View Full Version : Question to current Econ PhDs

wetslope

11-13-2010, 06:39 PM

Hi, I am a graduate student in a non-econ social science program interested in using mathematical models in my research -- not statistical or econometrics related. I was wondering which out of the following graduate level econ sequence would be the best option to pursue in this regard:

a) Math for Economists (Two semester sequence)

b) Graduate Microeconomics (also Two semester sequence)

or are there any others courses/sequences you would recommended?

Thanks in advance.

michaelmas

11-13-2010, 07:02 PM

What about a Calculus sequence at your math department?

enginecon

11-13-2010, 07:15 PM

Hi, I am a graduate student in a non-econ social science program interested in using mathematical models in my research... which out of the following graduate level econ sequence would be the best option to pursue in this regard:

a) Math for Economists (Two semester sequence)

b) Graduate Microeconomics (also Two semester sequence)

It all depends on your present math and econ preparation... You'd probably need much of the equivalent of (a) to be able to absorb well (b)... You may already know much of (a) from previous courses not necessarily aimed at econ... Even if you have (the equivalent of) (a), (b) might be tough (although not impossible) with zero econ background... you may consider an "intermediate micro" sequence prior to (b).... sometimes the business school teaches a course at the intermediate micro level (accepted for grad credit possibly for MBA students) which could be interesting as a basic prep for (b), even if you have the necessary math already.

wetslope

11-13-2010, 07:39 PM

It all depends on your present math and econ preparation... You'd probably need much of the equivalent of (a) to be able to absorb well (b)... You may already know much of (a) from previous courses not necessarily aimed at econ... Even if you have (the equivalent of) (a), (b) might be tough (although not impossible) with zero econ background... you may consider an "intermediate micro" sequence prior to (b).... sometimes the business school teaches a course at the intermediate micro level (accepted for grad credit possibly for MBA students) which could be interesting as a basic prep for (b), even if you have the necessary math already.

Sorry, I should have added more information about my previous math/econ training. Here are the courses I completed as an undergrad:

Math: Calculus (3 semesters including multi-var), Linear Algebra, Intro to Proofs, Real Analysis (1 sem), Mathematical Statistics, Differential Equations, Basic Probability

Econ: Intermediate Micro/Macro, Intro to econometrics

Based on these, what would you recommend? And the econ dept. I am thinking about taking these classes is a Top-10 econ grad program, would the amount of math I have be enough?

enginecon

11-13-2010, 08:31 PM

If you are comfortable with the material in the listed undergrad courses, and assuming they were taught at a challenging level, you can take grad micro without concerns... You could also take both (a) and (b) concurrently if you have the time, but it is not strictly necessary... For extra confirmation, you could look at the excellent math appendix of http://www.amazon.de/Microeconomic-Theory-Andreu-Mas-Colell/dp/0195073401

If you are comfortable with the appendix, you should be fine for grad micro... (for other stuff you may need/want measure-theoretic probability).

Depending on your social science, you may want to look into so-called "behavioral economics" (which relates to psychology) and/or "economic sociology" ( e.g. Economic Sociology Program (ESP) - MIT Sloan PhD Program (http://mitsloan.mit.edu/phd/esp.php) ). Of course, political science departments may offer relevant courses also.

wetslope

11-13-2010, 10:36 PM

If you are comfortable with the material in the listed undergrad courses, and assuming they were taught at a challenging level, you can take grad micro without concerns... You could also take both (a) and (b) concurrently if you have the time, but it is not strictly necessary... For extra confirmation, you could look at the excellent math appendix of hhttp://www.amazon.de/Microeconomic-Theory-Andreu-Mas-Colell/dp/0195073401

If you are comfortable with the appendix, you should be fine for grad micro... (for other stuff you may need/want measure-theoretic probability).

Depending on your social science, you may want to look into so-called "behavioral economics" (which relates to psychology) and/or "economic sociology" ( e.g. Economic Sociology Program (ESP) - MIT Sloan PhD Program (http://mitsloan.mit.edu/phd/esp.php) ). Of course, political science departments may offer relevant courses also.

