Bryan Caplan  

Where Worker Productivity Really Comes From

Back Alley Regulation... What about Capital?...
During my day with Eric Hanushek, he repeatedly asked me, "If rising worker productivity doesn't come from education, where does it come from?"  Good question.  In the pre-modern world, workers got little education and had low productivity.  In the modern world, workers get much education and have high productivity.  Productivity (and education!) keep going up.  If formal academic training isn't the main reason, what is?

The safe answer is, "I don't know.  But whatever the answer is, it's not education."  However, safe answers are not my style.  While I'm not sure, I think I've got a pretty good guess.  Or to be more precise, two pretty good guesses.  To what does the modern worker owe his high productivity?

1. Practice.  Modern workers get really good at modern tasks by practicing modern tasks, under the the tutelage of other workers who have already mastered them.  Modern programmers get good at programming by practicing programming, under the tutelage of skilled programmers.  Modern offshore oil drillers get good at offshore oil drilling by practicing offshore oil drilling, under the tutelage of skilled drillers.  In the pre-modern world, these practice opportunities were simply unavailable.  Now they're everywhere.  Whenever a new industry arises, people get new practice - and new tutelage.  The same holds whenever an existing industry innovates: people get new practice - and new tutelage.

2. Management.  Production is a team sport, and the coaches are called managers.  Even if all the team members are great at their jobs, productivity will be low unless these managers expertly lead and direct their skills.  This is crucial because good management does not come naturally to human beings.  Forging an effective production team isn't quite as hard as herding cats, but it's in the same ballpark.  In the Third World, management practices are plain bad; common-sense policies like "Show up on time," "Pay for performance," and "Count the inventory and the money" are widely violated.  And remember, the whole world used to be the Third World!  What changed?  Common-sense slowly caught on - though as Alex Tabarrok reminds us, even First World businesses have ample room for improvement. 

So that's my story.  The foundation of the modern economy is not teachers, but craftsmen and bosses.  Not book learning, but experience and leadership.  Not studying, but doing and directing.  As a professor, this doesn't do much for my ego.  But why should everything be about me?

Update: Many readers have listed "tools" or "capital goods" as a third critical factor.  This is perfectly reasonable if you're just trying to explain GDP/worker.  I'm trying to explain why "worker productivity holding capital constant" is so high.

COMMENTS (17 to date)
Lewis writes:

Practice under the tutelage of experienced people. Sounds a lot like education to me

E. Harding writes:

I'd also add "changing technology", though that could go under #1.

Lewis, employer training is much more job-relevant than academic training.

Yes, experience and management are important, but so is education. They are not mutually exclusive.

Certainly at a largely-engineering school (like say MIT) there's a lot of practical knowledge provided. I'd argue that's true not only in engineering courses (principles of software engineering that endure beyond a particular programming language or application), but also in humanities and social sciences (how to write well, how to understand economic principles, etc). Perhaps most importantly, progress in science is what drives more fundamental advances (semiconductors, nuclear power, advanced materials, medicine, pharma, etc). Hard to see that happening well without advancing and learning the core science in something that, whatever you want to call it, will look like a research university.

Some of the education is consumption (how to appreciate art and music), but much is also productivity-enhancing.

Floccina writes:

Seems like an easy question:

  • Schooling>Education


    Physical capital

    Organizational capital

    Increasing division of labor


Alan Goldhammer writes:

Computer programming is really a bad example for #1. I as well as a lot of other people I know who are good programmers are largely self taught. My last computer programming class was in 1969 learning Fortran on an IBM mainframe with punch cards. Everything since that time (Basic, C, C++, Excel) is self taught.

Ben Pacini writes:

There's something to be said here for specialization.

I would submit that a 1700's farmer likely knew more than I do in my current job--but it was about a great number more things, and more of it was probably wrong.

A programmer learns a tremendous amount, but it's focused. They don't need to repair their axes, or make new plows, or hunt game. They focus their efforts.

Along with that specialization has also come increases in IQ, probably because of fewer health challenges.

But it's also worth noting that IQ is malleable, and that education--done right--can increase IQ. Which brings us back full circle.

andy writes:


Programmers are self-taught a lot. But you have lots of documentation/information at your disposal and you have lots of libraries to use. And lots of it is free (as it incidentally has zero marginal cost). And we have lots of it.

JFA writes:

1) I wish Caplan would distinguish between schooling and education.

2) Most programmers are not "self-taught". Most programmers don't go through the day saying, "I wonder what this bit code would do if I changed this part" and not asking anyone for help. Programmers (especially those who claim they are "self-taught") set a goal, and then go to stack overflow to either ask others how to do it, or (more likely) someone already asked the question and you are using the answers to those questions to achieve your goal. Is it tutelage? Maybe not (maybe its tutelage of the crowd). Is it self-taught? Only on the broadest definition of self.

Also, "self-taught" programmers have the worst programs.

