Bryan Caplan’s signaling model and on-line education

The very useful model is here, and there is further commentary from Bryan here.  My question is this: does the model imply that on-line education should succeed, or not?

Let’s say that education signals conscientiousness.  A purely on-line class, with no ogre standing over your shoulder to discipline you, should be blown off by those who are not conscientiousness.  The on-line class would seem to offer a better signal and a cleaner separation of types.

Alternatively, let’s say education signals IQ or some other notion of “smarts.”  On-line education would seem to offer less opportunity to get through by buttering up the teacher, spouting mumbo-jumbo in basket-weaving classes, and so on.  For better or worse, a lot of on-line education seems to be based on relatively objective tests.  Then on-line education would seem to offer a better signal of smarts.

One possible application of Bryan’s model might be this.  Income inequality is rising, so there is greater care to get the signal, selection, and screening right for top jobs.  Relatively high levels of education should be all the more discriminatory, and that may mean more on-line education.  In fact, in normative terms that might well be a problem with on-line education, namely its inegalitarian nature with regard to curiosity and effort and smarts.

Oddly, the signaling model could be true, but through an invisible hand mechanism — schools competing to separate quality in the most effective ways — you can end up with a state of affairs where upfront signaling costs are fairly low.  Imagine a chess school, needing to sort talent, and unable to teach its students very much, but setting up a quite cheap on-line tournament and declaring some winners.  Aren’t the Khan Academy users some really talented people?

Alternatively, through an invisible hand mechanism, if the learning model is correct, you could end up with an equilibrium in which upfront signaling costs appear to be relatively high, namely that you impose “taxes” to make sure people end up learning what they need to know.  Think Paris Island or KIPP schools.

It is important not to confuse “seeing high upfront signaling costs” with “the signaling model of education is essentially correct.”  They sound like they should go together, but quite possibly they don’t.


The $$$ cost of education, once factored into worker's utility in Bryan's model, immediately suggests that online education (which should be cheaper) would encourage more diploma-taking.

The conscientiousness argument pulls K upward but the lower cost of production pulls K downward; what is the net effect? The basic intuition that signals have to be costly still holds in Bryan's model.

Wake me up when graduates of online classes are recruited by Goldman Sachs the way history majors from Brown are.

Selection(ability) bias. Presently, the game is already over when the acceptance(Ivy) letter arrives. You will be sleeping undisturbed for centuries.

But you're focusing on the tip of the pyramid and ignoring the base -- there are many, many time more students attending community colleges and regional universities that have essentially zero signaling value now (and whose students are - or certainly should be - highly price sensitive). Look for the revolution to start at the bottom and work its way up.

And Goldman Sachs et al are not looking for geniuses, they're looking for adequately smart hires who have been socialized into the mores of the upper classes and have shown themselves willing to do all the fraternizing/nose grinding/hoop jumping needed for the Ivy League.

Attending a traditional college signal conformity (which employers look for) while online education signals the opposite. Also, as there is yet an online school on par with the top universities in prestigiousness that's another signal you can't send.

What about xMIT ?

Online education is unlikely to take off unless cheating can be avoided. I cannot see how that will happen.

Auditing/Sampling. For example, one random test will require presence at a testing service. I also wonder how good current practice is at preventing cheating on a per-cost basis. Anti-cheating will likely require a greater degree of expense regardless of education regime.

I don't have a problem with Bryan's math, but I do with his assumptions:

'The only signal of IQ is a diploma, and a worker can only get one diploma.'

Ask yourself: what are the impediments, under various scenarios, to one getting a diploma, or, for that matter, with individuals with the same IQ, each getting a diploma under various scenarios.

Is the only signal of IQ a diploma, and does a person without a diploma signal they have a low IQ.

This is an amazingly comical piece of work. Because Bryan has no idea how students actually learn anything he's not teaching and become productive, he presumes that it's all a waste and that signalling fairies explain it all. As a honey-trap for liberal critics, he uses "IQ" instead of some other term for competence.

Ah, what else do we expect from the George Mason Sophistry Department? I hope Bryan does not expect a Nobel Prize in Economics for this. Perhaps the Kochs can endow a Nobel Prize in Propaganda.

Did you have an argument?

Btw, he uses IQ not because it inexplicably offends liberals, but in a pure signaling model THAT IS WHAT IT IS.

He's doing exactly what Krugman does and gets inexplicably lauded by liberals for, adding some overly simplified math to something very complicated to show his model isn't disproven by his model.

So the model is a useless tautology ?

A tautology, yes. Useless, no.

One of my ponderings is how much stress is required to actually change the brain to achieve actual learning. Another point for engineering education, although it also requires you to not have much else going on in your life.

Let's say education signals (in part) something approximating "socialization". Then the implications are the opposite, no?

