Thursday, March 27, 2014

Coming into econ from physics (and other fields)



Chris House makes a very good point over at his blog: A physics background will not help you in econ as much as one might imagine. Actually, this is true not just of physics, but of all subjects, math and computer science included. But physicists are notorious for adventuring into other fields (especially given the lack of things to do in theoretical physics these days), so its understandable that House singles them out.

As a physics-undergrad-turned-econ-PhD student myself, I think I'm qualified to comment on the degree to which undergrad physics will help you in graduate econ. And the answer is: a little, but it will also hurt you a little.

A physics background will make you better prepared for the math of econ than someone who was a pure econ major in undergrad (though not better than the econ/math majors that you often see nowadays). You will be able to sleep through most of "math camp", and probably pass out of the first-year math class requirement, though you might want to take it anyway just as a refresher course (or, you know, because the prof has a funny accent). You will know what a Hamiltonian is, and a Lagrange multiplier. You will not be afraid of real analysis. But by the time you're done with your first year, everyone else in your class will be caught up in terms of math.

Then there are the downsides. Physics intuition is very different than econ intuition. Physics intuition is all about symmetry, and about finding elegant (i.e. easy) ways to solve tough-seeming systems. In econ, that rarely matters at all; the intuition is all about imagining human behavior. Having been trained on the former type of intuition in your formative years can actually make you less receptive to the latter, because it's hard to change the basic way you think and reason. And if you haven't taken a stats class, take one, because stats is not math, and the intuition is not the same. Basically, when you come to econ you're going to have to learn a whole new way of looking at the world, and your econ-undergrad classmates will be way ahead of you in this regard.

Also, physics doesn't prepare you to deal with empirical data. Physicists have lab experiments; economists rarely do. But the wonderful bounty of exogeneity that physicists enjoy can become a sort of handicap in econ - like growing up in California and suddenly having to survive in the Arctic.

Of course, like I said, all of this is true for mathematicians and computer scientists. Don't expect to be able to prove your way to good results with math badassery unless you are a hardcore game or decision theorist (in which case you never really left the math field at all). And while programming skill is just as useful in econ as anywhere, deciding what to program will be the most important thing in the end.

(Note: As people have been pointing out in the comments, a lot of computational and applied physics, applied math, and CS people are less about the "pure" theory elements of the disciplines, and more about modeling complex systems tractably. I think that skill will come in very handy in econ. But you're still going to need to build that intuition about human behavior, and that's pretty unique to economics.)

But should all this discourage you from switching to econ? No way! First of all, the list of former physics nerds who made it big in econ is long and illustrious, including such names as Joe Stiglitz, Dan McFadden, Mike Woodford, John Cochrane, and Robert Barro. People who switch disciplines may find it tough to adjust, but if they can manage it, they often bring fresh perspectives that can push research in new directions.

So don't be afraid, physicists and mathematicians and other lost nerd-souls of the world. Econ is different than what you've done before, but it's a great life path. That goes for biologists and chemists and other overworked, under-appreciated lab scientists too. Don't let Chris scare you off - he's gruff on the outside, but underneath he's a big softie.


Postscript: Chris also makes one other point - about professional physicists trying to horn in on econ's turf, or show up and "save the day". It's true that physics theory usually can't be applied in a cut-and-paste way to economic problems (option pricing theory being the famous exception). And Chris' intuition is right that group theory is unlikely to be very useful in econ (group theory is just about symmetry, and symmetries in econ tend to be few and far between, and also highly unrealistic, as Miles Kimball could tell you).

But I do appreciate the kind of external criticism that physicists (like Mark Buchanan) have applied to economics. If you allow an academic discipline to be judged only by insiders, you run into the problem of groupthink and vested interest. Smart outsiders are needed to look in from time to time and check on what other kinds of researchers are doing. Of course, it works both ways - outsiders have been instrumental in forcing the theoretical physics community to realize that string theory hasn't really gotten anywhere.

