Bryan Caplan  

Overselling Online Education 2.0: Comment on Carey

PRINT
The Henderson Misery Index... Everyone needs to be accountab...
First-generation fans of online education pushed a simple story: Online education is a better, cheaper way to learn, and will therefore soon do to brick-and-mortar colleges what downloading did to the record companies.  I've long argued that these fans are naive.  The status quo doesn't just have hundreds of billions of government subsidies on its side.  It also has the power of conformity signaling to shield it from disruptive innovation. 

I'm pleased to see, then, that second-generation fans of online education have arrived.  Kevin Carey's recent NYT piece explains the intellectual evolution:

Three years ago, technology was going to transform higher education. What happened?...

[E]nrollment in traditional colleges remains robust, and undergraduates are paying higher tuition and taking out larger loans than ever before. Universities do not seem poised to join travel agents and video stores on the ash heap of history -- at least, not yet.

The failure of MOOCs to disrupt higher education has nothing to do with the quality of the courses themselves, many of which are quite good and getting better. Colleges are holding technology at bay because the only thing MOOCs provide is access to world-class professors at an unbeatable price. What they don't offer are official college degrees, the kind that can get you a job. And that, it turns out, is mostly what college students are paying for.

So far, so good.  But Carey remains imminently optimistic:
Now information technology is poised to transform college degrees. When that happens, the economic foundations beneath the academy will truly begin to tremble.
Carey's fandom is a big improvement over first-generation fandom, but remains naive.  Here's my point-by-point critique.  Carey's in blockquotes, I'm not.

Traditional college degrees represent several different kinds of information. Elite universities run admissions tournaments as a way of identifying the best and the brightest. That, in itself, is valuable data. It's why "Harvard dropout" and "Harvard graduate" tell the job market almost exactly the same thing: "This person was good enough to get into Harvard."

No, they don't.  Degrees signal an array of traits: not just intelligence, but work ethic, conformity, and more.  "Harvard dropout" tells the job market, "This person was promising enough to get into Harvard, but so lazy and/or non-conformist that he wasted this golden opportunity." 

Most important, traditional college degrees are deeply embedded in government regulation and standard human resources practice. It doesn't matter how good a teacher you are -- if you don't have a bachelor's degree, it's illegal for a public school to hire you.
A fair point, but overrated
Private-sector employers often use college degrees as a cheap and easy way to select for certain basic attributes, mostly the discipline and wherewithal necessary to earn 120 college credits.
Apparently "Harvard dropout" doesn't say the same thing as "Harvard graduate" after all.  In any case, Carey omits a vital "basic attribute": conformity to social norms.
The Mozilla Foundation, which brought the world the Firefox web browser, has spent the last few years creating what it calls the Open Badges project. Badges are electronic credentials that any organization, collegiate or otherwise, can issue. Badges indicate specific skills and knowledge, backed by links to electronic evidence of how and why, exactly, the badge was earned...

The most important thing about badges is that they aren't limited to what people learn in college. Nor are they controlled by colleges exclusively. People learn throughout their lives, at work, at home, in church, among their communities.

It's very hard to see what's revolutionary about these "badges." Individuals and organizations have always been free to award them.  See the Boy Scouts.  How does switching to "electronic evidence" fundamentally improve them?

The fact that colleges currently have a near-monopoly on degrees that lead to jobs goes a long way toward explaining how they can continue raising prices every year.

"Near-monopoly"?  Anyone is free to issue degrees.  The problem is that, outside of key niche occupations, employers take traditional academic degrees much more seriously than "badges."  Why should we expect that to change? 