Good point. It reminds me of the study a few years ago showing that dropouts who get a GED don't do any better in the job market than those who don't. The ones who get a GED are smarter, which often means that they dropped out not because they couldn't do the work but because they couldn't handle rules, bosses, etc.

The conundrum in all this is that it is easy for us to gain access to learning in the present, but exceedingly difficult to challenge, inspire, motivate, heal, help or interact with anyone on a purely individual economic basis. Until we allow ownership of knowledge skills in true usable and individual contexts, no additional educational choices can gain real marketable value.

I think (we) computer scientists have led in online learning for a number of reasons.

- we were here first
- or data is already on the computer
- we can demonstrate competence in the medium
- we never got all that personal interaction anyway

apologies, I wasn't clear what I meant by marketable value. For me the computer option is just as valuable as the book option. I meant that both need more value in the aggregate or larger sense of what they can ultimately provide, as some people lose faith in education altogether.

What are employers looking for? People bright enough to learn the job. People with enough training to show commitment and demonstrate the ability to complete tasks on time. In addition, employers will look at outside interests to see if you are a cultural fit and will work as part of a team (often through networking skills).

Ideally education should reflect all three core competencies.

Look at sports teams. Does a potential player have the basic skills or demonstrate the ability to learn basic skills. What is the players work ethic and have they demonstrated a commitment to improve. How do they fit into your team.

Some IQ level may be required but it is hardly sufficient to succeed (outside of a few places).

Traditional college students go through a "name brand" process that, in general, will test the three core competencies that I mentioned above. Current employees have gone through the system and you have experience with the "product".

Another option is to sub-contract work. In that case you are less interested in team fit etc. A potential may exist, in some fields (especially high demand fields), for independent contractors with agreed upon certification to set up shop. For example, automotive technicians with ASE certification. It is an imperfect signal of quality (and says nothing about trustworthiness) but it seems like a minimum requirement for employment. You have CPA for accountants etc.

Online education that tests specific required skills can be an effective signal in some cases. But only a partial murky signal.

Would it raise productivity in society? Not as much as public policies that encourage risk taking, rewards to creativity, and raise potential GNP by reducing regulations and taxes.

One can note that Western European nations spend less on their "crappier higher education systems" than the US, and yet get about the same outcomes when it comes to being wealthy, and certainly get better egalitarian outcomes. So, why aren't we copying their systems? Shouldn't the US being spending far less?

Edit function in Wordpress would be very nice. "Be spending".

1. You would have to accept a much more fixed social elite than in the USA. Think of France and its énarques. Bank executives become government ministers become transit CEOs, but almost always started out in one of a handful of schools. This is also the case with Britain's parliament; it is dominated by two (!) schools, and only parliamentarians take government office. This system also implies very early determination of one's fate; Germany's strictly streamed vocational schools don't hold much promise for kids, but maybe their relative social position will be better than that of their Detroit brethren.
2. You (and the rest of us in the Western world) would have to accept a rate of scientific innovation comparable to that of post-war Europe. We're good at sending our best biologists, economists, etc. to the US East Coast. We have a lot of practice.

You do realize that America's social elite is even less flexible than Europe's right? Like the guy above said, the game is over by the time your Ivy league acceptance letter arrives or doesnt. Unless you are comp sci autist and lucky.

Do you have a quote on the ability of European colleges to generate similar wealth outcomes? I haven't seen that data in a long time, but my impression was the opposite: Going to a college in Europe does much less for your future income than going to an American college.

I am talking society as a whole, not the individual. My point was to be provocative since I have often been fascinated by the seeming disconnect between the arguments about American healthcare and education when compared to the rest of OECD. John Thacker made a similar observation a few posts down.

European countries do spend less and generally have lower quality of higher education than the US. But I don't agree that they get the same overall outcomes, at least in STEM.

I think Europe and the US probably get similar outcomes in the middle of the competency range, where a degree signals conscientiousness, good literacy and numeracy, reasonable technical skills. You can definitely build a good workforce and economy on this...

But the US gets superior outcomes at the top of the competency range: The best Europeans in STEM often migrate to the US for graduate school or post-doc and employmnet. This cohort might not be large numerically, but it's of disproportionate importance for innovation.

In Europe, it's often harder to get into college. The US puts a lot more resources into providing college at the lower end of the competency range: The outcome here is probably mixed. There are advantages to not locking people out of college based on poor or even middling high school performance. At the same time, a lot of those resources are being wasted on students who make little or no progress...

Can an online degree signal that you're not a lemon?

Skills are one thing, but my impression is that many people want to avoid a really bad hire. So, to some extent, you do want the person who buckles down and writes a fifteen page history paper while juggling three other assignments due the same week, even if you don't care about history per se.

