The GMU/UVa wage disparity and the signalling model of education

It’s a well-known fact — well-known around GMU that is — that GMU graduates earn higher average salaries than do UVA grads (direct link here), that is for four year undergrads in their first year of employment.

It’s not just that UVa is in decline, or that some of them end up richer later in life.  Or others may use their wealthier parents to live in Williamsburg, Brooklyn and avoid direct employment.  A major reason for the wage discrepancy is simply that a disproportionate chunk of GMU students are likely to get jobs in the relatively high-paying Washington, D.C. area.

OK, so how does this relate to the broader ongoing debate over the signaling theory of education and wages?

It is widely accepted that UVa is a more exclusive school than GMU by the usual standards.  Yet here we see labor markets “seeing through” those credentials, and paying more to the GMU graduates.  In other words, labor markets are seeing that GMU students are, on average, “less exclusive by origin but will have a higher marginal product very quickly.”

The signaling model, in its simplest, most stripped down form, assumes that employers cannot judge the marginal products of individual new hires but instead pay them according to their credentials.  Yet here we have a case where employers seem quite willing to make a judgment about marginal product and indeed that is a judgment which contradicts data on exclusivity of academic origins.  Once you postulate that employers are willing to make estimates of individual marginal products which differ from the rankings that might be given by “raw ability,” the signaling model is  less applicable.  I don’t want to claim that the wages converge exactly on marginal products, but the credentials clearly are just one factor of many.  Employer judgments of expected marginal products are not dominated by credentials, and you can imagine that after having a worker for a year or two the credentials are even less important as a means of judging prospective marginal product.

Another way to put this point is that the speed of employer learning is in fact fairly rapid, and some of it happens before the job even starts.


UVA graduates many more of its students than GMU. Access to DC may be more valuable for those who do graduate from GMU, but presumably the high likelihood of graduating in four years is valuable to the UVA students, particularly the many students who do not seek employment. Specific numbers from Kiplingers: UVA students graduation rates for 4/6 years (85/93%); GMU graduation rates for 4/6 years (41/65%).

I have to defend GMU just a bit here - a number of its students are not exactly students. They are people taking a class or two to polish a resume, get/keep a certification, etc. In the past, GMU has built entire programs just to meet their needs, actually. Again, such programs aren't a joke, they are a business plan.

What Joe is getting at (I think) is that the signal provided by UVA is largely defined by its admissions process. Once you get through the admissions screen, you're extremely likely to become a UVA grad. Whereas the signal from GMU is defined by two screens, one at the admissions level and another screen via the coursework and flunking out of students.

Employers may recognize that the two-screen method ends up signaling higher quality, even if neither of the two screens alone is as fine as the single screen at UVA.

I doubt employers have any idea about the comparative graduation rates of the institutions. Even if they did, they'd probably just chalk it up to GMU having inferior students, or like prior approval says more part-time and "non-traditional" students. It may also be that within majors, graduation rates are closer, but the institutions have different compositions of majors.

It is true that UVA has a closely guarded underground reputation as "hard to get in, easy when you get there", but based on my experiences this is limited to alumni. If anything, though, I think this is more likely to fit in with the human capital model than the signaling model though. Some UVA students basically breeze through high school and college without ever having to work particularly hard or consistently. If GMU requires more effort from its students, that itself is an important work skill regardless of whether they're learning anything worthwhile. That always seems to be the complaint about recent college grads - tech-savvy, flexible, smart, but no work ethic/professionalism.

I don't know that much about GMU, but I could draw a similar analogy about Drexel and Villanova in Philadelphia. Villanova is clearly the more prestigious school, but I know when I hire a Drexel grad (in certain majors at least), I'm getting someone who is willing to scrape and claw and do whatever it takes to get the job done, whereas I know if I get a Villanova grad I'm getting someone who was smart enough to get into Villanova.

Having said this,

"simply that a disproportionate chunk of GMU students are likely to get jobs in the relatively high-paying Washington, D.C. area."

I don't get this:

"Yet here we see labor markets “seeing through” those credentials, and paying more to the GMU graduates. In other words, labor markets are seeing that GMU students are, on average, “less exclusive by origin but will have a higher marginal product very quickly.”

Is there any evidence that when GMU and UVa students (of the same age) apply to the same job that the GMU ones get hired, or at least get hired at the higher paying cross app jobs?

