Samuel Bowles and Arjun Jayadev write,

We distinguished between those who directly or indirectly produce goods and services that we consume—who Adam Smith called productive labor—and those who we term guard labor: the police, private security guards, military personnel and others who make up the disciplinary apparatus of a society.

…roughly one in four in the United States economy is now engaged in guard labor—providing security for people and property and imposing work discipline. Since 1890 the guard labor fraction of the United States labor force has increased four-fold. And in Sweden today the guard labor fraction is less than half that of the United States.

…In many countries, the job of getting people to abide by the rules is not left up to the specialists that we have included in guard labor. Anyone who has tried jaywalking in Germany will know what we mean: it’s not the police who you have to worry about, but your (equally formidable) fellow pedestrians

Their definition of “guard labor” includes the unemployed (over and above some frictional level), under the theory that the unemployed population acts as a discipline on those who are employed. That might explain why they used Sweden as a favorable example, rather than, say, France.

They also include supervisory workers in their definition of “guard labor.” In fact, numerically, this is the largest group, which makes their description of guard labor as “police, private security guards, military personnel and others” seem a bit misleading.

Douglass North also documents a decrease in the portion of population involved in production. As I noted in this essay, he estimated that half of U.S. output consists of what he calls “transaction processing.”

The difference is that North sees supervisors and others as adding value. They enable property rights to function, leading to more efficient production. In the Bowles-Jayadev model, supervisors serve the function of expropriating the work of productive labor.

If Bowles and Jayadev are right, then supervision does not increase output. That would imply a profit opportunity for firms that leave workers less supervised, or for workers who become self-employed. It would seem to me that if my boss does nothing but steal my output, then I ought to quit and become my own boss.

An earlier paper by Bowles and Jaradev is not behind a subscription wall.