John Oliver’s Segment on Jobs and Automation Doesn’t Quite Get It Right

Riley Ohlson

Riley Ohlson Vice President of Federal Gov't Affairs, AAM

John Oliver tackles some wonky issues on his HBO show, Last Week Tonight. Sunday’s episode was no exception, as he examined a topic very dear to our hearts: Automation.

And Oliver’s bit certainly garnered a lot of attention, with more than 4.3 million views on the show’s YouTube channel alone. But unfortunately, Oliver didn’t quite get it right — and given the host’s incredible reach, it’s important we clarify a few things (even if my seven Twitter followers don’t have quite the same reach of Oliver).

Oliver began by poking fun at President Trump's bluster about stolen jobs during the 2016 presidential campaign. After a begrudging admission that “some” manufacturing job loss is due to trade, Oliver quickly pivoted to automation, the phenomenon whereby industrial robots and other equipment take over parts of the production process that used to be done by humans.

Trump certainly has been known to put forth misleading arguments, and Oliver is right to point out that we do produce far more stuff these days with fewer workers. But Oliver also misses the mark a bit. 

Why? Because it glosses over a historic collapse in manufacturing employment that is, in fact, not explained by automation. If you look more closely at the graphic that appears in the background of his segment, you’ll notice employment remains basically flat between 1980 and 2000, even while output trends upward. It’s not until 2000 that employment really begins to trend downward.

So what happened? Naïve and short-sighted trade policy happened.

This period is different — and it’s not because of robots

First, look at employment. The graph above makes the employment fall look modest, but it actually was stunningly steep. Below is a graph of manufacturing employment generated from the Bureau of Labor Statistics website.

You’ll notice that since the late 1960s, manufacturing employment remained relatively steady for decades, with occasional fluctuation around a recession. Factories closed, jobs were lost, automation and other productivity improvements required fewer workers to make the same products, but manufacturing employment stayed strong, and U.S. exports and imports remained relatively balanced.

A widening of the trade gap in the 1980s, meanwhile, was met with action to address currency cheating, leading to the Plaza Accord in 1985. But for the most part, the macro picture was relatively stable in the sector.

All of this hews to the standard productivity narrative: as productivity increases, workers’ earnings rise, so they can purchase more. Factories can produce more for less, lowering the price of their products relative to other goods, boosting demand. That leads to more production, requiring more workers.

This cycle is expected, and generally appeared to work in the latter decades of the 20th century.

But in the 2000s, employment in U.S. manufacturing nosedived, shedding over 5 million jobs in a few short years. Job loss sustained in manufacturing during this period was more severe than during the Great Depression, devastating communities across the country. Many places still have not recovered.

Along with Oliver, lots of folks have blamed automation for this massive job loss. But that argument simply doesn’t hold up.

First, manufacturing productivity growth was 25.8 percent faster compared to the overall economy between 1990 and 2000. From 2000 to 2012, it was only 22.7 percent faster — but job losses in the 2000s were more than ten times greater than in the 1990s, even though productivity gains had slowed.

Second, a closer look reveals that much of the manufacturing productivity gains in the 2000s were overblown, because of the way the government measures productivity and output in the computer and electronics industry. Computers, semiconductors and other electronics are way more powerful today than in the past, and because these improvements show up in government figures, it makes it seem as if more of these products have been manufactured, instead of simply becoming more advanced.

It’s pretty wonky stuff, but here’s an example. Let’s say I’m using a flip phone from the early 2000s. I decide to finally get with the times and buy a smartphone (to better connect with my seven twitter followers), which is obviously much more powerful than my old phone.

To keep math to a minimum, let’s say the smartphone has an American-made semiconductor five times more powerful than the one in my old phone, and so I pay five times more for all the benefits of this newfangled technology. I didn’t buy five new phones, I bought one, more powerful device. But the way it’s measured, it shows up like five devices rolled off the assembly line to meet my new demand.