Inevitably, there will be a lag between the creation of such new credentials and their widespread acceptance by employers and government regulators. H.R. departments know what a bachelor's degree is. "Verified certificates" are something new. But employers have a powerful incentive to move in this direction: Traditional college degrees are deeply inadequate tools for communicating information.
It depends on what information employers are looking for.  In our society, traditional college degrees remain the only dependable way to communicate, "I'm a smart person who conforms to social expectations."  We can easily imagine societies that don't work this way.  But ours does.
The new digital credentials can solve this problem by providing exponentially more information. Think about all the work you did in college. Unless you're a recent college graduate, how much of it was saved and archived in a way that you can access now?
Why should we believe that employers even want all these extra details?  If you've ever been involved in hiring, the main problem is information overload: Hundreds of applicants with diverse backgrounds and talents.  To be fair, Carey anticipates this objection:
This does present a new challenge for employers, who will have to sift through all this additional information. College degrees, for all of their faults, are quick and easy to digest. Of course, processing large amounts of information is exactly what computers are good for. Scientists at Carnegie Mellon University are designing open badges that are "machine discoverable," meaning that they are designed to be found by employers using search algorithms to locate people with specific skills.
It's easy to believe that Big Data hiring will slowly become more important.  But why the continued assumption that employers hunger for "specific skills"?   Current hiring heuristics reveal rather different motives.  Sure, employers want workers with up-to-date practical experience.  But employers focus at least as much on workers' general competence and people skills. 

In the long run, MOOCs will most likely be seen as a crucial step forward in the reformation of higher education. But their true impact won't be felt until students and learners of all kinds have access to digital credentials that are also built for the modern world. Then they'll be able to acquire skills and get jobs for a fraction of what colleges cost today.

I wish Carey were right, but he's not.  Like his predecessors, he neglects my standing two-part critique.  Namely:

1. Due to conformity signaling, the status quo has a massive built-in advantage. 

2. Governments at all levels annually cement the status quo's advantage with hundreds of billions of dollars of subsidies.

If Carey or anyone else disagrees, I renew my offer to bet on the future of traditional college enrollment.  Happy to update the time frame, of course.

P.S. I used to restrict my bets to people with some cyber-reputation to lose.  But now I'm happy to bet anyone, anywhere who trusts me.  The system: When the bet starts, PayPal me whatever you owe me if you lose.  If you win, I refund your money + whatever I owe you + some interest if you like.




COMMENTS (17 to date)
Radford Neal writes:

You're right that getting an education at Harvard signals certain things beyond what you learned while there. But consider that demonstrating you have the same knowledge as a Harvard grad *without* having gone to Harvard, or any university, also signals something beyond what you learned. It signals that you have the ability to learn on your own (without personal hand-holding, though of course with the help of books, MOOCs, etc.), and the discipline to actually do that, without the constant motivational incentives of grades, regular lecture times, etc.

At some point, an employer may think, "He went to Harvard? Not smart enough and tough enough to learn on his own, I guess..."

Ricardo Cruz writes:

How do MOOCs compare with Open Universities?

There exist Open Universities for at least two decades.

UK: http://en.wikipedia.org/wiki/Open_University
Portugal: http://en.wikipedia.org/wiki/Universidade_Aberta
(probably many others ... these are the ones I know about ...)

Why have they not gain traction?

These institutions exist for far longer than MOOCs, and the finals take place in physical universities, so there is more legitimacy to them than doing exams over the "Internet", where anyone can do the evaluation for you.

Furthermore, Coursera, as an example, is a great place to learn but most courses allow you to repeat tests several times over. I have managed to conclude courses with 100%. Work assignments is also a joke; peers evaluate the work done by their peers, which sounds cool, but in my experience I saw little correlation between my effort and my grades. In countries with tough labor regulations, you cannot afford to hire someone on the basis of such primitive evaluation methods. I don't know why they don't partner with a real university and final evaluation is performed there at the end of the school year, at least as a starting point. Sure, it would be more expensive for students (it would imply one trip per year plus enrollment costs), but it would add a lot more value for them as well.

Anyhow, I don't understand why any discussion of MOOCs does not start by trying to understand why Open Universities have an insignificant amount of market share. This is like studying WWII without first studying WWI.

Brian writes:

So when I read this, I generally agree with Bryan, but I want to see a look at professions that are exceptions to the rule and understand why.

Computer programmers are generally hired and paid without regard to college degrees. Actuaries are licensed based on exams that anyone can take; only about one semester of college classes is required and those can be taken online.

Those are professions where actual skills are required to do a good job and those skills are hard to develop and challenging to apply. People who get those jobs with social skills and hope generally are a drag on their organizations. The result is that actual skills are more important than signaling. It helps that most of the work doesn't require salesmanship and office politics; those would be best proven by conformity and striving measures.