Nothing in the argument requires it to be "for the same job"...

I get that, but for the market to be "seeing" higher marginal product, then GMU should win head-to-head. Otherwise, it is more like GMU students seeing the market by choosing higher pay majors or GMU itself seeing the market by providing more programming for higher pay majors. I see that directionality to be important.

I had a similar thought as liberalarts. Surely self selection is possible? My sense, which is purely anecdotal, is that ambitious students at different institutions seem to have different goals. Imagine that in one student population all the best students become graduate students or get fellowships while in another student population all the best students get jobs in consulting or finance.

Even if students in the first population are of higher objective ability, wouldn't the second population be expected to do better in this data?

I'm not saying this is the dominant effect in the UVa GMU comparison as I don't know enough about either school.

If anyone has any numbers to support\refute the theory above, I'd be very interested.

Or apparently, the same geographic location.

'Yet here we see labor markets “seeing through” those credentials, and paying more to the GMU graduates.'

So, the same percentage of UVA graduates apply for jobs with the federal government as GMU graduates? Though, apparently, that is unimportant, in light of the following -

'The signaling model, in its simplest, most stripped down form, assumes that employers cannot judge the marginal products of individual new hires but instead pay them according to their credentials.'

So, is the number of former military graduates of GMU, in an older age bracket, as high as it used to be? And what would that say about credentialing (not to much mandated preference for federal employment when a verteran applies for a federal job)? No numbers to be found at or

Intriguingly, the top 10 google results for 'GMU veteran graduation' are all fairly directly related to applying to GMU as a veteran - in a region chock full of military facilities. A comparable search for 'UVA veteran graduation' led to a top link from UVA where it was necessary to actually find the link to the fairly pitiful, with the remaining relevant links mainly clustering around around a veteran student association.

Personally, as someone born in Northern Virginia, who both graduated from and worked at GMU, I would say that GMU is continuing to do what it does best - take maximum advantage of its proximity to federal government funds. Veteran benefits may not be quite as generous as before, but they still represent a decent market segment worth capturing, especially with a commuter school model that offers a large number of evening classes to accomodate working students. (With the caveat that GMU has not changed itself radically recently. Nope - Econ 103/104 are still offered as evening classes, from 7:20pm to 10:00pm - as an example

Because really, the Commonwealth of Virginia is just a bit player compared to the towering majesty of the federal colossus. And its workforce, hungry for credentials. With an employer not only willing to pay for those possessing such credentials, but also willing to pay for its employees to acquire them. What, does anyone think it is only veterans that can take advantage of federal programs to help pay for further education of federal employees? There is an entire section of GMU devoted to disabusing you of the notion that a federal employee is on their own when acquiring college credit. That was not a joke about taking advantage of GMU's proximity to the federal government, it is a business plan.

"....a disproportionate chunk of GMU students are likely to get jobs in the relatively high-paying Washington, D.C. area.”

I think you could have just said this full stop. Most likely a disproportionate chunk of GMU grads also enters government.

And government isn't really concerned with the value of its workers' marginal product.

They also overpay for degrees. This is just a fact as evidenced by degrees being formally required for various jobs, bonuses and bumps in pay. It is not as common for private employers to name a degree requirement for a non-entry position.

"A major reason for the wage discrepancy is simply that a disproportionate chunk of GMU students are likely to get jobs in the relatively high-paying Washington, D.C. area."

In other words, this is the networking model at work. Especially when you consider that, given the region, a disproportionate chunk of that disproportionate chunk are getting employment with the Federal Government or quasi-government entities where the relationship between worker productivity and salary is fuzzy at best. I think this result says more about "The Parasite that is Washington D.C." than any model of education value.

That's what immediately jumped out at me, too. Because of the "DC labor market"? Whatever that means, it makes highly skeptical that the jobs have anything to do with productivity, marginal or otherwise.

What does the pipeline to jobs suckling the teat of Leviathan say about GMU's free-market pretensions?

Good opportunity to post a quote that comes to mind every time I visit here:

"Reminder: Tyler Cowen believes in the free market so much, he works entirely in a government protected and subsidized sector that exploits people who are coerced into seeking credentials from it."

To be fair, though, it is consistent with free-market economics to believe that big government crowds everything else out and therefore given that we have big government it's probably the best deal around. It's approximately equivalent to ultra-rich people who want to raise taxes but still engage in maximum tax avoidance for themselves.