Given my math background and the choice between (a) and (b) which one would you recommend based on its value to mathematical modelling for research purposes?

enginecon

11-13-2010, 11:57 PM

Given my math background and the choice between (a) and (b) which one would you recommend based on its value to mathematical modelling for research purposes?

I am not 100% sure what they mean by Math for Economists in your specific school... my understanding is that typically it is just a collection of math topics which have been found useful in economics research. But it is essentially a math course (with occasional examples and motivation geared toward econ).

Depending on the specific application you have in mind, Math for Econ could be useful, or a total waste of time...

Grad micro is more likely to be useful to other social scientists, imho... not so sure about the part about the theory of the firm, but the chapters about utility theory/consumer behavior, game theory, and possibly, social choice/welfare could be quite useful to anyone studying issues related to human behavior.

dinosaurus

11-14-2010, 01:08 AM

Grad micro will give you insight about economic models, if you want your models to have some (possibly small) economic flavour then micro would be a good fit. Otherwise, could you elaborate what you mean by "mathematical" models? And why you feel that your current (quite substantial) background is not enough?

wetslope

11-14-2010, 02:31 AM

Grad micro will give you insight about economic models, if you want your models to have some (possibly small) economic flavour then micro would be a good fit. Otherwise, could you elaborate what you mean by "mathematical" models? And why you feel that your current (quite substantial) background is not enough?

I meant "mathematical model" in the sense of showing mathematically some relationship between different social phenomenas before testing the relationship (if possible) through empirical means.

AREStudentHopeful

11-14-2010, 02:46 AM

I meant "mathematical model" in the sense of showing mathematically some relationship between different social phenomenas before testing the relationship (if possible) through empirical means.

Based on your description, is an econometrics course possible? Generally, testing for relationships, empirical data tests, ect., all fall in the realm of econometrics.

wetslope

11-14-2010, 03:30 AM

Based on your description, is an econometrics course possible? Generally, testing for relationships, empirical data tests, ect., all fall in the realm of econometrics.

Yes, I have already decided to take econometrics -- which was the reason I mentioned in the OP that I was not asking about it. I am not sure about the modelling part, as in which courses/sequences cover those aspects. As an example of what I mean by mathematical modelling:

Theorizing about conflict - By Jack Hirshliefer

http://www.econ.ucla.edu/workingpapers/wp727.pdf

AREStudentHopeful

11-14-2010, 03:33 AM

Okay; I was just making sure that you would be testing using econometric methods for the type of analysis you describe. That being said and looking at that paper, I would go with the micro sequence. You should be well-enough prepared given your math background; it should serve you best.

jeeves0923

11-14-2010, 01:25 PM

I don't know about other places, but the two micro sequences I have taken didn't really teach me to write models so much as they taught me the classical basics of economics (consumer theory, producer theory, game theory, general equilibrium, expected utility, etc). If you actually want to learn to write models, a course in Organizational Economics or Contract Theory or Political Economy might be better suited, though those usually have the micro sequence as a prereq.

enginecon

11-14-2010, 03:21 PM

I am not sure about the modelling part, as in which courses/sequences cover those aspects. As an example of what I mean by mathematical modelling:

Theorizing about conflict - By Jack Hirshliefer...

Micro could be a good foundation for modelling, and some simple models of interesting situations (e.g. the Robinson Crusoe economy) are introduced ocassionally as examples or problems. However, it is not a modelling course per se.

Given the example you have cited, if you have access to a Mathematical Sociology and/or Economic Sociology course you may find them deserving of attention.

For example see, research on Mathematical Sociology at Cornell (http://www.soc.cornell.edu/research/mathematical_sociology.html)

Likewise, you may perhaps find interesting/relevant Univ. of Chicago Prof.

Steven D. Levitt - Freakonomics (http://pricetheory.uchicago.edu/levitt/Freakonomics.html)

wetslope

11-14-2010, 03:57 PM

Micro could be a good foundation for modelling, and some simple models of interesting situations (e.g. the Robinson Crusoe economy) are introduced ocassionally as examples or problems. However, it is not a modelling course per se.

