Recent Trends in Labor Intensity. Or, the History (and Future?) of Steady Work in the US

by Andrew McAfee on January 17, 2012

When I’m trying to understand something, I start drawing graphs using whatever data’s available; pictures help me more than tables of numbers or regression coefficients. So here’s a picture I drew to see recent trends in US labor productivity — how much more output the American economy gets from its workers over time.

Instead of looking at productivity as output per hour worked (the usual measure) I instead looked at output per worker. And I inverted the usual measure, so I’m looking at workers per output — how many workers it takes each year to generate a given level of economic output (in this case, GDP). Let’s call this Labor Intensity.

Of course, labor intensity goes down as productivity goes up since one is the inverse of the other.

Because the US Bureau of Labor Statistics counts and categorizes workers every year, we can see how the labor intensity of various professions has changed over time. Here’s a graph of data from 2000-2011, with all American workers placed into nine groups, that I came up with after rearranging the BLS’s categories a bit (click on it for a bigger version):

It shows that the only category with substantially increased labor intensity over that period is health care jobs. This is because our population keeps getting older (and so requiring more caregivers as a percentage of all workers), cost containment is difficult, labor-saving automation is scarce, and offshoring is often not possible. A great deal of health care has to be delivered by hand; it’s highly labor intensive.

I find it very hard to believe that payors will continue to tolerate steadily increasing labor intensity; this graph helps me understand why there’s so much pressure to cut health care costs. I don’t know exactly how this will be done, but that purple line can’t keep drifting upward forever.

The three greenish lines below it belong to job types that have seen only minor (less than 5%) change in labor intensity between 2000 and 2011. ‘Services’ here are all highly manual jobs: firefighters, fish and game wardens, cooks and dishwashers, janitors, barbers, and so on. So it’s not surprising that our economy needs as many of them (per unit of GDP) in 2011 as we did in 2000. Here again, these jobs can’t be offshored, and robots haven’t yet replaced many of these workers at all.

And even though professions like lawyer, scientist, architect, and computer programmer can be offshored, they apparently haven’t been to a huge extent yet. The US economy is 95% as professional-intensive as it was in 2000. Which is all the more reason to stay in school and get that engineering degree.

Or even better if you’re interested in job security, get an accounting or human resources degree. We need even more ‘business and financial’ professionals that we did 12 years ago. This surprises me a bit. I would have thought that automation and offshoring would have made a dent in this job category by now, but evidently not.

The two blue lines show that the managerial and creative classes have seen roughly equal decreases in labor intensity; we need about 90% as many of them (again, per unit of GDP) as we did in 2000. My guess is that this is because computers, databases, and networks have taken over the traditional management function of gathering, summarizing, and transmitting information up the org chart. The district manager, in short, doesn’t prepare the monthly report for the regional manager any more. The ERP and CRM systems do that, and as a result we need fewer district managers, each of whom can cover more territory. As I wrote earlier, management as a profession definitely isn’t vanishing, but it is getting a bit more productive. It’s also true that as the overall number of workers shrinks (see the ‘office and sales’ line, discussed below), the number of people needed to manage them also goes down.

I’m not as sure about what’s behind the roughly 10% labor intensity drop for arts, design, and media folk. Is it the death of mainstream media? The greater productivity brought by the technologies they use? What do you think?

The three reddish lines at the bottom belong to job categories that have seen the biggest decreases in labor intensity. The almost 20% decline in ‘natural resources, construction, and maintenance’ is almost surely due to the collapse of the real estate industry. This category was holding steady until 2006, at which point it fell off a cliff as the housing market did.

The huge (>30%) drop in ‘production and transportation’ is due to a combination of at least three powerful factors: automation, offshoring, and the fact that US manufacturing is shrinking as a share of GDP, even though it’s still growing in absolute terms.

For me, the steep drop in ‘sales and office’ labor intensity is the clearest example constant technical progress. Digitization has eliminated many jobs in this category, and made the rest of the workers much more productive. Offshoring has certainly had an impact here as well, but I think the story behind this steadily downward-sloping line is primarily a technology story. Workers here are still a huge chunk of the American labor force – 23.6% in 2011 – but they are being rapidly encroached on by digital labor. I don’t think this trend is anywhere near played out yet –we’ll continue to see large, steady decreases in labor intensity here as the digitization of the economy proceeds.

I’ll talk more about these and other trends in later posts. For now I just wanted to show and explain the graph, and get your reactions to it. Which of these trends, if any, surprise you? Do you expect to see any of these lines change direction much in the coming years? Leave a comment, please, and let us know.

  • Katepugh

    Andrew – These are fascinating data. I’m struck by the oscillatory nature of “Business and Financial” (~5%) and “Management” (11%).  Not that 16% dominates the GDP, but that population has a major impact on political structures. And, we all know that people smack in the middle of an oscillation tend to panic, and then overshoot in their “repair” of the disturbance. What might this mean, at the aggregate, in terms of political influence? Pursuit of degrees? Respect for our profession?

  • http://twitter.com/boysenandrew Andrew Boysen

    Your graph is very pleasing to look at – what do you use to create it? As a manager, I agree with the idea that firms are trying to have each manager cover a larger territory, and at times even more diverse functions. Lately, for teams that are relatively self-sufficient, if we lose a manager the team is rolled under an existing manager with no new hiring.

    The drop in labor intensity for arts/design could relate to the ease of outsourcing specific design jobs, including to other countries. More firms are relying on digital design (or at least cutting into their print budgets), and digital is easier to hire freelance. There seems to be more international competition for those jobs in the middle of the market, and at the low end templates and easy setup allows people to make decent looking websites easily. I don’t consider myself a skilled designer, but I can throw together a decent looking WordPress website for a non-profit, using existing templates, in less than an hour. It would be interesting to compare the changes in labor intensity to import numbers for design services over the same period.

  • http://andrewmcafee.org/blog Andrew McAfee

    Andrew, I just used Excel to make the graph, but I tweaked the default settings a lot and spent a fair bit of time on it. I really hate ugly graphs…

  • Josef Dietl

    It’s a very interesting visualization, and out of the ten or so things that come to my mind, let me focus on one: “sales and office” vs. “business and financial” (and engineers, though they seem to be stuck between two categories:
    I believe that the tasks that can be easily standardized will be more and more automated or offshored at least in the near future. But every standardization creates a certain residual volume of exceptional tasks that can not be automated or offshored as easily – on the contrary, some of the residue moves from the “office” category to the “engineer, business and financial” category because they turn into some kind of project work.
    So I agree with your prediction of a continued fall in “sales and office”, at the same time I see that fall strengthening “engineers, business and financial”, though it is hard to quantify the effect. Any ideas?

Previous post:

Next post: