 1760 to 1860 rose faster than GDP per capita (Clark, 2008), putting Luddite riots in a new light. In words of Frey and Osborne: “While this implies that capital owners were the greatest beneficiaries of the Industrial Revolution, there is at the same time consensus that average living standards largely improved.”

With the progress of technological advances, skilled workers started to benefit more. Workers with low acquisition costs employ new technologies faster, so it is more profitable to increase their capital endowment. Lower skilled workers generally have lower capital-to-labor ratio, therefore “experience an absolute decline in wages” (Caselli, 1999). In other words, with more mechanized assembly lines, a greater amount of tasks could be automated, and the new automats required workers to acquire different skills in order to operate them. Moreover, these revolutions enabled the emergence of new white-collar jobs (Goldin, Katz, 1998). In order to best position themselves for the market demands, future workers sought more time for education, leading to sharp increases in tertiary educational attainment.

What seems to be a common trait between both skilling and deskilling revolutions, is hollowing out of middle income jobs. According to Autor and Dorn (2013), changes in the American employment can be characterized by the reverse bell-curve, and these results were observed in other places as well (Goos, et al., 2009). In the markets exposed to computerization of routine tasks, high skilled workers joined the labor stock while low skilled workers moved to low-skill service jobs.

 
marc-olivier-jodoin-291607-unsplash.jpg
 

Impact of advancement in AI and automation in the 21st Century:

According to the data presented in the World Bank’s estimates, “the stock of robots in the U.S. and Western Europe increased fourfold between 1993 and 2007,” and this rate can be expected to only grow. Moreover, “one more robot per thousand workers is estimated to reduce the employment to population ratio by about 0.18-0.34 percentage points, and wages by 0.25-0.5 percent”. Some companies are at the forefront of the digital frontier and will benefit from the increased use of industrial robots, and it is important to note what distinguishes these companies. Usually these companies are in the tech-sphere since outside Silicon Valley adoption, especially early adoption of AI technology has received lukewarm reception. Adoption is closely correlated with how digitized the sector is, since these industries have historically been leaders in using new technology in their companies. According to a McKinsey survey "firms that combine a strong digital capability with robust AI adoption and a proactive AI strategy see outsize financial performance," suggesting that commitment to AI can provide growth for companies. However, it is important to note that this growth may not be equally distributed. 

Even more challenging task than estimating which jobs will be gone, is stating which new jobs will be created. So far each technological revolution created more jobs than it destroyed and number of people employed has been steadily increasing, leading to, among others, significant alleviation of extreme poverty from 43% in 1990 to 21% in 2011 worldwide. In the American Context, apart from the small downfalls during economic recessions, employment increased since the end of the World War Two on average by 3.9% every year.

In the end, they will depend on the complex interaction of how each of these channels affect the economic forces that drive the organization and location of production and work, i.e. the relative price of labor and capital, the cost of transacting, the economies of scale and market competition. Without question the US government will need to take initiative in tackling the challenges that come with AI adoption. This will involve careful observation and research into the effects of growth of AI as well as policies that will work to support both technological progress and the economic prosperity of US citizens. 


In order to estimate the changes we drew from methodology developed by Frey and Osborne (2013), and applied their 702 occupations susceptibility to computerization dataset to 2016 Bureau of Labor Statistics regional data. As a result we can compare the degree of states’ proneness to automation of employment. The map below is a graphic representation of a more nuanced level of analysis. The data shows state-wide probabilities, accounting for size of employment, together with specific breakdown of jobs grouped in one of three categories suggested by Osborne and Frey, 2013 (Low, Medium, and High Probability of Computerization). Generally, a higher state-wide probability corresponds to the relatively bigger number of High Probability jobs, and Nevada is a state with both the highest state-wide probability (62.78%) and the highest ratio of High Probability to Low Probability jobs (2.33).  

Even more challenging task than estimating which jobs will be gone, is stating which new jobs will be created. So far each technological revolution created more jobs than it destroyed and number of people employed has been steadily increasing, leading to, among others, significant alleviation of extreme poverty from 43% in 1990 to 21% in 2011 worldwide. In the American Context, apart from the small downfalls during economic recessions, employment increased since the end of the World War Two on average by 3.9% every year.

In the end, they will depend on the complex interaction of how each of these channels affect the economic forces that drive the organization and location of production and work, i.e. the relative price of labor and capital, the cost of transacting, the economies of scale and market competition. Without question the US government will need to take initiative in tackling the challenges that come with AI adoption. This will involve careful observation and research into the effects of growth of AI as well as policies that will work to support both technological progress and the economic prosperity of US citizens.