An online program with all elective classes and no enforced socialization seems like just the thing to stick you with a bright but problematic employee.

The flip side is of course that completing an online program demonstrates that you have the ability to actually buckle down when there is nothing around you pushing you to do so.

I still think the way forward is to disconnect the testing for proficiency in any subject from the education itself. It is still kind of shocking to think that the higher education system largely gets to evaluate it's own effectiveness itself, rather than having more third party testing of it.

Employers are the third party testers

Yes, and this is hideously expensive to do, and they don't really get to do it before they hire the person, either.

At the community college we are finding that our evening class enrollments are dropping while online increases. The traditional evening students (full time workers) do well in online classes. Other students, not so much.

I am interested.

Do you do any group projects requiring the students to work in a team? Do you have any times where you require the students to meet as a group? How important are social networks in the student later finding a job or getting a reference or referral from a classmate?

You can assign group projects and all of the learning management systems (LMS) support group projects. Students typically meet online and most communication is asynchronous. (Not unlike a multinational business spread across different time zones.) Social networks may develop - more so at the upper division and graduate level rather then lower level. Then again they may not.

The newer LMS's do take advantage of social networking.

Students that are working full time and choose to continue their education typically have good time management skills and tend to be more focused than other students. That makes them a good online student.

"Online Education" as a degree-granting signal may not be valuable as other commenter have noted, but publishing a portfolio of academic papers, design work, or software projects is a strong signal of competence that would not have been possible before the Internet.

Buttering up the teacher and spouting mumbo-jumbo are two core business skills.

I will say many of today's managers have a bias against employees who do all their work online, probably because it limits the manager's ability to signal bossitude ... "Johnson, get in here!" "I am virtual today; would you like to schedule a video conference?"

I bet a key component of the signalling of brick and mortar schooling is the ability to cope with unreasonable demands from management (professors); it is helped by the fact (or the reason why?) most professors are selected for research capability not teaching.

Isn't the answer at the moment a lot of both, online study with in person instruction. Tyler has previously mentioned that the near future higher education model should be a mix of online instruction, coaching(study guidance, feedback and strategy from grad students), and some traditional formal lecture teaching. This would require less physical space. I am a, non-traditional, Grad student and I cant think of one undergraduate lecture style class that would not have greatly benefited from forcing the professor to convey the same material in half the time. The worse offenders are the ones who could care less about lecturing, power point users, and those who think they are really good at, self-aware performers. Both kinds never evolve and are time wasters. Both kinds would be pushed out in the new model.

In finance, I have been told by many industry folks that simply being signed up to take the CFA exam is at least as valuable as having completed a masters degree in finance.

ya that is actually a lie. there are plenty unemployed CFAs or doing non finance work. The road to finance leads through an elite Ivy league undergraduate, and the further away from elite you are the more sophisticated your degree has to be. Thus an engineer from the University of Michigan and a Dartmouth history of art grad have the same odds of being recruited for a job with Morgan Stanley. The only hope a non-Ivey kid has is rolling the dice a second time and trying his luck at getting into a top 5 MBA program or the long arduous slog through law school, large law firm associate, lateral hire to a bank sometime in his mid 30s. Otherwise he either be an excellent computer programmer and join one of the quant shops or fuck off and die as far as Wall Street is concerned.

I wasn't referring to becoming a partner at Goldman Sachs. I was talking about us proles in flyover country, who might just try to pass our pitiful lives as portfolio managers in Wichita or something.

The majority of jobs do not require an enormous amount of intellect, something that academics can easily forget since their jobs do. Even for graduate jobs, employers care more about an applicant's ability to turn up reliably, work well with other people, and give answers when put on the spot than they do about IQ. Online classes neither teach nor test any of those things but regular classes do.

In addition, people trust people. So a teacher's assessment of a pupil will tend to carry more weight than an automated online assessment. Partly this is sentimentality but it's also partly due to the fact that teachers are harder to cheat. If you get asked a question in an online class you can look up the answer in a couple of places online and synthesise an answer which is very hard for anti-cheating systems to spot. If you're asked something more factual, you don't even need to go to that trouble.

Look at how online education works now. Western Governors University gives you some stuff to read and then they have you do assignments/pass a test. All courses are graded pass/fail. This is not designed to send a "high quality" signal, it's designed to provide a degree for jobs where a degree is just a box you check. Which is fine, we need the equivalent of a GED for college.

Look at how online colleges market themselves. They don't emphasize how tough they are to get in to or how demanding the curriculum is. They market to everyone and emphasize how quickly the degree can be completed. That's the quality signal they send.

What they do though is reduce the cost and effort of a college education, meaning that more people will have college degrees, meaning that more people will want graduate degrees to send a better signal.

"What they do though is reduce the cost and effort of a college education, meaning that more people will have college degrees, meaning that more people will want graduate degrees to send a better signal."