(Note: Also, I should have said, and forgot to say, that advanced math techniques often migrate into econ from other fields, and end up proving useful in the eyes of many economists. Economists did not invent functional analysis, for example, or stochastic calculus, but lots of economists use these now, so economists shouldn't be too quick to write off esoteric-sounding math techniques.)

107 comments:

  1. John Soriano6:24 PM

    Something I don't quite understand about this discussion is why physicists cite their more expansive mathematical repertoire as the primary reason they'd have new ideas to contribute to econ. Don't math PhDs have an even bigger repertoire? I've seen math PhDs with jobs in econ departments. Is it because physicists have more practice applying the math to scientific problems?

    ReplyDelete
    Replies
    1. Probably. Physics and econ are both applied math disciplines.

      Delete
  2. I think an important difference is around randomness. Physicsists, whatever they say, struggle with non ergomdic systems-in economics there are agents with volition actively looking to prove the theorist wrong. Nature is benign, and physics consequentially easy.

    ReplyDelete
    Replies
    1. Well, yes, but "easy" is not the right word for it. Humans will push until they get to the hard stuff.

      Delete
    2. Yes, I agree that a key aspect here is randomness. In physics, all of the theories of aggregate systems -- thermodynamics, statistical mechanics -- are based on the principle that the underlying system is random.

      As you say, in economics there are agents who act willfully. However, I don't agree that they are trying to prove theorists wrong. Rather, the application of statistical methods to systems which are not inherently random is inappropriate. A new theoretical mechanism needs to be created which takes into account the willful, non-random, actions of individual actors. Perhaps, some elements of complexity is appropriate here?

      Delete
  3. So true. I was an applied math major, and first year in econ grad school was pretty easy - if by easy you mean being able to solve first order conditions or loglinearize Euler equations without breaking a sweat. But it took me a while (probably longer than it should have) to realize that understanding the content (what is this equation actually supposed to mean?) is much more important than mastering the technique. And I'm still working on it :)

    If I could go back few years, I would try harder to build the intuition in that first year, although I'm not sure what would be the best way to go about that. Probably not reading dumbed-downed undergrad textbooks. Perhaps, particularly for macroeconomics, a course in history of the field (like going through Snowdon & Vane book, reading some seminal papers) would be a good way to get some context and see a bit of the big picture.

    ReplyDelete
  4. noiselull7:02 PM

    You forgot Robert Barro.

    ReplyDelete
    Replies
    1. Good point, I'll add him to the list! Any others you know of?

      Delete
    2. Kevin Denny3:13 AM

      There is, or was, a real rocket scientist ( guy who worked on Apollo) who became a time series econometrician.

      Delete
    3. Daniel McFaddon, Robert Engle...a bunch of Nobel Prize winning former engineers as well

      Delete
    4. Oh you know, just some nobody named Michael Woodford.

      Delete
    5. Awesome. Thanks, guys. You rock.

      Delete
  5. Physics intuition is very different than econ intuition

    I thought intuition was a human quality and no field has a monopoly on it.

    Applied Physics - aka Engineering - practices the science of approximate and unknown well.

    Econ and Finance are constructed on belief systems that are not easily testable, if at all. One can take lumps in Econ and Finance (like go bankrupt and begin again, or go to jail, or go directly to GO, like in Monopoly), it is simply not realistic in real sciences.

    Finally, I have concluded that Econ guys and gals are the most argumentative people on this earth!

    ReplyDelete
  6. Hi Noah,

    The following confused paragraph strikes me as exactly the reason that math and physics folks *could* make contributions to economics:

    "Postscript: Chris also makes one other point - about professional physicists trying to horn in on econ's turf, or show up and "save the day". It's true that physics theory usually can't be applied in a cut-and-paste way to economic problems (option pricing theory being the famous exception). And Chris' intuition is right that group theory is unlikely to be very useful in econ (group theory is just about symmetry, and symmetries in econ tend to be few and far between, and also highly unrealistic, as Miles Kimball could tell you)."

