Without Robots, How Can We Afford to Grow Old?

By Alberto Moel, Vice President Strategy and Partnerships, Veo Robotics

Welcome back, fearless reader, to another installment of our series on the macroeconomics of automation. It’s been a while since I’ve written about the topic—we’ve been too busy launching our new product and laying out the theoretical and practical groundwork for safe human-robot collaboration to pontificate about broader topics. But after our successful public debut at FABTECH, we have a bit more time for free-associating.  

As we have written about extensively, there is a narrative out there that robots are taking our jobs, with the assumption that there is a causal relationship between the robots showing up and people losing their jobs. But what if the causality is in the other direction, where robots show up because there aren’t enough people to fill those jobs?  

In today’s post we argue that that is what is happening and, more pointedly, we not only provide evidence that robots are filling jobs left vacant by humans, but also that robots filling these jobs is a necessary condition for continued economic growth. These linked arguments are not new, but only now are they coming to the fore as a sensible counter to the dystopian narratives out there.  

Fewer workers mean more automation and robots

Let’s start with the evidence that aging and demographic headwinds result in more automation. Influential MIT economist Daron Acemoğlu and his former graduate student Pascual Restrepo (now a professor at Boston University) make the case pretty convincingly in their working paper Demographics and Automation:

> We argue theoretically and document empirically that aging leads to greater (industrial) automation, and in particular, to more intensive use and development of robots. Using US data, we document that robots substitute for middle-aged workers (those between the ages of 36 and 55). We then show that demographic change—corresponding to an increasing ratio of older to middle-aged workers—is associated with greater adoption of robots and other automation technologies across countries and with more robotics-related activities across US commuting zones. We also provide evidence of more rapid development of automation technologies in countries undergoing greater demographic change. Our directed technological change model further predicts that the induced adoption of automation technology should be more pronounced in industries that rely more on middle-aged workers and those that present greater opportunities for automation.

These findings are backed up by recent research on the use of industrial robots in 382 US metropolitan areas from 2010 to 2015 produced by Professor Mehmet Tosun of the University of Nevada:

_> Our research results so far show that the population aging indicators we use (share of population 65 and older, median age and old-age dependency ratio ¹

) are positively correlated with the growth in the number of industrial robots. In addition, we haven’t found any negative association between these indicators and the growth in real (inflation-adjusted) per capita income in metropolitan areas._

Although it is possible that an older workforce is more amenable to automation, all the evidence points to the fact that increases in automation are correlated with reductions in available labor as workers age out and retire from full-time employment. It’s simply an issue of a reduced labor supply. Baby boomers are retiring and the cohorts that follow them are smaller, so the share of people who are working in the US is shrinking. It is estimated that by 2030, only 59% of adults over 16 will be in the workforce, compared with 62% in 2015.

Automation filling in for the “giant sucking sound” ² of people leaving the workforce is not a new phenomenon. From 1930 to 1980 the US farm labor force declined by about 70%, the machinery to laborer ratio went up by 10x, and farm productivity increased by over 3x. The farm laborers weren’t driven away by technological change, but by the simple fact that labor was more profitable outside of the farm. A 1986 study³ of cotton harvesting estimates that 79% of the decline of picking cotton by hand was driven by the rise of non-farm wages in the same and surrounding communities. Economists Richard Hornbeck and Suresh Naidu examine the natural experiment of the 1927 Great Mississippi Flood to find a causal relationship between workers leaving the farm and increases in farm mechanization. The exodus of laborers after the flood prompted farmers to invest in automation, not the other way around.

Figure 1.

Back to the robots. Although I am not an academic economist, ⁴ I’ve run some simple comparisons that point to the conclusion that fewer jobs cause more robots, rather than the opposite. To start, consider Figure 1, which shows 2010–2020E population compound annual growth rates for the 50 states in the US. ⁵ Red means populations are shrinking on average, while blue means they are rising. Trends are pretty clear—Florida and Texas are growing strongly, while the Eastern states and the Rust Belt are shrinking at the expense of the Western states.

Figure 2.

And where are the robots? Figure 2 shows robot densities (defined as number of robots per 10,000 workers) in the 50 US states ⁶ The average density for the US is 12.4 robots per 10,000 workers. States with below-average densities are shown in blue, and those above the average are in red. You’ll see that the robots are pretty much all in the Rust Belt.