Given the example you have cited, if you have access to a Mathematical Sociology and/or Economic Sociology course you may find them deserving of attention.

For example see, research on Mathematical Sociology at Cornell (http://www.soc.cornell.edu/research/mathematical_sociology.html)

Likewise, you may perhaps find interesting/relevant Univ. of Chicago Prof.

Steven D. Levitt - Freakonomics (http://pricetheory.uchicago.edu/levitt/Freakonomics.html)

The course at Cornell seems to be very close to what I am looking for but unfortunately I do not have the option of commuting to Ithaca --not to mention Chicago-- to take this course. Are there other Northeastern schools you are aware of that might be doing similar courses?

enginecon

11-14-2010, 08:21 PM

The course at Cornell seems to be very close to what I am looking for but unfortunately I do not have the option of commuting to Ithaca --not to mention Chicago-- to take this course. Are there other Northeastern schools you are aware of that might be doing similar courses?

I am not specifically aware of any similar courses in the NE; however, I was under the impression that most strong Sociology departments would offer something similar (perhaps under a different name?)... it is not really my field.

You can however easily get the Freakonomics books... they are oriented to a "general audience", thus, you would have no problem at all following the arguments... at least you'd likely find them extremely provocative and interesting.... Not that the books get deeply into math modeling, but they at least provide the conceptual model to analyze certain social phenomena from an econ POV... At amazon it seems the price range is $7 to $22 depending on the edition... definitely worth it assuming you cannot get them at your libraries....

asquare

11-14-2010, 08:51 PM

I am not specifically aware of any similar courses in the NE; however, I was under the impression that most strong Sociology departments would offer something similar (perhaps under a different name?)... it is not really my field.

You can however easily get the Freakonomics books... they are oriented to a "general audience", thus, you would have no problem at all following the arguments... at least you'd likely find them extremely provocative and interesting.... Not that the books get deeply into math modeling, but they at least provide the conceptual model to analyze certain social phenomena from an econ POV... At amazon it seems the price range is $7 to $22 depending on the edition... definitely worth it assuming you cannot get them at your libraries....

The Freakonomics approach to research is about as far away from modeling as is possible. It's about statistical identification -- finding a source of exogenous variation that provides plausibly exogenous variation in an underlying policy or "treatment" of interest. It's a reduced-form approach to research. It is one popular approach to empirical research, but it is definitely not a good source for someone who wants to learn about structural modeling. It is one component of the techniques you'd learn in a PhD econometrics class (or, at least, in an applied econometrics class), but not at all what you'd learn in the micro core or in IO, labor, or PF field classes. But there's a difference between economic (or political, or sociological) models, and econometric models. Crudely, the former characterize the behavior of agents, while the latter enable you to estimate parameters of those structural models or of reduced-form relationships using data.

The paper the OP linked to sets up the sort of model that would be covered under game theory in a first year micro sequence. If that's the kind of work you are interested in, I agree that first year micro is probably appropriate, though there will be lots of material covered that you may find irrelevant (imperfect information models, for example). I think most strong quantitative political science departments also teach graduate game theory classes that might cover similar material for a non-economics audience, and perhaps that would be better targeted for you. You've taken all of the math classes that are recommended for new econ PhD students, so you should be ok on that front. Do keep in mind that people who aren't in the econ PhD program often have a bit of a harder time with econ classes, because they aren't in study groups with the econ students and because there are positive externalities from other classes the econ students are taking.

(Here's a class (http://www.columbia.edu/cu/polisci/course-unify/2008-2009/POLS6802/index.html) in Columbia's political science department that seems designed to teach modeling. And, for contrast, here is a class (http://www.columbia.edu/cu/polisci/course-unify/2008-2009/POLS4291/index.html) that is about statistics and estimation.)

enginecon

11-15-2010, 12:07 AM

The Freakonomics approach to research... is one popular approach to empirical research, but it is definitely not a good source for someone who wants to learn about structural modeling. It is one component of the techniques you'd learn in a PhD econometrics class (or, at least, in an applied econometrics class), but not at all what you'd learn in the micro core or in IO, labor, or PF field classes. But there's a difference between economic (or political, or sociological) models, and econometric models.