    I mean......that's just simply untrue, no?

    So.....Why is Chris right? Who is Miles Kimball as an authority and just how does he know that symmetries are 'few and far between' in economics? Why are these people deeper thinkers about symmetry than say economists like Irving Fisher, let alone physicists or mathematicians?


    Bewildered,

    Eric Weinstein

    ReplyDelete
    Replies
    1. Hey, Eric! Miles Kimball is my advisor and a coworker of Chris'. He teaches a class on symmetries in value functions. Symmetries are few in econ because not many quantities are invariant. But there are some, and as people invent new theories, they find more within those theories. For example, habit formation has some symmetry. But those theories are often very unrealistic (e.g. CARA utility), making the symmetries suspect.

      But if you have an idea for an overlooked symmetry in economics, let me know!

      Delete
    2. Hi Noah,

      I'd be happy to.....but I'd like, if I am not wrong, to get some support for the idea that these are just destructive prejudices which, if economics is honest should stop. You've made clear statements that are sharp should be subjected to invalidation. My goal is to invalidate the statements conclusively and not to invalidate you. I assume we both want that. If I am wrong, I'd like to know that now. What I don't want is a changing goal line.

      In physics, the groups U(1), SU(2) and Diff(X) are essential and are representative of:

      A) Abelian Lie Groups in finite dimensions.
      B) Non-Abelian Lie Groups in finite dimensions
      C) Infinite dimensional non-abelian Lie groups
      I will point out important examples of each in economics. But before I do, I want to point out what economists often get wrong about this game. The symmetries don't generally act on the messy economics. They act on where the economics live. That is a price path of a portfolio in a financial market is unlikely to have any symmetry. But the vector space in which the price path meanders is .... well a vector space. Like every vector space, It has lots of symmetry and doesn't have any less symmetry just because it can be used for economics.
      Okay. Here are three important symmetries of economics of each type.
      A) Scale in-variance. First on Irving Fisher's list of axioms for economic measurement is invariance under the group R_+^* of positive real numbers under multiplication. You may not think of it as a Lie group but that is what it is and the index of Francoise Divisia is path dependent because it is actually the holonomy of a non-trivial connection on a bundle with structure group R_+^*. And remember, while it seems simple, all of electro-magnetism is generated by a group with R_+^* as it's universal cover so such groups are not to be trifled with.
      B) GL(2,R) as the natural structure group for bilateral trade. When nation A trades with Nation B, we have B's imports as A's exports, and A's imports as B's exports. If they have a reason to trade the vectors will not be co-linear and will span a 2-dimensional space of baskets. But here their imports and exports can vary, grow/shrink, and rotate over time. The natural index for bilateral trade is therefore group valued in the group GL(2,R) just as GDP is valued in the group GL(1,R) which is a fancy name for R_+^*. But here the abstraction pays off because the analog of the Divisia index is not an exponential due to the fact that the theory is non-linear owing to the non-abelian nature of GL(2,R). In order to compute the analog you have to use Freyman Dyson's version of the perturbation series. This lead to time-ordered products in order to get the analog of Divisia's exponential.
      C) G=Diff(R_+^*) as the group of reparameterizations of the space of cardinal Utils. This is the most interesting by far. Somehow, almost nobody seems to know that at the epicenter of economic theory is a perfect principal bundle ready for gauge theory. The space of cardinal utility is acted upon by G via composition. The quotient space is ...... wait for it........ordinal utility theory. That is, welfare theory is a principal G bundle with cardinal welfare theory as the Total space of the bundle and ordinal welfare theory as the base space and G is the structure group of the (infinite dimensional) theory.
      --------
      I used math language to keep it short but there isn't supposed to be a philosophical discussion about the lack of groups. Almost any economics problem with prices, goods or welfare is about groups! The groups are just there. You can't lock them out. If economics hasn't figured out that the reparameterization of utils acts on cardinal utility to get ordinal utility because it's convinced symmetries are 'few and far between' that's just an error.

