�鶹������ý

Today, work as we know it is shifting, and rapidly. Over the next decade and beyond, how will artificial intelligence and automation change work and opportunity in cities like Phoenix?

The Phoenix metro area and other communities across the country will be at the front lines of this change.¹ To prepare for this future, �鶹������ý will host the first ShiftLabs on April 20th in Phoenix, Arizona in partnership with Arizona State University and with support from the Rockefeller Foundation. At the day-long design lab, leaders from the Phoenix region and across the country—from technology, industry, policy, philanthropy and culture—will come together to consider the impact of technology and automation on work in Phoenix and to develop a long-term, place-based vision for opportunity.

To bring a data-driven lens to ShiftLabs, �鶹������ý partnered with leading labor market analytics company and their analysis of to conduct a first-of-its-kind analysis of the potential of automation to impact jobs in the greater Phoenix region. We ask: Of the thousands of jobs held by Phoenix workers today, which could be performed by existing technology? Which occupations and skills are at greatest risk of automation, and who holds those jobs today? To answer these questions, we combined and analyzed Burning Glass data on the likelihood of a computer being able to do a job using existing technology, as well as data from the Bureau of Labor Statistics on occupations in Phoenix and nationally.

To be sure, emerging technologies will also create many jobs, including entirely new jobs that don’t even exist today. This is a familiar pattern—about came from the creation and expansion of brand new jobs. On balance, automation and technology may create more jobs than they eliminate, but predictions of the number and types of jobs that will be created are outside the scope of this study.

The findings of our analysis are clear: Automation will have a widespread impact on jobs in Phoenix in the years ahead, and especially on low-skilled jobs. In some cases, technology will eliminate high-risk jobs. In many more cases, technology will change them—sometimes dramatically.

In the Phoenix metro area, 649,040 people are employed in occupations that are at high risk of automation—35 percent of total jobs. Another 537,110 jobs (29 percent of total jobs) are at moderate risk of automation. Only a little more than a third (36 percent) are at low risk.

Workers with less education

Among workers, the least educated workers are at greatest risk of automation. This is especially true of workers with a high school degree or less, who comprise 45 percent of the workers at high risk of automation in Phoenix and just 18 percent of workers at low risk. Those numbers are flipped for low-risk jobs: Half of workers in low-risk occupations have a BA or higher, while just 18 percent have a high school degree or less.

This vulnerability is further magnified by Phoenix’s lower levels of post-secondary education attainment, compared to the country overall. Across the United States, between 2012 and 2016, had a high school degree or higher and just over 30 percent held a BA or higher. In contrast, in Phoenix, had a high school degree or higher and just over 27 percent held a BA or higher.

Workers who earn the least money

The jobs at high risk of automation are nearly half as well paid, on average, as the jobs at low risk of automation. The average annual salary of workers in the more than 200 jobs that are at high risk of automation is $32,959. Meanwhile, the average salary of the workers in the more than 300 jobs that are at low risk of automation is $67,416—more than double that of the high-risk workers.

Women

The high risk occupations in Phoenix disproportionately employ women. Applying national averages of women employed across occupations,¹ women constitute 58 percent of workers in high risk occupations in Phoenix. Women dominate in many food and retail-related industries that are especially high risk. For instance:

• Cashiers: Over 44,000 people in Phoenix worked last year as a cashier – a job with a 97 percent risk of automation. Nationally, 73 percent of cashier jobs were held by women.
• Office clerks: More than 36,000 people in Phoenix worked as office clerks – an occupation with a 96 percent risk of automation. Nationally, 83 percent of those positions were held by women.
• Secretaries and administrative assistants: More than 29,000 people worked as secretaries and administrative assistants in Phoenix, which carry a 96 percent risk of automation. Nationally, 95 percent of those positions were held by women.

The Largest Occupations Most at Risk

Of the 50 occupations that employ the most people in the Phoenix metro area (totalling just over half of all workers), the following 15 occupations are the most at risk of automation.

Largest Occupations Least at Risk

Of the 50 occupations that employ the most people in the Phoenix metro area (or about half of all workers), the following 17 occupations have a low risk of automation.

