Bryan Caplan  

The GMU Difference

PRINT
TANSTAAFL vs. Futurism... Applied Ignorance...
How are GMU economists different from normal economists?  Many ways, but here's one that's struck me lately:

Them: Impossibly high intellectual standards for a handful of high-status research questions - and embarrassingly low intellectual standards for all other questions.

Us: Reasonably high intellectual standards for everything we say and think.

Random application: Many elite economists scorn OLS econometrics, but see no shame in publicly defending politicians who couldn't explain the difference between a mean and a median.


Comments and Sharing





COMMENTS (17 to date)
razib writes:

somewhat self-aggrandizing, but i think you have a good case.

rapscallion writes:

But econ only gets the respect it does because of the pretense that it's a science. To maintain that pretense, it has to look really, really hard. If you make your arguments and findings comprehensible to non-specialists, you're kind of letting the cat out of the bag.

Daniel writes:

No, it's just plain self-aggrandizing.

Please name the horde of economic issues that GMU grads have vastly higher intellectual standards on than Harvard/Chicago/whatever grads do.

You should have stuck with something a bit closer to reality like: GMU students (may) have a better appreciation for public choice economics.

Nor is your one example even slightly revealing. The fact is, in many situations, there are better methods than OLS, whereas in politics, you often choose the best of a bad lot. If better politicians could be conjured up as easily as FGLS models out of a textbook, I'm sure the (unnamed) economists you mention would do so.

razib writes:

Please name the horde of economic issues that GMU grads have vastly higher intellectual standards on than Harvard/Chicago/whatever grads do.

why are you asking a question about a topic which bryan didn't address? you should read the post before you comment.

mtraven writes:

How about the fact that they all seem to be sucking on the teat of the Koch Brothers? That's kind of different.

Cole writes:

@mtraven

The salaries for GMU professors are publicly available. They earn quite a hefty chunk from the university already (about 100k). It would be a surprise to me if the share of Mercatus funds outweighed their state university salaries.

Plus the website you linked to is silly. It uses one of Bryan Caplan's pieces as evidence of criticism against the whole GMU econ department. I really don't understand the depth of stupidity it would take to make a post about GMU's supposedly corrupt econ program on a post by Bryan Caplan, using Caplan's critique of Austrian economics as evidence.

You would think that if they were all just a bunch of corrupt corporate whores they could at least get their act together and present a consistent story rather then disagreeing with each other all of the time.

Sam writes:

why are you asking a question about a topic which bryan didn't address? you should read the post before you comment.

Maybe your reading comprehension is so superhuman that you can read not only what Bryan has written but he was actually thinking. Bryan did not provide individual names, institutions or topics. How is Daniel supposedly off-topic or not addressing what was written? Please...

RGV writes:

GMU doesn't matter when it comes to academic citations. It clearly punches above its weight in page rank. That to me is the key difference.

Student writes:

I'm an undergrad in economics. So far, I've ONLY been taught OLS!

If OLS is so frowned upon, why am I bothering?

jc writes:

Fwiw, others, e.g., McCloskey, have expressed similar sentiments.

It's often said that innovation comes from the periphery, as central actors are overly bound by structural norms (including modes of thought and incentive systems) that inhibit creativity and risk taking. GMU seems to be one of the brighter lights in the periphery.

JPIrving writes:

@Student

The in thing these days in microeconometrics in instrumental variable. One can always make up stories about why the right hand side variables in OLS are not 100% exogenous. So economists go through a long song and dance and use IV methods and....often find the same results as standard OLS.


You got to start somewhere though and one needs to learn OLS to use IV or most other econometric techniques.

John Hall writes:

William Gosset,
Even if OLS is not the perfect statistical method, many other methods are built upon it. Think of it like doing physics in a world with friction vs. one with friction. Much of advanced statistics is accounting for the problems of OLS and making it better. One of the best ways to learn econometrics is to just open up stata or some other package and go through the help sections.

John Thacker writes:

However, Bryan, I'm certainly embarrassed by Tyler Cowen's repeated and increasingly silly public faith in and defenses of President Obama.

Mo writes:

The in thing these days in microeconometrics in instrumental variable.

Probably an accurate statement in 2001. IV is still used (b/c it is useful in many cases) but calling it the "in" thing might mislead "Student."

I'd say field experiment, lab experiment, and RD are all much more "in" these days in the applied micro realms. You could probably think of a few others too.

mtraven writes:

Yeah that web page is not the best. Try here or here.

According to sourcewatch, the Koch have given millions to GMU directly, as well as Mercatus. Are you implying the economics professors are somehow unaware of where the money they get paid comes from? If so, it's professional incompetence, at least.

Most science journals require authors to declare funding sources and conflicts of interest. Why don't economists do the same?

Rob writes:

Bryan,

I don't think elite economists scorn OLS, but merely caution against making big inferences with it. The natural experiment approach, i.e. using IV, panel data, etc. isn't really more technical than basic OLS, but it focuses on identification so the researcher can make causal inference. I see that as progress.

Dale writes:

Student Wrote:

I'm an undergrad in economics. So far, I've ONLY been taught OLS!

If OLS is so frowned upon, why am I bothering?

1) OLS gives us the tools to teach regression analysis, which is very important. Without OLS the tools to do regressions would be beyond the scope of most undergrads and as such would make learning how to analyze empirical data beyond the scope of most undergrads.

I do not doubt that most institutions view that learning the basics of understanding the system is more important than not learning anything, or learning complicated models that the students can't comprehend.

It is likely the case that institutions feel that students who go into professional settings are more likely to be introduced to these more complicated models later than simple OLS and since they will not start out at the head of their department, will quickly be educated in the updated information

I do however, wish they told the students this. I think that this lack of disclosure (both in statistics and in economic modeling and theory in general) has caused real problems in the world. But that is an side.

2) OLS is the basis for more complicated techniques. And we need an understanding of OLS in order to have an understanding of more complicated techniques.

Specifically; in many situations the assumptions necessary for OLS to be correct fail and because of this in order to be as careful as possible, using a method which is correct under those assumptions that we believe hold is more rigorous.

Making statistical errors is a big deal because when we do, we often don't even know in which direction that we are wrong. A good way to say it is this. If our assumptions are right then our answers are correct. But if our assumptions are wrong, then our answers are not correct(note: not wrong, just not correct). Results that are not correct are almost like no results at all. Inferring from them is dangerous. Note that we can also make mistakes which we can identify and will move our results in known and predictable ways.

Understanding OLS is a requirement for understanding what is going on in more complicated examinations. And, importantly, knowing which more complicated examination to use in order to best answer the question that you have asked.

____

It should be within the material of most introductory econometrics courses (or second level stats courses) to really break down the OLS/MLS assumptions and give students the tools to understand when they don't know enough statistics (and when they do) to know if something is correct or not correct and precisely what it is saying when it is. It is also probably the case that many institutions don't do this.(but i have no clue on that level)

If you haven't had that level of statistics yet that may be the case as to why you haven't seen anything but OLS.

Comments for this entry have been closed
Return to top