      Best,

      Eric

      Delete
    3. That's interesting. So what results can you obtain from these symmetry considerations?

      Delete
    4. To be honest, this stuff sort of fascinates me a little. But I'm completely lost when it comes to "G(2,R) as a structure" groups and the like (and calling them "Lies" is not great marketing either). I would be infinitely grateful if Eric or Noah could give just a simple example here. Say some two period model, and show how powerful these methods are.

      Eric is also the man behind the new Theory of Everything: http://www.theguardian.com/science/2013/may/23/eric-weinstein-answer-physics-problems

      So he seems quite prolific.

      Delete
    5. Hi Noah,

      You wrote "That's interesting." I appreciate that, and I'm happy to move on to your next question as soon as we settle this one. I don't know what to make of 'That's interesting' though. You made a strong statement and I made one in response by saying that your strong statement is factually incorrect. That's not personal as this should be a matter of fact rather than expertise or opinion. I'm trying to be constructive.

      I have shown you how groups of the kind that are central to physics are ubiquitous wherever welfare, goods and prices are found. I'm looking for acknowledgement that I am not, in fact, incorrect, or an argument about how I, as an outsider, have failed to understand something. If I'm wrong about those groups, I would like to know.

      But, of course, they are simply there. In mathematics or physics, we could possibly be debating their importance, but here we are somehow in economics debating their existence. I don't even know what to make of that. It's like saying 'All economists know non-white swans are exceedingly rare." while it is pointed out by an outsider that you are surrounded by a sea of black swans.


      You wrote: " If you allow an academic discipline to be judged only by insiders, you run into the problem of groupthink and vested interest. Smart outsiders are needed to look in from time to time and check on what other kinds of researchers are doing."

      Exactly. I suppose, in your language, I am trying to call groupthink on "symmetries are few and far between in economics" before we go into the implications. I've given three important examples (two of which are found everywhere in the discipline) with specificity. If I’m not wrong, can you please amend your post. Otherwise it spreads prejudice against the work of others.

      Optimistically,

      Eric

      Delete
    6. My strong statement was that group theory was unlikely to be useful in econ. Of course, that statement might be wrong, but in order for me to say "OK, I was wrong", I want to see some usefulness.

      Delete
    7. Anonymous2:39 PM

      While I find the Gauge Theory type ideas appealing (for obvious reasons), I feel that the real idea from physics that has been underappreciated is "renormalization group". I could be wrong, and perhaps these ideas have been used under other names.

      Delete
    8. Anon: Where do you think renormalization would come in handy? Maybe in theories that include death...

      Delete
    9. Anonymous4:28 PM

      Noah: (@ 3:49PM)

      Complex systems, primarily. I think Doyne Farmer et al have some work of similar flavor.

      Best way to formulate an RG flow might be through some agent based framework: say you have a large number of dimensions, and each agent looks at (the response of the agent is a function of) only a few of these. You could create a graph of how many agents are on which dimensions, and what is the sharing structure. Put in some "market impact". Then you take this graph, and run something akin to real space RG on it.

      But, a physicist would focus on the RG, and probably could discard the agent based framework once some intuition is built.

      But this probably needs more solid state physicists, rather than particle physicists. In this setting, the socially relevant question becomes: how is the job market for solid state physicists, vis-a-vis string theorists? (I do not have a clue about the answer to that question, I am neither an economist, nor a physicist of any flavour)

      Delete
    10. Anonymous4:39 PM

      The other thing I forgot to mention (but was on my mind originally) was that the idea of dressed vs bare parameters seem to capture the intuition that different agents might be measuring different things (say unemployment rate) even though they are trying to measure the same underlying thing.

      As the simplest example, say seasonal adjustments might come out different depending on which "scale" you are at. Economists would do this sort of thing, but RG would provide a "machine" you just need to turn the cranks of.

      Anyway, back to work...