Looking at official statistics, it appears there “was [an] astonishing 179 percent increase in computer manufacturing output from 2000 to 2010” when, in fact, we made fewer computers during this period.

This all means that there were even less productivity gains in the United States in the 2000s than the big picture data suggests. Yet, job loss soared.

Meanwhile, if we look outside the United States, we find that manufacturing powerhouses like Germany (19 percent of population employed in manufacturing), South Korea (16.9 percent) and Japan (16.9 percent) all top the list of industrial robots per capita, but still maintain a thriving factory workforce.

The United States has less robots and a smaller portion of its population, 10.5 percent, working in the sector. If anything, it would appear the increase in productivity that accompanies the higher use of industrial robots allows for firms in the countries to be more competitive.

So unless we want to compete globally by racing to the bottom on wages, environmental protections and health and safety rules, increasing the use of industrial robots is actually key to growing jobs. Siemens President and CEO Joe Kaesar put it this way:

“First, it is true that digital manufacturing does cut out the middle-man. More and more routine, repetitive assembly tasks will be taken over by machines. But as certain jobs disappear, new ones open up in other parts of the factory. Germany in many ways exemplifies this trend. Today, German manufacturers deploy three times more robots than U.S. companies, but they also still employ more humans. Relative to the size of our economies, German’s manufacturing workforce is twice the size of America’s.”

If automation wasn’t to blame, then what was?

China.

While there are a number of ways the United States can improve its trade policy, the biggest miscalculation has been on China. In 2000, the United States and other countries opened trade with China after it was granted Permanent Normalized Trade Relations status. The following year, China joined the World Trade Organization. That opened the floodgates, as it seemed entire industries shifted production overseas, from textiles and apparel to furniture to toys to consumer electronics.

This was incredibly unfair to American workers (along with companies that wanted to maintain production in the United States). China’s economy is state-run, allowing the government to pump money into areas in which it wants to dominate globally and manipulate its currency to tilt the playing field in its favor. China’s labor and environmental standards are lax compared to the United States and other Western nations. Meanwhile, China also routinely steals intellectual property and trade secrets from U.S. companies.

The mistaken assumption supporters of open trade with China made was that exposure to global institutions and deeper trade relationships would have a liberalizing effect, making China's economy, and society, more like those in the West. But instead China doubled down on its unfair trading practices.

There’s little doubt that China played a decisive role in U.S. manufacturing job loss in the first decade of the 21st century. Economists from Yale University and the Federal Reserve found a link between manufacturing employment and the change in China’s trade status;  Massachusetts Institute of Technology researchers estimated that growing Chinese import competition led to the loss of between 2 and 2.4 million jobs. Meanwhile, the Economic Policy Institute reported 3.4 million jobs were lost or displaced by Chinese imports from 2001 to 2017.

Now, I want to be clear that automation and productivity improvements have caused disruption and changed the workforce. We need to be ready for new developments in Artificial Intelligence, automation and Skynet, and we need to get it right. Policymakers, unions, and business leaders should be proactive in addressing rapidly changing technology, including by training workers and students for the jobs of the future.

But automation isn’t the boogeyman it is often made out to be. Had China not engaged so aggressively in unfair trade, the job loss from productivity increases from automation in the 2000s likely could have been absorbed as it had in earlier decades, and factory jobs could have remained far more steady. After all, one recent working paper found that only one of 270 occupations listed in the 1950 census has been fully automated out of existence: elevator operators.

In fairness, Oliver did note that the hysteria surrounding automation has been overblown. But it’s time to dismiss the idea that trade doesn’t play a role in factory job loss when it so clearly has.

These weren’t just “some jobs.” These were millions of America’s most productive workers across the country losing their livelihoods and seeing their communities devastated because of dysfunctional trade. We continue to see the terrible impacts of this job loss today, including in some surprising ways.

It’s unfair, and irresponsible, to shift the blame for this historic policy blunder to the robots.

***

Reposted from AAM