Modelling means different things to different people... I do not necessarily disagree with much of what you wrote... I already suggested micro to the OP, and mentioned utility/GT, social choice/welfare, etc as foundational concepts for further work... The OP also indicated an interest in the econometric approach... in the end, I suggest all three: micro, econometrics, and the freakonomics books... since they are very inexpensive both in terms of money and the amount of effort needed to absorb that level of material, I do not see how they could hurt.

dreck

11-15-2010, 01:12 AM

Modelling means different things to different people...

I'm not so sure that it does. asquare's two definitions--one about the behavior of agents and the other about estimation of parameters of the first kind of model--cover pretty much everything. I will resist the urge to derail the thread by continuing to debate Freakonomics; to be brief I think it's great for getting people interested in economics, but hardly a good tool for training researchers.

OP: I would gather some information from a few faculty members or staff advisors in different departments (econ, sociology, poli sci). You seem to have a pretty clear idea of the kind of training your after, and could probably find the best fit--if a good fit exists--after a few short conversations. You might also look at a course in the mathematics department on modelling, though the topical coverage of these courses is erratic, so you would again want to talk to the professor first.

kevinp123

11-15-2010, 02:11 AM

Why are those two classes your only options and why do they have to be through the econ department? If you only want to learn to construct mathematical models, you are probably best off taking a course in mathematical modeling through the math department.

wetslope

11-15-2010, 05:39 AM

Why are those two classes your only options and why do they have to be through the econ department? If you only want to learn to construct mathematical models, you are probably best off taking a course in mathematical modeling through the math department.

They do not have to be from econ department. The only reason I listed them was because I notice extensive modelling in quite a few econ papers, so I naturally assumed that the econ department had such training available. Not to mention that being in the social sciences, it only seemed appropriate to take a class in another social science department rather than at the math department --where things can get pretty abstract at times. But I do not seem to find a "Modelling" course per se in the math department, if you have suggestions please do let me know.

kevinp123

11-15-2010, 04:40 PM

Oh, at my school there is a course entitled "Mathematical Modeling" that details how to construct mathematical models of reality, but the topics generally vary according to the professor. Dover has published a couple of cheap books on mathematical modeling (Amazon.com: An Introduction to Mathematical Modeling (9780486411804): Edward A. Bender: Books: Reviews, Prices & more (http://www.amazon.com/Introduction-Mathematical-Modeling-Edward-Bender/dp/048641180X/ref=sr_1_2?ie=UTF8&qid=1289838981&sr=8-2)) that would probably be useful to you, although I have never read it so I cannot attest to it myself.

wetslope

11-24-2010, 02:11 AM

thanks for all the responses guys. i have one more question: as i mentioned earlier, i also plan to take econometrics and my question is would it better to take a class using the Greene textbook (applied) versus one using the Hayashi textbook (theoritical). I am not planning to work on econometric methods per se, meaning I have no intention to do econometric theory once I graduate, but having said that I do want a solid theoretical foundation on econometrics and not just be able to apply it. Would the Greene approach be good enough for my requirements or do you recommend a Hayashi based course? And which math coureses are the most extensively used in phd level econometrics courses? Thanks.

dk801

11-24-2010, 05:35 AM

I think Greene will be fine for applied statistics.

In terms of modeling - you might want to consider looking at some agent based modeling. I think people in polisci are more open to it than economists. (Just my perception)

For example:

Complex Systems Summer Schools | Santa Fe Institute (http://www.santafe.edu/education/schools/complex-systems-summer-schools/)

New England Complex Systems Institute (http://www.necsi.edu/)

On-Line Guide for Newcomers to ABM (Axelrod and Tesfatsion) (http://www2.econ.iastate.edu/tesfatsi/abmread.htm)

Robert Axelrod's Home Page (http://www-personal.umich.edu/~axe/)

wetslope

11-24-2010, 04:19 PM

thanks for you advice. any more suggestions out there for which of these classes i should go for?