      Delete
    11. Anonymous8:08 AM

      Scale in-variance. First on Irving Fisher's list of axioms for economic measurement is invariance under the group R_+^* of positive real numbers under multiplication. You may not think of it as a Lie group but that is what it is and the index of Francoise Divisia is path dependent because it is actually the holonomy of a non-trivial connection on a bundle with structure group R_+^*. And remember, while it seems simple, all of electro-magnetism is generated by a group with R_+^* as it's universal cover so such groups are not to be trifled with.

      This might be outdated but I *think* section 6.2 of this 2009 paper by Lee Smolin [1] explains this stuff in more detail. I still don't understand half of it but ok:

      http://arxiv.org/pdf/0902.4274.pdf
      http://pirsa.org/06050010/

      Delete
    12. But if you have an idea for an overlooked symmetry in economics, let me know!

      I suggest to you that properly defined "risk" is conserved - like energy. Or perhaps, risk monotonically increases - like entropy. As we introduce complexity to reallocate risk we introduce additional risk.

      It seems to me that there is a basic symmetry that a lot of economic discussion loses sight of. If one considers perturbing an economic system by plus or minus a small amount, the result should be continuous through the origin and the derivative on both sides of the origin should be the same.

      Delete
    13. This is probably why generalizations based on physics concepts are useless. What is properly defined risk? It increase monotonically with respect to what? Complexity of what? Complexity of a firm, business process, economic system?

      Delete
    14. Complexity of what? Complexity of a firm, business process, economic system?

      Yes. :-)

      Delete
  7. Eric,

    Gauge symmetry applies to economics, but hardly anyone but you recognizes the fact yet ;)

    ReplyDelete
  8. The optimal undergrad training for grad econ, probably dominating even econ itself is something like a math-history or physics-history double major. And I don't mean "optimal for grad econ history". I mean econ econ.

    ReplyDelete
  9. As a math undergrad turned econ grad student, the natural choice would have been to study micro theory. I felt right at home in my first year micro courses. But macro was just so much more appealing even if a lot of it was utterly baffling and took a long time to get used to. The most useful thing I'd suggest to other math/science people going into econ is to take some economic history courses. I had the great privilege of learning economic history from DeLong and Eichengreen and that really helped with starting to acquire the "econ intuition" that Noah refers to.

    ReplyDelete
    Replies
    1. I'm in a similar boat...for some reason I looked at micro theory and thought "If I wanted to do this I'd have stayed in physics". Weird, huh.

      Delete
    2. I think that maybe it has to do with why one started studying physics? (besides an innate desire to understand the world...) I loved physics when I could visualize the systems and how different mechanical objects were interacting, or would interact in the future. Equally with how particles/waves would swap back and forth between what was being observed or even how the general energy of a system could affect what exists. With Micro I could always "visualize" the "push-pulls" of the systems that made people behave and how those aggregated. I love just staring out the window and thinking about how, say, wealth distributions change as micro-agents age and their spending/saving patterns change or the distribution of earnings/capital changes and that affects the next generation's choices, etc.. In contrast, while I "got" macro on the mathematical level, I could never "visualize" the emergent phenomena to the extent that my classmates could and that just made me care less about it - much as reducing "particle formation" to an operator applied to an equation just made me drift away from the field.

      To each their own. One of the joys of science is talking to people who see the same world as you do and letting them open your eyes to seeing it in a new way!

      Delete
  10. Anonymous3:15 AM

    I don't think of physicists as anywhere near as bad as engineers/programmers when they blather on about some one simple rule for belly fat model of society that will deliver us all to liberatopia.

    ReplyDelete
  11. Guess I shall throw in a brief comment here. However, I suggest anyone interested go over to Chris House's thread where I made a long comment that he did not reply to, perhaps because I got a bit insulting at the end suggesting that he was a Whig who did not know what he was talking about (not very nice of me). Mark Buchanan in particular raised the issue of complexity, and several of the approaches that Chris simply dismissed fall into that area, including chaos theory and path dependence. I also noted as did Mark that what is labeled "econophysics," which neither Chris nor Noah here dealt with, spends a lot of time looking at fat-tailed time series, which happen to hold empirically in many areas, particularly in finance. A general point I would make is that either some of the topics Chris initially dismissed are either still under study, such as agent-based models, or have been partly subsumed by regular econ to some extent, such as chaos theory and path dependence.