Lawyering does require a hard exam, but you need a seven year degree first. (It's an undergrad degree -- you don't have to do any original work -- but they call it a doctorate for social signalling prestige.) A lot more people can pass the exam than there is work for, unlike programmers and actuaries.

CPAs require a hard exam, but also require a masters degree. Like lawyering, there seems to be more workers able to pass the exam than work for them.

More states are requiring more and more teacher exams under the ESEA. Lately I see that math teachers that can pass the math teacher subject exams are hired with higher salaries in better districts and the masters degree requirements and teaching degree requirements are waived. Other teachers are hired with a minimum requirement of six years of teacher training and a masters degree as a minimum. Apparently the existence of a legitimate exam has changed the credentialing and signaling preference drastically.

john hare writes:

My problem with bets of this nature is that I am not going to tie up money for extended periods without expecting very strong returns on investment. A piece of equipment or even ready cash can make a difference in my business over the next several years. OTOH I don't have an online reputation that would allow me to make an honor based bet. And besides that, I think it will take decades for the inertia of subsidies and signaling to change, so I wouldn't make this particular bet anyway.

Yaakov writes:

Israel has an open university. It is quite popular, but the main reason is because it actually has classes that help you go through the books. Most people in the open university do not learn on their own, but rather go to classes. This is because people do not like to learn on their own and feel they must have classes in which they will be taught.

This is even more evident from the psychometric test (equivalent to the SAT, I gather), which is required to get accepted to university or college. It is very easy to prepare for the test on your own. The overwhelming majority of applicants, however, take a course which costs a fortune, and delay going to university for a year to take this course.

MOOCs are great for the 1-2% elite of the population. Maybe if a very high standard MOOC was established, with tough tests and high esteem degrees, it could attract that elite. The rest of the people will probably continue demanding a teacher next to them.

Sam writes:

For anyone considering Bryan's bet (like I am), the updated table number is 302.60, and figures through 2013 can be found here: http://nces.ed.gov/programs/digest/d14/tables/dt14_302.60.asp?current=yes. The 4-year enrollment rate has dropped 4.4% in the 4 years since 2009 (it increased slightly 2009-2011, then fell).

Ricardo Cruz writes:

@Yaakov
"It is quite popular, but the main reason is because it actually has classes that help you go through the books."

Yes. But the proponents of MOOCs do not defend MOOCs because you can learn at your own pace - in Coursera and others, there are also classes and a schedule, and you have to take courses in the right order if you want the track certificate. The main reason why proponents of MOOCs push for MOOCs is the fact that costs are lower, and classes are better.

Thus my question. Why do they think MOOCs will work when Open Universities have failed. (Do you have any data on the market share of Open University in Israel?)

"MOOCs are great for the 1-2% elite of the population."

I disagree. I have all kinds of acquaintances that have tried Coursera to a lesser or greater extent. Within my former colleagues, the brightest ones were enrolled and had completed a bunch of courses there (they are usually pretty easy), but I knew students who were less smart that did some stuff there. The classes there (and in such sites as Khan Academy, and even Youtube) are much easier to follow than actual classes here, even if in English.

Bostonian writes:

I have a son in middle school. This summer he will take online intermediate Python programming and math contest preparation classes at Art of Problem Solving. He likes to program and thinks he may major in computer science in college. He is good at math, and doing well in math competitions may boost his college admissions classes.

Online courses will be used by some ambitious students with supportive parents (these classes cost a few hundred dollars each) to prepare for college and to burnish the college application. As Caplan says, I don't see the importance of the college degree falling.

Michael writes:
    "It's easy to believe that Big Data hiring will slowly become more important. But why the continued assumption that employers hunger for "specific skills"? Current hiring heuristics reveal rather different motives. Sure, employers want workers with up-to-date practical experience. But employers focus at least as much on workers' general competence and people skills."