Wes Winham writes:
Practice under the tutelage of experienced people. Sounds a lot like education to me

But experienced at what?

Anecdote: All competent programmers use a tool called source control, with git being the most popular. Not using it is like a home builder using hammers instead of nail guns.

Because the average Computer Science professor never worked on real software, source control is often not used in class. To pick on a local university, most Purdue Computer Science graduates have never been taught to use this most-basic of tool. Employers complain to me about the months of wasted time teaching those most basic of software engineering skills.

In contrast, the coding bootcamps usually teach version control within the first couple of weeks.

Larry Sanger writes:

Bryan, if you're right, then maybe we pursue liberal arts education for reasons other than job training. But then, that's always been perfectly obvious to anyone with actual training in the liberal arts. Nobody who got a philosophy degree, for example, ever expected to be able to do any particular job upon graduation. That's pretty damn obvious. Hence the purpose of a philosophy degree is clearly not job training. I always knew that, after I got my B.A. in philosophy, I'd need to be trained in something else if I were to get anything other than unskilled or entry level work.

As to the argument in your post, it doesn't quite work. You begin by saying "the safe answer" is that worker productivity has nothing to do with education. (Your book cites comparative data that controls for such things as IQ and socioeconomic status of parents—making this startling claim "safe"—right?)

Practice and management are two sources of worker productivity, you posit—easy to concede—and neither involves "book learning." That you are able to find two things that don't immediately depend on book learning does not establish that book learning is unnecessary for worker productivity. That just doesn't follow.

While we are speculating, let me do a little. If we were simply to jettison all book learning from Western civilization and attempt to pass on the know-how for economic productivity via practice and management, I daresay civilization would break down in very short order, for several reasons. Book learning and more theoretical teaching of the sort typically sought from universities are important for many purposes, which might be at a few removes for writing code or drilling for oil, but which are very important nevertheless. What about petroleum research, which tells us where to drill? What about writing annual reports? Without quite substantive "book learning," executives won't be able to articulate their thoughts satisfactorily to each other or to shareholders. What about the persuasive skills needed to convince a VC to part with some cash? How about the analytical (even mathematical) skills needed decide whether a loan or investment makes sense? Then, of course, there is all the academic background deeply important to have under your belt before you enter higher level government positions, law, medicine, and engineering. These involve skills that need practice, you might say—I'd agree—but the best teachers for many such things are teachers and professors, obviously.

Tiago writes:


psmith writes:

centuries of eugenic reproduction in certain areas and stagnation elsewhere

(t. A Farewell to Alms)

Matthias Görgens writes:

Book learning is great for programmers (and others). But all you ever needed for book learning was a library (and the internet these days).

University does help with some learning about programming. But most programmers aren't that kind of programmers. (Another important function of university in practice is network building within people sorted by uni admission.)

Floccina writes:

In my post above it should have been "Schooling not equal to Education".

The editor much have removed a character

John Wentworth writes:

Apologies for my ignorance, but... how do we know productivity has increased after controlling for capital & tech? How is that measured? What's the data there, and just how much higher is it?

I mean, there's the obvious answer "workers are more productive because technology has improved", but that's so obvious that presumably there's some data indicating that productivity has improved even controlling for that - otherwise we wouldn't even be talking about this. So what data is that?

JK Brown writes:

Practice, but also repetition. Modern workers of curious minds devise improved methods simply because they work a task until they are familiar with every detail. When your work is dispersed and any one task is seldom, the goal is to accomplish the task and move to the next with little pay off for taking extra time to improve it by a few minutes or even an hour. But when those minutes and hours will add up, just as the savings a hundredth of a cent on some small part, then the increased productivity is incentivized.

On the education side, we can see the spread of education, but not from schooling but rather publications, traveling speakers and wider mixing. Until around 1800, the spread of new methods was hardly quick or widespread. If you didn't cross the ridge to the next town, you'd never learn of their improved methods. But as newspapers and magazines spread so did the desire to write of new methods for reads far and wide. We can throw in the original purpose of the Dept of Agriculture which was to spread improved methods to isolated farmers.

Yes, traders and sailors did see a wider mix of methods in manufacturing and farming, but unless they were inspired to take up such work back home, those methods didn't spread except as tales of wonder to those tied to the land by their work.

quanticle writes:

If the reason that worker productivity has improved is due to tutelage and management (which, if your telling, at least implicitly stem from cultural factors), then isn't that a fairly strong case against open borders? Tutelage doesn't scale easily, at least not for the intensive 1:1 training that's required to produce high-productivity workers. Management doesn't scale that easily either, as indicated by the aphorism about A players hiring A players, but B players hiring C players. As we see in the technology industry, it's often far better to let a position go unfilled than it is to have someone unqualified or low productivity in the role.

If you take it as a given that high productivity workers come from intensive training and excellent management, both of which are extremely scarce, how can you be in favor of unlimited immigration, which would squander this scarce resource?


Return to top