Whenever a new semester begins, I thank the universe for good students.  Good students have four key traits:

First, good students genuinely want to learn.  They don’t study material
merely because they see it on the syllabus or expect it on the test.

good students fight the natural human tendency to forget material right
after the final exam.  Unlike most students, they consciously choose to
try to remember what they learn.

Third, good students strive for what educational psychologists call Transfer of Learning.  They earnestly try to apply what they’ve learned outside the classroom.

and perhaps most importantly, good students put Truth first.  They
aren’t afraid to entertain and embrace socially unacceptable ideas.

I’ve never taught Scott Alexander, but when I read his take on labor economics, I immediately identified him as a most excellent student.  It’s an extensive critique of what Scott sees as the libertarian view of labor-capital relations.  In truth, though, his critique is broader, because Scott targets models that economists across the political spectrum take largely for granted.

Since my latest labor economics class is now in session, this seems like a great time to sift through Scott’s critique of my subject.  Since most of you haven’t read what he has to say, though, this post will merely reproduce his key sections.  I’ll respond later this week.  Here’s Scott:


2.5: How do coordination problems justify labor unions and other labor regulation?

It is frequently proposed that workers and bosses are equal negotiating
partners bargaining on equal terms, and only the excessive government
intervention on the side of labor that makes the negotiating table
unfair. After all, both need something from one another: the worker
needs money, the boss labor. Both can end the deal if they don’t like
the terms: the boss can fire the worker, or the worker can quit the
boss. Both have other choices: the boss can choose a different employee,
the worker can work for a different company. And yet, strange to
behold, having proven the fundamental equality of workers and bosses, we
find that everyone keeps acting as if bosses have the better end of the

During interviews, the prospective employee is often nervous; the boss
rarely is. The boss can ask all sorts of things like that the
prospective pay for her own background check, or pee in a cup so the
boss can test the urine for drugs; the prospective employee would think
twice before daring make even so reasonable a request as a cup of
coffee. Once the employee is hired, the boss may ask on a moment’s
notice that she work a half hour longer or else she’s fired, and she may
not dare to even complain. On the other hand, if she were to so much as
ask to be allowed to start work thirty minutes later to get more sleep
or else she’ll quit, she might well be laughed out of the company. A
boss may, and very often does, yell at an employee who has made a minor
mistake, telling her how stupid and worthless she is, but rarely could
an employee get away with even politely mentioning the mistake of a
boss, even if it is many times as unforgivable.

The naive economist who truly believes in the equal bargaining position
of labor and capital would find all of these things very puzzling.

Let’s focus on the last issue; a boss berating an employee, versus an
employee berating a boss. Maybe the boss has one hundred employees. Each
of these employees only has one job. If the boss decides she dislikes
an employee, she can drive her to quit and still be 99% as productive
while she looks for a replacement; once the replacement is found, the
company will go on exactly as smoothly as before.

But if the employee’s actions drive the boss to fire her, then she must
be completely unemployed until such time as she finds a new job,
suffering a long period of 0% productivity. Her new job may require a
completely different life routine, including working different hours,
learning different skills, or moving to an entirely new city. And
because people often get promoted based on seniority, she probably won’t
be as well paid or have as many opportunities as she did at her old
company. And of course, there’s always the chance she won’t find another
job at all, or will only find one in a much less tolerable field like
fast food.

We previously proposed a symmetry between a boss firing a worker and a
worker quitting a boss, but actually they could not be more different.
For a boss to fire a worker is at most a minor inconvenience; for a
worker to lose a job is a disaster. The Holmes-Rahe Stress Scale, a
measure of the comparative stress level of different life events, puts
being fired at 47 units, worse than the death of a close friend and
nearly as bad as a jail term. Tellingly, “firing one of your employees”
failed to make the scale.

This fundamental asymmetry gives capital the power to create more
asymmetries in its favor. For example, bosses retain a level of control
on workers even after they quit, because a worker may very well need a
letter of reference from a previous boss to get a good job at a new
company. On the other hand, a prospective employee who asked her
prospective boss to produce letters of recommendation from her previous
workers would be politely shown the door; we find even the image funny.

The proper level negotiating partner to a boss is not one worker, but all workers. If the boss lost all workers at once, then she would be at 0% productivity, the same as the worker who loses her job. Likewise, if all
the workers approached the boss and said “We want to start a half hour
later in the morning or we all quit”, they might receive the same
attention as the boss who said “Work a half hour longer each day or
you’re all fired”.

But getting all the workers together presents coordination problems. One
worker has to be the first to speak up. But if one worker speaks up and
doesn’t get immediate support from all the other workers, the boss can
just fire that first worker as a troublemaker. Being the first worker to
speak up has major costs – a good chance of being fired – but no
benefits – all workers will benefit equally from revised policies no
matter who the first worker to ask for them is.

Or, to look at it from the other angle, if only one worker sticks up for
the boss, then intolerable conditions may well still get changed, but
the boss will remember that one worker and maybe be more likely to
promote her. So even someone who hates the boss’s policies has a strong
selfish incentive to stick up for her.

The ability of workers to coordinate action without being threatened or
fired for attempting to do so is the only thing that gives them any
negotiating power at all, and is necessary for a healthy labor market.
Although we can debate the specifics of exactly how much protection
should be afforded each kind of coordination, the fundamental principle
is sound.

2.5.1: But workers don’t need to coordinate. If working conditions
are bad, people can just change jobs, and that would solve the bad

About three hundred Americans commit suicide for work-related reasons
every year – this number doesn’t count those who attempt suicide but
fail. The reasons cited by suicide notes, survivors and researchers
investigating the phenomenon include on-the-job bullying, poor working
conditions, unbearable hours, and fear of being fired.

I don’t claim to understand the thought processes that would drive
someone to do this, but given the rarity and extremity of suicide, we
can assume for every worker who goes ahead with suicide for work-related
reasons, there are a hundred or a thousand who feel miserable but not
quite suicidal.

If people are literally killing themselves because of bad working
conditions, it’s safe to say that life is more complicated than the
ideal world in which everyone who didn’t like their working conditions
quits and get a better job elsewhere (see the next section,

I note in the same vein stories from the days before labor regulations
when employers would ban workers from using the restroom on jobs with
nine hour shifts, often ending in the workers wetting themselves. This
seems like the sort of thing that provides so much humiliation to the
workers, and so little benefit to the bosses, that a free market would
eliminate it in a split second. But we know that it was a common policy
in the 1910s and 1920s, and that factories with such policies never
wanted for employees. The same is true of factories that literally
locked their workers inside to prevent them from secretly using the
restroom or going out for a smoking break, leading to disasters like the
Triangle Shirtwaist Fire
when hundreds of workers died when the building they were locked inside
burnt down. And yet even after this fire, the practice of locking
workers inside buildings only stopped when the government finally passed
regulation against it.