A couple of points. In our experience at the community college, online does not mean lower cost. What you save in physical buildings you more than make up in additional labor and IT support. Online classes are typically capped at 25 while face to face classes cap at 36.

You can scale up online education faster because you don't have to build buildings. That also means you can target a much larger market. Hence the rapid expansion of online. The expansion is not because it's cheaper.

As for reduced effort, yes there is less effort in being at a particular place at a particular time. The student will put in more effort if you count more reading and writing as additional effort.

"Parris Island" has two "r"s

Good discussion, if I may say so.

The technology of delivery -- on-line vs. on-site has to be separated from the idea of the outcome. One can have present outcomes at far less cost, e.g. by removing the country club aspect of some undergraduate education, and removing the research capability of the undergraduate faculty. That is, after all, not necessary for teaching , well, whatever.

The point is that alternative delivery modes will enable some to succeed somewhere--we just don't yet who. As for the rest, it could be delivered at a lot less cost.

I'll state my bias first: I am (in chronological order) a STEM graduate, worker, supervisor and, now, manager of a very profitable high tech manufacturing company.

I suspect that signalling is of much less importance in STEM, compared to economics or to any career in which an MBA is highly regarded. If you're going to make stuff that works, ya gotta know actual objectively verifiable facts, especially if said facts are going to be checked, tested, measured and verified by someone who doesn't speak your native language, who is in another country and who has never heard of you or your big deal alma mater.

It's a complicated subject but here's a vignette that illustrates the point: one simple test I do when recruiting staff is to give them an object, a vernier caliper, a pencil and a piece of paper and say "You have ten minutes to produce, freehand, an engineering drawing of that." Some candidates are insulted by the menial task. Tough. I don't care how much their education cost or which Senator their Daddy knows. If they can't do it, the interview is over. If they learnt how to do it on-line, that's fine by me but I probably wouldn't even ask how they acquired the knowledge and skill.

Forget the signaling. The high multiple of the speed of learning online (that's right - several times) is enough. With three times the speed and increased quality of the learning the pure acceleration of skill should be more than enough to achieve your goals. Personally I manage a full time study schedule online despite having no time at all for ordinary studies.

It will do very well for majors that are primarily learning based and not signaling based. I suspect.

What if having an ogre standing over you having you work on material that doesn't relate to you actually deters curiosity, and with it conscientiousness?

On-line education will fail and has gone nowhere because college now serves as nothing more than a licensing mechanism, like the law licenses of lawyer or the medical licenses of doctors, or, increasingly, the ever-more-expensive and more complex licensing requirements now being imposed upon such mundane professions as hairdresser, masseur, et al.

As offshoring exponentiates and as AI + databases + computers + the internet automate more and more tasks, an ever-larger fraction of the American population becomes unable to find work. (And wait till google's self-driving car hits the market in quantity: what do you think will happen to all those delivery driver jobs and bus driver jobs?) As work becomes an increasingly precious commodity, basic economic theory tells us that ever-increasing barriers to work will be raised in order to reduce the tidal wave of people applying for ever fewer jobs.

The two most obvious and simplest ways to raise barriers to entry into the workforce involve [1] high upfront entry fees; in medicine or law these translate into enormously expensive law school and medical school fees, but increasingly common state colleges are raising their tuitions at such a rate that they'll approach the same entry-barrier amounts of money. To put it bluntly, don't think about going to law school nowadays unless your daddy's got a quarter million to burn on your law school tuition. [2] artificial entry requirements designed to sieve out most of the applicants. In the case of the elite Ivy League colleges these includes the requirement of not only stellar GPAs and stratospheric SAT scores, but conveniently include "legacy" admissions for the children of not-very-bright but immensely wealthy former alumni.

On-line college education would reduce both of these artificial barriers to entry and thus are dead in the water.

Quite the contrary, in fact -- look for more and higher barriers to entry into college as time goes on and the American middle class melts away. Soon, you'll need to pass (let's say) something like a genetic screening, a full background check, a deep data-mining search, and various other barriers before you can be admitted to college. That's in addition to having a greater-than-4.0 GPA (which students can get nowadays by getting straight As in advanced placement courses; the old 4.0 average is no longer remotely good enough), having perfect 1400 SAT scores, and having the usual required additional community service etc on your high school transcript.

The problem, of course, is that as the American middle class vanishes, competition for entry into college will become increasingly frantic, and as a result, barriers to entry will rise exponentially.

Dying collapsing societies typical exhibit these kinds of pathologies, with many artificial licensing requirements to prevent the ever-growing underclass from infiltrating overrunning the increasingly tiny ruling class, and an eventual imposition of a virtual (or actual) caste system of the kind Emperor Diocletian imposed on Rome.

Comments for this post are closed