Not really where I was going. We need good economists, if only to keep the less-good economists from running amok. We just don't need a lot of them.

Tyler is a good economist.

His school should be teaching kids things they can use in the real world, not in the beltway. In my experience, a good understanding of economics is one of those things. Go forth, young GMUs. Fly!

Let's be honest here. UVA is much better of a school than GMU, JMU, VT, etc. You just can't get a better education elsewhere in VA. This post selected some discrete data points that don't summarize the huge value UVA provides; nor does it take in to account the fact that employers yearn for these UVA graduates over similarly educated ones from GMU. Not to mention the Economics department at UVA is miles ahead of the GMU department. It's just that none of the professors take to the blogosphere to tout their recently found links.

UVA is well known among parents more concerned about academics as the well-off-high-achieving-NoVA-kids party school, with inflated grades. UVA benefits from having many alumni in NoVA who are all more concerned about their children getting into the "right school" (which for public schools in VA is UVA) and signalling their superiority. Yes, UVA can be a good school, but it is not clearly the best.

Not sure what the data says, but I wonder how many TJHSST grads pick UVA over VA Tech, MIT, Rice, etc., etc.

GMU is a mix of commuter and residential students, older and "traditional" students. My child chose a different school, but he had friends from out of state for whom GMU was their first choice (don't recall the major areas of interest, but I do remember being surprised it wasn't econ). If econ had been my child's primary area of interest GMU would have been his first choice of VA schools.

Fortunately, kids in VA have a wealth of good public colleges to attend if they are so inclined, and VA has great transfer understandings / agreements between its community colleges and many 4 year schools.

GMU, VATech, W&M, Mary Washington, NoVA Community College and many others are all great schools for many kids - especially if a kid can graduate with no debt. VA has a high number of good colleges for a relatively small commonwealth.


Many more TJ kids go to UVA than VT in spite of the fact that VT has a much better engineering program. It's like 150 kids go to UVA each year and 25 to VT, something like that. At least 15 years ago. And the VT kids tend to be a little less higher achieving on average. That said, I went to VT for engineering and did not apply to UVA and I certainly do not regret it.

I'm not sure of the reasons why- I think most people who go to UVA are not in engineering. Maybe most of the heavy STEM students go to better schools, although they are heavily discriminated against at the top schools (most good schools have quotas- e.g. no more than 10 TJ kids).

"You just can’t get a better education elsewhere in VA." B.S. You can get a great education at any halfway decent four-year institution if you are willing to work and visit your professors. Now if you need to be spoon fed, well, sure.

All of these assertions could be true, but they are pretty meaningless without evidence.

Graduates from the Jefferson College of Health Sciences out-earn both GMU and UVa graduates, and by a healthy margin at that (Tyler forgot to mention that the difference between UVa and GMU grads was less than $2000/yr). Gosh, do you think the mix of majors and the types of careers students tend to choose might vary between schools? Might that have something to do with the difference in earnings the first year out of school? There is an old saying about "a little knowledge" that seems apposite here.


The top six schools in the US for starting salary include the three service academies, MIT, CalTech, and Harvey Mudd.

No way...degrees and years of education are all the same. The government versus private sector pay researchers told me so.

This seems obviously true to anyone who actually reads the AIR study linked in the post without trying to shoehorn its findings into some preferred theory. At more venerable institutions like the private and public Ivies a much greater proportion of students choose more traditional majors in the liberal arts, humanities, social sciences and general sciences than do students at large, newer public universities like GMU where the students tend to choose majors where the answer to the "what are you going to do with that?" question tends to be self-evident. I read some stats on this years ago (wish I remembered where) and the differences can be stark--as many as one-third of students in the Ivy League major in liberal arts and humanities but just about 10 percent do so at large state universities and fewer than 5 percent at commuter-type state schools. Where UVa and GMU fit on this continuum is open to interpretation but I doubt anyone would put them in the same position. As the AIR study indicates, more general fields have lower starting salaries than job training oriented majors like nursing. Depending on proportional differences in major selection between the two schools, the fact that UVa starting salaries are less than $2000 lower than GMU may actually be kind of impressive.

