by Andrew Yang
Economists in particular seem predisposed to suggest that all will be well. People invoke the Industrial Revolution and say, “We have heard these fears before, all the way back to the Luddites. New jobs always appear.” There is an almost magical embracing of ignorance cloaked in humility: “It is unknowable what the new jobs will be. It is beyond human wisdom. It would be arrogant to guess. I just know that they will be there.” Oftentimes, the person who thinks all will be okay is guilty of what I call constructive institutionalism—operating from a default stance that things will work themselves out.
This is, to my mind, a disavowal of judgment and reality. History repeats itself until it doesn’t. No one has an incentive to sound the alarm. To do so could make you seem uneducated and ignorant of history, and perhaps even negative and shrill.
It also would make you right in this case.
There has never been a computer smarter than humans until now. Self-driving cars are a different type of leap forward than the invention of cars themselves. Data is about to supplant human judgment. And on and on. It’s like the warning you get when investing—sometimes the past is not the best indicator of the present or future.
It’s important also to remember that things got quite rough during the Industrial Revolution; in America this is the period between 1870 and 1914 when factories and assembly lines absorbed millions of workers before World War I. There was considerable upheaval, and the role of the state evolved in response to unrest. Labor unions rose up in 1886 and pushed for increased worker rights, 40-hour work weeks, and defined pensions. Labor Day was inaugurated as a national holiday in 1894 in response to a railway strike that killed 30 people and caused $80 million in damages—the equivalent of $2.2 billion today. The United States instituted universal high school; in 1910 only 19 percent of American teenagers were in a high school, and barely 9 percent of 18-year-olds graduated. By 1940, 73 percent of teenagers were in high school and the median American graduated. The women’s suffrage movement culminated in success in 1920. Socialism, communism, and anarchism were all vital political movements. There was a constant whiff of revolution. Even if you rely solely on history, you’d expect a lot of conflict and change ahead as the labor pool shifts due to technological advances.
“Isn’t the labor market simply going to adjust to the new reality and people will move on to other jobs?”
In college, I learned about the efficient capital market hypothesis: stock market prices reflect all available information, and attempts to beat the market are going to be ineffective over time. Now, most every investment professional believes that this is grossly incorrect or at least incomplete given the financial crash, the rise of behavioral economics, the success of certain hedge funds, and the fact that trading firms are investing millions in having a faster pipe to the exchanges to front-run other traders.
The labor market is assumed to be similarly high-functioning, too. That is, if someone gets fired or their job gets automated, they’ll find a new job that’s the right fit. A lot of our public policy is built around this. This, too, is fundamentally incorrect thinking.
For highly qualified and talented people in robust marketplaces, the labor market is pretty seamless. If you’re a great Silicon Valley programmer, you can practically just cross the street and get another high-paying job. You’ll probably also have a couple of nice headhunters helping you in order to collect the finder’s fee you come with, perhaps 12 to 15 percent of your annual salary.
The less qualified and talented you are and the less prosperous your local economy, the dicier things get. If you are a factory worker or salesperson in a store that just closed, chances are the other factories or stores nearby aren’t growing and don’t have job openings. Once you leave the market, it’s especially rough. People who have been unemployed for a while lose confidence and skills. Studies have shown that employers think you’re a major risk if you haven’t been employed for six months. Atrophy can set in quickly. Women who take a break to raise children often have a hard time ramping back up, even if they’re highly educated.
The employment market is loaded with friction. We all know that in real life. Yet so much of our policy assumes a dream world where people are infinitely mobile across state lines, know what jobs are there, have the savings to wait it out, make wise decisions about school, are endlessly resilient, and encounter understanding employers who are rooting for them and can see their merits. I’ve hired hundreds of people over the years. For the normal person, virtually none of this is true.
“Okay, I can buy that old jobs will disappear, but won’t there be new jobs we can’t predict that will take their place?”
Every innovation will bring with it new opportunities, and some will be difficult to predict. Self-driving cars and trucks will bring with them a need for improved infrastructure and thus perhaps some construction jobs. The demise of retail could make drone pilots more of a need over time. The proliferation of data is already making data scientists a hot new job category.
The problem is that the new jobs are almost certain to be in different places than existing ones and will be less numerous than the ones that disappear. They will generally require higher levels of education than the displaced workers have. And it will be very unlikely for a displaced worker to move, identify the need, gain skills, and fill the new role.
