Scott Sumner  

Second best policy arguments

A Question for Krugman... Identificationists Beware...

Economists use the term "second best policy" for a public policy that may not be optimal, but nonetheless might improve things if public policy is off target in other respects. Thus if we don't have a tax on pollution emitted from cars, a second best policy might requite all car companies to build cars that emit no more than a certain quantity of pollutants. Or one might argue that taxing "Cadillac" health plans is a bad idea when considered in isolation, but becomes a good second best policy when those plans are already being heavily subsidized via the tax deductibility of employer provided health insurance benefits.

Roughly 95% of the time I'm skeptical of second best policy arguments (although I think the two examples discussed above might be in the justifiable 5%.) In a recent post, Scott Alexander recently made a surprisingly powerful argument for a seemingly far-fetched second best policy. Here's the conclusion of his post:

If I were Sanders, I'd propose a different strategy. Make "college degree" a protected characteristic, like race and religion and sexuality. If you're not allowed to ask a job candidate whether they're gay, you're not allowed to ask them whether they're a college graduate or not. You can give them all sorts of examinations, you can ask them their high school grades and SAT scores, you can ask their work history, but if you ask them if they have a degree then that's illegal class-based discrimination and you're going to jail. I realize this is a blatant violation of my usual semi-libertarian principles, but at this point I don't care.
If you had merely showed me the conclusion presented above, with no supporting arguments, I would have thought the proposal was undesirable, even foolish. After reading his post I'm not so sure, indeed I wonder if it isn't a good idea. It's a surprisingly persuasive post. (Although given that it's written by Scott Alexander, I guess I shouldn't be surprised.)

PS. If you are short of time, skip over section one of his post, and start at section two.

PPS. Scott also has doubts about K-12, expressed in the form of the best graduation speech ever.

Comments and Sharing


COMMENTS (16 to date)
ThomasH writes:

I think the minimum wage may be another. The best way to raise the incomes of low income workers would be to increase the EITC (and or exempt the first x$ of wages from payroll taxes), but if that's off the table, then a modest increase in the minimum wage may be "second best."

lrC writes:

But the information is likely to be on a resume, or volunteered during an interview, since it is important in a way that other "protected" attributes are not. An interviewee could lie, but then if in the course of work the truth comes that misrepresentation would be grounds for termination of employment.

An idea with little practical application.

maynardGkeynes writes:

I think Scott Alexander's post, which I enjoyed, had more to do with Jonathan Swift than Pareto.

Richard writes:

That was an excellent post. Of all the libertarian causes, this should be on the top of the list.

We require people to go to school for 12 years, 5 days a week, 7 hours a day. And then, it requires at least another four years of college to signal something to your employers.

And if all this is for nothing, just a useless arms race and signaling, that's a really big deal. Not only the money spent on "education" in this country, but the opportunity cost of the best minds of the nation spending a huge portion of their lives on useless tasks.

Hopefully school vouchers are the first step towards ending this insane system.

Scott Sumner writes:

Thomas, Possibly, but I doubt it.

IrC, What do companies do when they later find out that employees lied about their credentials? I had thought they did nothing, if it was a good employee.

Richard, I've seen the education system from both sides (student and teacher) and it seems to me that there is an awful lot of waste.

Zeke writes:


Maybe Scott intended to make it illegal to volunteer the information? Otherwise, it seems to me that those with excellent credentials would offer them and those with lousy credentials would not offer them. As a result, absence of credentials would by implication be the same as bad credentials.

Also, what employer wouldn't do a LinkedIn/Facebook search of an employee before higher the employee?

Lying would be very hard to accomplish. Many things would have to be forged.

It seems to me any such system adopted would add very little value and cause significant compliance cost.

Nathan writes:

As is so often the case with Scott Alexander, this argument is best appreciated not as "This is something we should do" as "This is something clearly insane that we could do that would still be better than what we are currently doing."

Jay writes:

Didn't we resort to all this college degree signalling primarily because examinations and IQ tests open employers up to discrimination lawsuits?

I'm with Zeke and Irc above, it would be very easy to tell with >95% accuracy (from HS grades, google/social media searches, etc) whether or not someone went to college even if they made volunteering the information illegal as well. You'd get into all sorts of fuzzy legal language as to what constitutes "asking", could I as a college grad put my name on a 3rd party list of graduates and then companies have open access to that?

Scott Sumner writes:

Nathan, Exactly. Very well put.

Jared writes:

Wouldn't it be enough to simply end government subsidized loans? There is no way that students could afford to spend tens of thousands of dollars without those loans and no bank would give them the money. Schools would be forced to cut their prices in order to stay open. Perhaps it's not politically popular right now, but that can change.

Chris H writes:

Schooling (particularly high education) is currently over supplied for optimal welfare due to signaling. So how do we use economics to handle over supply?

Well normally we do so by raising the price of that good. But we've been doing that and whenever said price starts getting high enough to stop dropping off some of the college seekers we increase student aid. Given that, price rises either through natural market actions or taxes seems politically impossible.

Scott's idea has merit, but produces the problem that a mandate to "don't ask people whether they went to college" may wind up being unenforceable. From vocabulary to four years after high school with no work history it won't take employers long to piece things together.

But what about price controls? Economists normally disdain price controls because they have the effect of decreasing the supply of the very thing that people are trying to gain access to. Well that's what we want! And non-economists seem to have problems understanding the problems with price controls. So let's have a policy of intense price controls and no subsidies, thereby reducing supply as well as encouraging the remaining universities to focus on the most valuable assets (which is probably good teaching rather than a cool campus life).

It's a second best policy, but probably still in the category of "less crazy than the status quo."

David Friedman writes:

Scott is, as usual, wonderful, but I think he slightly understates his case. If you are the sort of person who might learn about the wonders of philosophy, or poetry, or mathematics, conventional schooling may well lower the chance that you will do so by presenting those as things you are being ordered to learn and are not really expected to like. The odds are probably better with unschooling, where you are learning things you want to learn.

But I may be biased, having unschooled two children, now adults. One of whom has just finished his first novel, a craft we made no effort to teach him.

Scott Sumner writes:

Jared, Yes, that would be a first best policy.

Chris, Or make the student loans only sufficient for the cheapest colleges.

David, I agree. I doubt many famous novelists learned their craft in school. They probably learned from reading other novels.

Glen Smith writes:

Exams for hiring open up employers to discrimination suits?Guess that means I could make money by suing every potential employer I've ever had or their parent companies.

Floccina writes:

It seems that voters prefer to get Government (taxpayers/hopefully rich people) to pay for things rather than try to make things cheaper. If people all decided to spend less on education would be cheaper.
Same with healthcare. Call me crazy but I do not see what is so bad about paying in arrears for healthcare. It might motivate providers to get us back to work quick.
I think that we only need health insurance because most people have health insurance. People complained that health insurance costs too much but mostly PPACA is about getting other people to pay.

Prakash writes:

An interesting second best policy i read somewhere was linking the minimum wage in an area to the cost of something like a standard apartment. The areas that restrict housing will also have higher minimum wages, driving business away. Areas with more relaxed housing standards get a lower minimum wage, attracting business.

Comments for this entry have been closed
Return to top