Surely significantly more students from UVA go to medical, law, or graduate school than GMU. This study would seem to not count these students. It's striking that nursing is the highest paid major and biology is the lowest. Surely this is because all of the best biology students go to medical or graduate school. I'd be curious to see the same stat for salary 10 or 20 years out of college. I don't think nursing is still the highest paid major. Analogously, would the same survey design (income after one year) suggest that the most successful high school students are ones that don't go to college but instead go into construction?

It's also because a BS in biology is useless compared to the immediate monetizability of some 6 month or something certificate in nursing.

(which, of course is related to why grad/med school is a requirement)

Indeed. I believe it would. Nice point. I'd also like to point out that Economics and Accounting majors (in various econometric studies) make more money cumulatively then do others. Economics majors also have a new law-mandated position in VA. Every student must learn elementary economics before they graduate.

Yes. This occured to me also. If a significantly higher percentage of top UVA than GMU graduates go on to graduate or professional schools then the results will be skewed in GMU's favor.

This is absolutely correct. That's why nursing schools or schools with large nursing programs have the highest paid (initially) graduates. One good example is the University of the Sciences in Philadelphia (its a nursing/pharmacy school) which is always near the top of such lists.

If a large reason for the disparity in earnings is because of the DC labor market, it might be interesting to do a comparison of the salaries for the students who gets jobs in DC metro.

Also what about grad school enrollment?

This is one datapoint. I bet if you compare GMUs in other states to their UVAs, the result is not the same.

A possibility is the tournament model. As an engineer my first-year salary blew away many but it stagnates relatively quickly if you stay in your role. Other fields don't even offer the ability to stay in your role because of the supply of fresh, cheap grads. So, what is their average, median, and top quartile later in life?

I get the feeling that a large part of economics is determining whether or not you can treat something as an aggregate and then what? Let's not forget the first question, especially when it is convenient.

Also, UVA's McIntire school of commerce is being forgotten in this discussion. He mentioned Jones, but UVA continuously pumps out people from its undergraduate business school with graduates making upwards of 60k with a >95% job placement rate(albeit with a 5k tuition differential charge). Perhaps the college of arts and sciences should adopt some of McIntire's techniques for job placement.

The signaling model of education is still alive and well. If someone hears the UVA name drop, over a similarly majored, qualified, involved, etc. student from GMU, the UVA student is going to get the job.

Just a note, VT outearns both schools. Both initially and mid-career:

Also, VT is much better at football.

So what I'm hearing is this is another Tyler post were he sees something he thinks confirms a pre-existing bias of his and then tries to push it without really thinking it through.

Otherwise known as libertarianism in general.

To be fair, Tyler's opponents in the signaling vs human capital debate are mostly other libertarians.

It is basically ONLY libertarians talking about whether the eye of the needle we force everyone through has any basis whatsoever. So, it is exactly the opposite of asdf's problem.

Have you controlled for full v part time but currently working student mix? During a recession if you are currently working in a full time job but going to school part time you may be better off than a full time student.

Well, GMU has about 20,000 full time and 13,000 part time students - and the average student age is 26. Of course, many of those part time students will become graduates of GMU. The numbers are interesting, when you look at the idea of the people taking longer being part time, and often already employed, students -

'Students Graduating Within 4 Years - 40.7%

Students Graduating Within 5 Years - 58.8%

Students Graduating Within 6 Years - 64.0%'

Yep, I think there are any number of differences between a typical UVA and a typical GMU graduate - and most of those differences would tend to tilt in favor of GMU graduates earning a couple of thousand dollars more upon getting their diploma.

That's a far reaching conclusion to draw from a rather narrow data set. Employers think GMU performs better than UVA because of the entry level salaries offered to graduates from either colleges? Never mind that it may not be for the same jobs, that the student composition is not the same, that there's no mention of who's going to grad school instead of applying for jobs, etc... There's a whole list of factors not accounted for, one that is mention and then ignored, that GMU students take advantage of their proximity to government.
The line from the data to the conclusions is not exactly solid, but I get why it was drawn. It's just a shot by a GMU professor at UVA, tongue in cheek. I hope. This is graduation season...

As they say in real estate: location, location, location.

Lets move this from D.C. to, say, St. Louis. Imagine you have two equally qualified candidates for a position that has a decent learning curve.

Applicant 1 is from St. Louis and has family there.
Applicant 2 is from San Diego, and has no family ties to the St. Louis area.

Applicant 1 is clearly a better bet. Applicant 2 is more likely to go through the learning curve, and then move back closer to home.