I think this is the crux of the issue, but it is not insuperable. The general skills will remain important no matter what happens to the initial part of the hiring process - sorting out people with specific skills. Of course, I am assuming the candidate did not bypass the initial screening through connections. The soft skills part of the hiring process will remain intact in phone screens and live interviews. That is exactly where the arbitrary hiring decisions are heavily influenced by soft skills. However, employers do not even make it to this stage without first screening whether the applicant has specific skills that fit the job profile. This initial screening period is when education, job experience, etc. are checked. Of course, it is quite an easy heuristic to rely on degree signalling at this point. But, it is worth pointing out that if, a big if, search algorithms make the process of sorting out skills by verified certificates painless and employers begin to put actual stock in these certificates, then the initial screening could be altered significantly without changing the secondary, and arguably more important, soft skills screeening.

Dan Carroll writes:

The most likely route for MOOC's will take is for established universities to incorporate part or all of the technology into their existing curriculum. The greatest needs are the very large state universities that have been using taped lectures for years (since before I was in college in 1988) to extend very large entry level classes (I once took a class that had 1100 students in 1989 or 1990 - basic macro econ, which was the smaller of the two basic econ courses) and to reach geographically remote classrooms. The other point of leverage is for the university "brands" to extend their reach and consolidate the industry by (slowly) expanding enrollment. They may even tier their degrees (Harvard Elite, Harvard Good, etc.). Of course, the community college effect has already taken place with U of Phoenix type degrees.

This is all speculation, so I'm not ready to bet, but technology gets applied in ways that are hard to predict.

Yancey Ward writes:

It will take time, but I do believe that the traditional higher education model is on an unsustainable trajectory. At some point, the cost is going to cause just enough talented students to access the online versions that employers will start weighting them accordingly. We haven't reached that tipping point yet, but it is coming, and it will be the high-profile universities that finally start issuing the degrees for on-line students.

R writes:

I think there might be something no on-line university or MOOC degree has ever tried: compete directly with Harvard.

Try making a top-notch on-line university. Start by selecting students through very demanding exams. Offer tough, excellent courses.

Once you start having students who paid 1/4th what Harvard costs but who are just as smart there is going to be a real change. MOOCs and open university simply haven't tried to conquer the high-end market. They have not gone after the establishment. Yet.


OWG writes:

There are plenty of obstacles to overcome before online degrees will have much veracity. The “Badges” idea seems remote to me. There are already websites that offer to take online classes for you, so we have a long way to go to produce credible evidence that much has been learned in them.

Massimo writes:

MOOCs will grow, but as a supplement to the status quo, not as a replacement. Many employers value specific niche Coursera classes, but they are expected in addition to a masters degree, not as a degree replacement.

College enrollment rates may fall due to other factors, but not MOOCs.

Mark V Anderson writes:

Michael has it right. It is the initial screening where the college degree is the most important.

But I am a bit confused about folks saying that since MOOCs haven't become important qualifiers for education that on-line education has failed. What about U of Phoenix? Of course the problem is that it costs even more than other colleges.

Employers just want it to say BA on your resume. My guess is that U of Phoenix is as good as any other. It seems the problem is no one has come up with a cheap solution to provide college degrees and also make money themselves. There have always been matchbook colleges that sell degrees. Is the problem that no one can come up with a cheap on-line degree program that is accredited?

Jim Rose writes:

MOOCs Is a flash name for a really big lecture hall. We can give university lecturers at football stadiums for some time now.

Teaching at college in graduate school is a customer input technology. What matters most is the quality of the other members of the class.

The Smarter the class, the faster the pace, the more difficult the questions that can be put and the more you benefit from your classmates.

Just about every educational institutions sets a minimum admission standard to ensure that the pace of the class is to the preference of those who want to enrol.

Smart students do not want to have to go at the pace of not so smart students. That's why they enrolled in Ivy League universities. The quality of the classmate is more important than the quality of the lecturer.

Yaakov writes:

I believe the open university in Israel has the most students in any university in Israel, but probably its completion rates are very low. The same is probably true with MOOCs, you can learn one, two, three courses, but the completion rate is low. And since employers still want degrees and not just knowledge, people are willing to spend a lot of money on lower quality lectures, which have higher completion rates. If MOOcs / open universities can solve the drop out problem, they will have a better chance in competing in the market.

Comments for this entry have been closed
Return to top