    A central issue that keeps circling around these discussions seems to be rational expectations. Chris (and to a lesser degree, Noah) simply assert that macro models in particular simply must assume ratex, even though we know empirically that it is simply false. Now there is this strand of response that DSGE models are beginning to loosen this a bit with such methods as incorporating habit formation, and so on. Maybe, but once the door is open, why not go further and look more seriously at such things as ABMs that are clearer about what they are doing? To a substantial degree my reaction to all the DSGE-models-with-financial-frictions is similar. Why not go further and let Minsky effects be more clearly modeled? Is it because he allowed for a model of endogenously changing preferences?

    On the matter of Eric's gauge theory approach, I would note that there are some economists pursuing this, with perhaps the most important being William Barnett. One issue with it may well be that while it attempts to solve certain problems of failures of invariance, it in the end assumes in effect higher forms of invariance that many economists are uncomfortable with, although Eric might disagree with that observation. This is almost more a matter of aesthetics and lack of familiarity than anything else rather than something fundamental about gauge theory.

    Barkley Rosser

    ReplyDelete
    Replies
    1. Can you recommend a specific Barnett paper as an entry point to the economic literature using Weinstein's approach?

      Delete
    2. Hi, Barkley!

      I also noted as did Mark that what is labeled "econophysics," which neither Chris nor Noah here dealt with, spends a lot of time looking at fat-tailed time series, which happen to hold empirically in many areas, particularly in finance.

      Well, yeah. But here at Stony Brook we have a lot of applied math people who work on fat-tailed time-series, and they tell me that physics analogies have not so far been helpful in finance problems. And I have a friend who runs an econophysics group in Japan, who is not sanguine about their results so far either. And also, I'm sure you've read Cosma Shalizi's rants on the subject. So I left econophysics out intentionally.

      Chris (and to a lesser degree, Noah) simply assert that macro models in particular simply must assume ratex, even though we know empirically that it is simply false.

      I am confused by this comment. I have never asserted this, and in fact I believe rational expectations is false.

      it in the end assumes in effect higher forms of invariance that many economists are uncomfortable with

      Yeah, it seems super-idealized and frictionless-neoclassical to me.

      Delete
    3. Blog posts where I talk about Rational Expectations:

      http://noahpinionblog.blogspot.com/2013/01/the-power-and-terror-of-irrational.html

      http://noahpinionblog.blogspot.com/2012/02/why-rational-expectations-models-can-be.html

      http://noahpinionblog.blogspot.com/2011/09/steve-williamson-proves-our-universe-is.html

      http://noahpinionblog.blogspot.com/2011/10/rational-expectations-vs-heliocentrism.html

      Delete
    4. Noah,

      OK, you are innocent of naive statements about ratex, but Chris remains pretty guilty in my eyes.

      Certainly econosphysics is controversial and one can easily find papers in Physica A that are pathetically awful and unuseful. OTOH, statistical mechanics models from physics do generate kurtosis in time series pretty easily.

      Sam,

      Barnett has mostly supported Eric in blogs and tweets, but he does use variations on gauge theory for his monetary aggregation work, with his recent book on Gettting It Wrong a good place to look at that stuff.

      JBR

      Delete
  12. The augment is not too convincing and I'm detecting a field of close mindedness.

    If orderstatistic does not understand gauge theory how can he dismiss it as being useful to economics? Einstein said things should be a simple as possible, but no simpler. Maybe complicated, tedious math is the simplest way of approaching an unresolved economic question, because everything else has failed. Economists have a hard enough time making notable contributions. To say physicists face a challenge is like the pot calling the kettle black. Finally, there are many instances of novices revolutionizing fields, like John Nash and and his equilibrium theory that he was inspired to create after only a few economic courses (he was not studying to be an economist)

    ReplyDelete
    Replies
    1. Anonymous2:57 AM

      The Nash equilibrium result is overrated and the whole Nash mythology is misleading. I have been told (but can't verify) that von Neumann found the Nash equilibrium years earlier and stashed it away as unsatisfactory. In any event we know he wrote it off as trivial.