Figure 3.

Eyeballing Figure 1 and Figure 2, you’d probably conclude that more robots and population declines are linked. And you’d be correct. Figure 3 shows a scatter plot of robot density per US state against 2010–2020E population CAGR. The bubbles indicate relative share of the US workforce. On the left we see a cluster of large, growing, low-automation states (e.g., California and Texas) and a bunch of small, shrinking, low-automation states (e.g., Maine). But as we move along the higher automation scale, we see a group of relatively large and important states where population growth is inversely correlated with robot density (e.g., Michigan and Ohio).

Figure 4.

As part of our commercial development, we have spent a lot of time in factories and manufacturing sites around the world, and one common refrain is that it’s hard to find enough qualified workers to run the machines and meet production demands. ⁷ So the drive to automate to make up for the labor gap is strong and continuing. And given forecasted population growth trends (Figure 4) we’d expect this drive to automate to continue.

So why are more robots a good thing?

Economies grow by increasing the production of economic goods and services, from one time period to the next, usually measured in terms of GNP or GDP. Aggregate gains in production accompanied by an increase in productivity leads to an increase in incomes, which ideally means a higher standard of living for the population in that economy. All good things, we can agree.

Growth is usually modeled as a function of physical capital, human capital, labor force, and technology inputs. The usual growth models rely on the classical economist's "growth accounting" toolbox, first developed by Robert Solow in the 1950s. Economic growth can come only from:

• Increases in the quantity of labor, as more young people enter the workforce than older people exit it;
• Increases in the quality of labor, usually proxied by educational levels;
• Increases in the quantity of capital relative to labor (from say, increased automation);
• Increases in the quality of capital, for example, from better or more sophisticated equipment or software tools; or
• An increase in the residual not explained by the above four factors, called "total factor productivity," or TFP. The TFP is also known as "the measure of our ignorance" and is a metric that captures the interaction between the other components of labor productivity.

Clearly, if the population is stagnant or shrinking, this would have an impact on economic growth. ⁸ Stagnant or negative population growth has many downsides, from underfunded pensions to reduced rates of firm formation. ⁹ And if we want to counter this buzzkill, something must make up the slack in weak or negative growth in the quantity of labor. This is where robots come in as a positive force. Because of their flexibility and ease of use, robots not only substitute for human labor, but also provide incremental productivity improvements above and beyond their capital substitution effects. ¹⁰

It is early days, but the initial evidence does indeed point to the fact that faster adoption of automation is a net positive for growth. For example, Acemoğlu and Restrepo find that countries experiencing more rapid aging have grown more in recent decades, and they attribute that to the effect of increased automation. ¹¹ In summary, we believe that automation, for all its disruptions and controversies, is a net economic good. ¹²

---

¹ The ratio between the number of people aged 65 and over and the number of people between the ages of 15 and 64.

² In keeping with the theme of job displacement.

³ The Cotton Harvester in Retrospect: Labor Displacement or Replacement? By Willis Peterson and Yoav Kislev

⁴Although I did play one on TV for a while.

⁵ Population data from Cooper Center.

⁶ Regional robot density data from Brookings.

⁷ This anecdotal evidence is amply confirmed by multiple surveys by the Robotics Industry Association and other industry bodies.

⁸ And there is plenty of evidence of this. See, for example, N. Maestas, K. Mullen, and D. Powell, The Effect of Population Aging on Economic Growth, the Labor Force and Productivity. They find that, all else equal, a 10% increase in the fraction of the population ages 60+ decreases the growth rate of GDP per capita by 5.5%.

⁹ Some of them alarmingly but convincingly outlined by economics writer Eduardo Porter of the New York Times. By the way, I also borrowed the title of this blog post from the last line in his op-ed, so thank you Eduardo Porter for that.

¹⁰ Not to toot our own horn or anything, but at Veo we are building technology that allows for more flexible, productive, and valuable interaction between labor (factory workers) and capital (robots).

¹¹ Daron Acemoğlu and Pascual Restrepo, Secular Stagnation? The Effect of Aging on Economic Growth in the Age of Automation.

¹² Although the careful reader will note that we skipped talking about some of the downsides of automation that merit consideration, such as its impact on inequality or long-term unemployment.