Top 50 Occupations by Number of People Employed

Overall, the risk of automation facing workers in the Phoenix region is just slightly above the risk to all workers nationally. Workers in Phoenix and nationally have the same rate of high risk, but Phoenix is one percentage point higher than the United States on medium risk, and one percentage point less in low risk.

Looking more closely at specific occupational groups within the economy, there are some pockets of greater vulnerability and greater resilience.

The chart below illustrates the occupational groups in Phoenix that differ significantly from the national average.

Ways the Phoenix area workforce is more resilient to automation than the nation overall

• Management, business and finance have a low risk of automation. As a percent of employment, Phoenix’s population has 20 percent more workers in management positions than the national average. As a percent of employment, Phoenix’s population has 9 percent more workers in business and financial operations, which carry a low risk of automation.
• Phoenix also has 21 percent more of its workforce in computers and mathematical positions.
• Compared to the national average, Phoenix has 51 percent fewer workers in manufacturing and 13 percent fewer workers employed in transportation and material moving. Both of these occupations are high risk of automation.

Ways the Phoenix area workforce is more vulnerable to automation than the national overall

Phoenix has a greater share of its workforce than the national average in several big occupational groups that have a medium or high risk of automation. These include:

• Phoenix employs 10 percent more workers in office and administrative support than national average. These occupations have a medium-high risk of automation, and include many high risk jobs.
• Phoenix has 6 percent more workers in sales and related occupations than national average, and many of these jobs (like cashiers) are high risk of automation..

Phoenix area employees are less well-represented in certain low-risk occupational groups than the national average:

• Education, training and library occupations are at very low risk of automation. Phoenix has 32 percent fewer workers in these occupations than the national average. Nationally, 73 percent of those positions are held by women. Thus, compared to the rest of the country, Phoenix has significantly fewer low-risk jobs in education that overwhelmingly employ women.

In our analysis, the rankings of automation risk describe the technical feasibility that an occupation can be computerized or automated with start-of-the-art technology available today. This data comes from Burning Glass Technologies, and is derived largely from a well known 2013 study from two researchers at Oxford, Carl Benedikt Frey and Michael A. Osborn. To calculate the automation risk, the Oxford researchers evaluated the ability of computers to perform the underlying tasks associated with the given occupation.

1. “High risk” occupations are the top quartile of risk, with at least 85 percent risk of automation for a given occupation.
2. “Medium risk” occupations are in the second quartile of risk, between 50 percent and 85 percent risk of automation for a given occupation.
3. “Low risk” occupations are in the bottom two quartiles, with less than 50 percent risk.

A few key caveats are important to consider when interpreting the data.

First, the rankings are not a probability that a given job will actually be automated. Because a job or task can technically be done by a computer does not mean that it will. A range of legal, logistical, business, financial, political, and social factors could lower the real rate at which businesses and employers adopt technology and automate functions. Moreover, predictions about technology have a relatively high degree of uncertainty.

Second, jobs that have some tasks that can technically be automated will not necessarily be displaced. Instead, the nature of many jobs will change—in some cases, dramatically—but will not be eliminated. (McKinsey estimates that percent of jobs will be outright eliminated, but that half of job tasks could be automated.) The implication of this change is the need for workers in at risk occupations to continuously upskill to keep pace with the changing requirements of their occupation.

Finally, while technology and automation will displace some jobs and change others, new jobs will also be created and other jobs will expand. Our analysis does not capture the impact of projected job creation.

• The data on automation potential comes from Burning Glass Technologies, which is derived largely from a well known 2013 study from two researchers at Oxford, Carl Benedikt Frey and Michael A. Osborn, titled “”
• Occupational and wage data for the Phoenix metropolitan area is from the Bureau of Labor Statistics and covers the period from January 1, 2017 to December 31, 2017. The geographic area spans Phoenix, Mesa, and Scottsdale, including Pinal and Maricopa counties.
• Data on national averages of women in occupations comes from the Bureau of Labor Statistics.
• Data on education levels of employed individuals comes from the American Community Survey (ACS) five-year estimates (2011 – 2015).