      The Nash result comes from games that aren't repeated, but the real world is all about repeated games. The proof wasn't constructive, so it wasn't especially insightful, it just used a fixed point theorem. There are some real geniuses working on the computational complexity of finding Nash equilibria (hint: non-trivial), but that's beside the point because it's not a useful model. Nash won the Bank of Sweden Prize largely, I suspect, because others seized upon his model for whatever reason.

      Delete
  13. Thanks for this Noah! A balanced view. I like the cartoons also. One short comment. You note that "Physics intuition is all about symmetry, and about finding elegant (i.e. easy) ways to solve tough-seeming systems."

    This is obviously true of classical and quantum mechanics, field theory, etc., and this is what most people think of as "physics". But this kind of symmetry-centric physics leaves out a huge part of modern physics which studies messy disordered systems, things like glasses and other amorphous materials, protein folding, soft condensed matter (gels and that kind of stuff), crystal growth out of equilibrium, random media, crackling noise, etc, where symmetries are few and far between. I actually think this is the part of physics from which we're likely to find the most inspiration for new metaphors and models for economic systems. It's a very different kind of physics from the symmetry guided field theory stuff, and uses different mathematics. One rarely finds elegant solutions to apparently complicated systems, and more frequently finds rich, complicated behavior emerging from apparently simple systems.

    ReplyDelete
    Replies
    1. That is a very good point! :-)

      Delete
    2. I added a note saying this.

      Delete
  14. Anonymous7:42 AM

    I discovered once, although I don't now remember the details, that some of Paul Samuelson's innovations came from pondering J. Willard Gibbs and thermodynamics.

    ReplyDelete
    Replies
    1. Samuelson drew a lot from physics, actually.

      Delete
  15. Phil Koop8:22 AM

    The amazing thing is not that it might be difficult for a physics major to succeed in an econ PhD program, but that it is possible at all. An econ major would not be adequate preparation for a physics PhD.

    The "outsiders" in the "theoretical physics community" are either mathematicians or experimental physicists - hardly outsiders at all. Woit has a Princeton PhD in particle theory; on the one hand, that doesn't put him very far "outside", and on the other, it's not like the string theorists are paying attention to him anyway.

    ReplyDelete
    Replies
    1. True, but really I was thinking that most of Woit's audience are outsiders, and I think that public pressure, along with the failure of the LHC to find supersymmetry, has really put pressure on the string theorists.

      Delete
    2. Actually this is a very interesting point. In fact I would go even further.

      I would say that by itself an econ undergraduate degree is not adequate preparation for a PhD program in econ.

      My own personal experience is that by the time we got to the comprehensive exams the only people left all had at least an undergraduate degree in math or similar training. There's probably very few fields where only holding an undergraduate degree in the field automatically dooms one to failure at the PhD level in the same field.

      That's not to say that an undergraduate degree in econ has no value. There are a lot of economists who skipped econ 101 entirely and their lack of intuition shows. They may do brilliant theoretical work, but say stunningly stupid things in casual conversation for example.

      Delete
    3. I would say that by itself an econ undergraduate degree is not adequate preparation for a PhD program in econ.

      It is true that the people with only an undergrad econ degree have to work hard to catch up on the math in the first year.

      Delete
    4. "An econ major would not be adequate preparation for a physics PhD"

      Actually, only true in the sense that is is not commonly observed. But you're confusing causality with correlation here and not controlling for self selection.

      Hell, I know one of those useless "art history" undergrad majors who went on to get a math PhD, studied under a Fields Medal winner, kicked all the "I-was-an-undergrad-math/physics-majors" asses in her grad cohort in terms of placement and is happily doing serious math s**t in a tenure track position at an R1 right now. It's just that... she was originally genuinely interested in art history.