An employer's bias towards local people is completely understandable.

“A major reason for the wage discrepancy is simply that a disproportionate chunk of GMU students are likely to get jobs in the relatively high-paying Washington, D.C. area.”

Agree, agree, agree. I am an employer in New Hampshire. I have had many nominally qualified candidates claim that they can handle the winters because they come here to ski. But living somewhere is not the same as vacationing there.

Similarly it is well known in finance that one of the qualifications you need to have to get a NYC-based finance job is an NYC address.

I haven't seen the subject of internships raised. Quite a bit of Mason faculty have various connections with Northern Virginia employers. I work in IT at Mason and have written numerous recommendations for students seeking internships as well as other employment. I have nothing but anecdotal evidence, but have noted over the years that students who intern have a greater likelihood of landing paid gigs due to their connections. I would not be surprised if internship connections therefor had a role in this.

Or to put it older terms - the good ole boy network centered around Richmond has now completely lost out to its formerly sleepy northern neighbor.

Though Loser's Lane does have bigger statues of local heroes.

I don't follow. If the wage disparity is driven primarily by the geographic choices of graduates, how does this observation tell us anything at all about signalling value and employer learning? Doesn't it tell just tell us that more GMU grads are choosing to work in an area with higher than average wages?

One can find SCHEV's latest estimates of median wages of recent graduates by institution and program here

The median for UVA econ grads was 42,121 and the median for GMU econ grads was 37,861. What should we infer from this?

By the way, it would probably be a good idea to look at how the information was collected before coming to any conclusions.

Employer judgments of expected marginal products are not dominated by credentials,

Not demonstrated by the case above. What is demonstrated is that employer judgment of credentials is not purely based on tightness of admission screening. So despite UVa having tighter admissions criteria, employers have learned that GMU *grads* have a higher expected marginal product than UVa grads.

Assuming of course that the discrepancy isn't entirely due to the geographic dispersion of GMU vs UVa grads, or different mix of degree subjects - does GMU graduate proportionally more STEM grads? Or more economists relative to historians among its LASS grads?

I didn't read through all the comments and somebody may have already pointed this out, but I was curious about two things. GMU has a higher proportion of both older and foreign students. The older people who attend gmu have access to a a better job market (i.e they may already have a job and experience). Also, if students come here on an H1B1 visa can't get a job and go back to their home country are they included in the survey data for GMU or is there a different procedure all together. If these two things are correct it might give the wrong impression of a high school student choosing what school to go to.

I think that this is a classic case of carrying a modeling assumption a bit too far. I'm a bit surprised to see this here.

As a modeling assumption, I have no argument with the proposition that wages are a function of marginal product *all*else*equal*. It's just that in the world all else is never equal. A rational employee is not interested primarily in the absolute wage level. The better measure of compensation is the employee's net utility. To make a meaningful comparison from the employee's perspective, the wages must be adjusted for the local cost of living (including any net non-pecuniary cost or disutility from crime, traffic, etc.), as well as risk, expectations for future growth etc. Organizations that choose to locate employees in high-cost areas must compensate employees for these additional costs. Presumably, the reasoning is that the marginal product of those employees is increased by the organization's choice of location.If not, the organization would relocated those employees to a lower cost location.

The GMU grads working in DC may indeed have a higher average marginal product than the UVA grads working in other geographies despite the UVA grads possessing on average superior human capital, because the location in which one works can have a major impact on one's marginal product but is not a part of ones human capital.

A simpler explanation is that "exclusivity of admission" is not the only dimension along which university credentials differ. A GMU degree may be perceived as "better" than a UVa degree by employers for any number of reasons.

What you're saying may be true but even so it would only just barely be considered evidence against signalling, and only weak evidence at that.

The high cost of living in Northern Virginia compared to the rest of the state is surely a factor for some UVA grads. If we adjusted for living costs, UVA grads might come out ahead of us GMU grads. The costs of living in NoVA extend far beyond housing and services -- some of my friend from Central Virginia are so appalled by the traffic up here that they hate to even visit me. (At least that's their excuse.)

This paragraph provides clear idea for the new visitors of blogging, that truly how to do blogging and site-building.

Currently it appears like Wordpress is the preferred blogging platform available right now. (from what I've read) Is that what you're using on your blog?

Comments for this post are closed