Look at retail. Some might say, “It’s okay that the malls and main streets are closing, because you’ll still need warehouse workers and truck drivers to deliver the goods plus web designers for all of the e-commerce storefronts.” Yet all of the new roles are likely to be located far from the mall and other population centers. Over time the warehouse workers will be replaced by a handful of technicians who supervise and operate warehouse robots and the delivery drivers will be replaced by a handful of logistics specialists. We can celebrate the 200 new robot supervisors in suburban California and the 100 new logistics specialists in Memphis and the 50 new web designers in Seattle and say, “Hey, we didn’t know we’d need these 350 college-educated people—hooray!” Meanwhile there will be 50,000 unemployed retail employees who will be looking fruitlessly for opportunities in their shrinking communities.
When newspapers began shifting from paper-based to online publishing, people complained that “we’re trading analogue dollars for digital nickels and dimes.” That’s what’s going to happen with workers. We’re going to trade 100 high school graduates for 5 or 10 college graduates someplace else.
The test is not “Will there be new jobs we haven’t predicted yet that appear?” Of course there will be. The real test is “Will there be millions of new jobs for middle-aged people with low skills and levels of education near the places they currently reside?”
The closest thing to a growth opportunity for the unskilled is being a home health care aide, which isn’t a good fit for most—the former truck drivers will not be excited to bathe grandma. It’s also a terrible job. On average, home care aides work 34 hours a week and make an average of $22,600 a year. One in four live in households below the federal poverty line and many don’t have health care themselves. The field has a high rate of turnover—some estimates put it as high as 60 percent per year. Of the 10 occupations that added the most new jobs in the past several years, personal-care aides earned less than all except for fast food workers.
“Some would call it a dead-end job,” said Deane Beebe, a spokeswoman for the Paraprofessional Healthcare Institute. “These are very hard jobs. They’re very physically taxing—they have one of the highest injury rates of any occupation—and they’re very emotionally taxing: It’s intimate work; it’s isolating work.”
If home health care aide is our answer to the future of jobs, we’re in deep trouble.
“The government should provide education and retraining programs to help transition workers to new jobs.”
In theory, this sounds great, and it makes for wonderful soundbites.
In reality, studies have shown that retraining programs, as
currently practiced, tend to show few, if any benefits. The biggest recent efforts revolved around manufacturing workers over the past 15 years. One study of the Trade Adjustment Assistance (TAA) program, a federal program for displaced manufacturing workers, found that participants in the program garnered less income over a four-year time period than the control group, with older workers showing particularly little benefit. An independent analysis by Mathematica Policy Research compared TAA recipients to workers who got traditional unemployment assistance and found that TAA recipients had lower earnings than people who received regular unemployment assistance, and only 37 percent of those who were trained for specific jobs were actually working in that profession. A similar evaluation of Michigan’s No Worker Left Behind program found that one-third of workers did not find any work after participation in the program, not vastly lower than the 40 percent unemployment rate that laid-off factory workers experienced generally in another study.
One laid-off Chrysler worker, Mal Stephen, commented to an interviewer after completing a $4,200 course at a private training center paid for by the government, “I still haven’t got a job in my skill” a year after finishing the course, and “[Government-funded retraining is] just a way for these little cheap schools to make money, everybody’s scamming the money.” Stephen, 51, received a certificate in computer skills and business math after 16 weeks at public expense. Other workers describe new and for-profit schools of dubious quality offering retraining targeted to laid-off workers with little benefit. The sociologist who interviewed Stephen described him and his fellow laid-off workers as having gone through “the fiction of learning so that they could put it on their résumés and the state could write them off as retrained.”
This is when one is able to access educational benefits; the No Worker Left Behind program had a waitlist of tens of thousands in Michigan in 2010 and stopped taking new applicants shortly thereafter. A study of several dozen laid-off workers in Michigan could only find one who was taking classes that were paid for through a government retraining program. The others had been denied retraining benefits because too much time had passed, the courses they were trying to take were in another state, the subject matter wasn’t supported by the program, or there were pauses between classes and the program required retraining to be continuous. Others said they were unable to determine what benefits were available to them and were told to leave their name on a list, only to never hear back.
Successfully retraining large numbers of displaced workers would require a heroic number of assumptions to prove true. The government needs to be able to identify displaced workers over a range of industries and have both the resources to pay for mass retraining and the flexibility to accommodate individual situations. Each person needs to have the capacity and will to be retrained in an in-demand field. The government needs to be an effective disseminator of information to thousands of individuals in real time. The worker needs to actually learn new marketable skills from the course or school in question. Last, there need to be new employers in the region that want to hire large numbers of newly trained middle-aged workers as opposed to, say, younger workers.
All of the above will hold true for some proportion of the displaced, but not for most of them. The reality is more often displaced workers spending government funds or racking up debt at the University of Phoenix or another for-profit institution in desperate bids to stay relevant and marketable.
We should 100 percent invest in successful retraining of employees. But we should also know that we’re historically very bad at it even in situations where we know displacement is happening. Expecting this to be effective over a large population over a range of industries is more wishful thinking than policy recommendation.
“If jobs are already vanishing, wouldn’t it be showing up in the unemployment rate?”
Not necessarily, because the unemployment rate doesn’t measure what you likely think it measures.
As of September 2017, the unemployment rate is only 4.2 percent, close to the lowest rate since the 2008 economic crisis. That sounds great, and economists are talking about the very optimistic case of “full employment,” which is when an economy has as many jobs for people in the workforce as want them.
The problem is that the unemployment rate is defined as how many people in the labor force are looking for a job but cannot find one. It does not consider people who drop out of the workforce for any reason, including disability or simply giving up trying to find a job. If you get discouraged and stop looking for any reason, you no longer are considered “unemployed.” The unemployment rate also doesn’t take into account people who are underemployed—that is, if a college graduate takes a job as a barista or other role that doesn’t require a degree. Conservative economist Nick Eberstadt says the unemployment rate “no longer serves as a reliable predictor of the numbers or proportions of people who are not working—or for that matter, for those who are working.”
The unemployment rate is like checking how a party is going based on everyone who’s at the party. It doesn’t take into account the people who were never invited to the party or couldn’t get in. It also doesn’t take into account the people who are in the wrong room at the party and having a bad time.
The proportion of Americans who are no longer in the workforce and have stopped looking for work is at a multi-decade high. There are presently a record 95 million working-age Americans, a full 37 percent of adults, who are out of the workforce. In 2000, there were only 70 million. The change can be explained in part by demographics—higher numbers of students and retirees—but there are still 5 million Americans out of the workforce who would like a job right now that aren’t considered in the unemployment rate. Both the historically low labor participation rate and broader measures like the “U-6” rate that include underemployment show high levels of dislocation and a less healthy labor market, particularly for younger workers. The New York Federal Reserve recently measured the underemployment rate of recent college graduates and came up with 44 percent.
The U-6 unemployment rate was 8.4 percent in May 2017, almost twice the headline number. The U-6 rate is a much more revealing measurement and has ranged between 9 percent and 16 percent for the past 10 years.
The unemployment rate is a terribly misleading number that we should stop relying on unless it’s accompanied by a discussion of both the rate of underemployment and the labor force participation rate.
“If we were undergoing a technological revolution, wouldn’t we be seeing it appear in increased productivity?”
You probably weren’t thinking about this one. But it’s a question that economists and academics have been debating. The thought is that we’d see a productivity spike if we were doing a ton more with technology and fewer people. Productivity numbers are actually lower now than they have been in quite some time, leading people to say that any automation-related job displacement concern is misplaced.
There are a few possible explanations. One is that productivity indicators are backward-looking. For example, productivity numbers will show zero indication of self-driving vehicles until tens of thousands are on the road. Yet, we can be pretty sure they’re coming. We’re not ostriches—we can look around and make reasonable projections of the future. Counting on the measurements to tell us what’s going on is like waiting until the storm is here before battening down the hatches.
Another is that it’s possible that low productivity numbers actually reflect an overabundance of labor looking for things to do. Ryan Avent of the Economist poses a theory that technology has created an abundance of labor, both human and machine, and that companies when faced with both low labor costs and a low-growth environment invest less in new technology, which leads to lower productivity growth. This would suggest that we’re in an environment where employers are faced with low incentives to innovate because people are quite cheap to hire.
For example, imagine if over the years I slowly invented a machine that did the work of 10 percent of
American workers. Would unemployment surge by 10 percent over that time? No, because the displaced workers would have to keep working in order to feed themselves, and so would take any jobs in sight, thus depressing wages and keeping productivity low. It would also keep incentives low to automate away further labor and depress the labor participation rate. This is a pretty perfect description of where we are right now.
There is another big reason that productivity statistics do not show a massive increase in output balanced by fewer workers: We’re still technically in an expansion, and employers save the toughest, most unpopular decisions for when times get hard.
When I was the CEO of my education company in the mid-2000s, we had many years of robust growth. Times were flush, which made it a lot easier to be a generous boss. We had free food a lot and regular company outings. I bought the company Knicks and Mets season tickets that we divided among the team. People got raises and bonuses very reliably.
Then, we had a month where revenue was lower than the year before. It was January 2009, so it seemed like it might be a very important sign of things to come. I went into my office and started planning for different scenarios. The contraction path had me looking at personnel and ways to make things more efficient. These included not hiring new people, outsourcing certain non-core services, scaling back on planned raises, and renegotiating with vendors. We had an exceptional team, but there were a couple recent hires that I thought we could potentially let go of if things became really rough. In flush times, I would not have given those recent hires a second thought.