      In other words, what the "preparation" here is, is just acquiring familiarity with the terminology, nomenclature, social practices and norms, the culture and taboos and the language of a particular discipline. Any smart enough person can do it, given a year or two. The whole point of formally studying a subject is to speed up that process. This set of skills actually isn't as trivial as it sounds, is probably underrated, and it's sort of the reason why we have formal education to begin with. Think of trying to become semi-functional and independent after arriving in a foreign country where you don't speak the language - you'd wish you had an "immersion" (major in ...) education before hand. But there's nothing pivotal about it. An econ major is a perfectly fine preparation for a math PhD, it's just that most people who self select into econ are a little bit dumber - along a certain dimension - than the folks who select into math PhDs. Maybe this just reflects my general doubts about education in general though (in case it's not obvious, I'm more in the "education as a signal" camp than the "education improves productivity" camp)

      The general implication is that whatever you study, study it seriously. If you're gonna do "Hotel and Restaurant Management", take it damn seriously. Ok. That's stretching it a little bit. So the narrower implication is that if - and let me capitalize that - IF you're interested in scholarly study pick a major that will challenge you to think, kick your butt, and make you work. For most people that means picking the science-y majors. For some people - as weird as it may sound on this blog - it could actually be "art history". But yeah that's an exception. Something like traditional "classic" Humanities - History, Philosophy, even... heresy of heresies... a well designed Cultural Studies major ... will do plenty to get the mind going.

      The main issue of course is that those classic Humanities majors will get you an epsilon (where epsilon is defined as the smallest number greater than zero) chance of getting a decent job after you graduate. Which is why you want to combine that Gender Studies major with something like Physics or Math, and make yourself take it (both parts) very very seriously.

      Delete
    5. Allow me to blow Peter Thiel's mind for a moment here: a good undergraduate degree will give you the critical reasoning skills and deductive/inductive abilities that can then be applied to any field. Being prepared for graduate study is then just a matter of learning field material (which shouldn't be the focus of a BA anyways).

      Delete
  16. Econ is a charlatan field. There is not a single axiom in econ that is not a fabrication. Nothing is verifiable and nothing is really testable (can not isolate anything!). It is pile from a bull or cow, infested with worms! Enjoy, Noah!

    ReplyDelete
    Replies
    1. What about discrete-choice models, auction theory, and stable matching algorithms? ;-)

      Delete
    2. They are part of the pile with bigger worms!

      If we just take the abstract math from each model/theory, it might make sense in math context. As soon as we introduce their applicability to any Econ situation, we immediately inherit unproved axioms, i.e. efficient market, or efficiency. And, when we try to apply (like applied physics or engineering) we run into dealing with ignorance and ignorants trying to protect something they have while they are scheming to take what others have. Don't believe, just look at all the laws that are designed to only favor a few who were part of the fabricating them.

      Delete
    3. Anonymous1:55 PM

      ^ A guy who thinks that anything with numbers that he doesn't understand is 'abstract math'

      Delete
    4. KV I think you're just plain wrong.

      Delete
    5. Noah,

      With all due care, I must say that saying somebody is wrong does not make it so. I have proposed that Econ is based on made up axiom and they are simply not provable. This is not the case in Physics and most hard sciences. One only needs Newton's laws to go to the Moon and come back, predictably, and can be done as often one wishes, including by the scientists and engineers from any country. Heck, we can't even figure how to issue a currency that remains stable for a few thousand years, or even a few hundred years. I can cite many examples but will ask to only consider the insurance. If the population pool is kept large, it might work; however, insurance companies want to keep only those clients that are minimal risk and dumping others out of the pool. As soon as such a selective discrimination begins, the theoretical foundation is corrupted. Then, we let loose the laws and regulations and the rest of the catastrophe. You will find the same in almost all Econ matters. The reason is very simple: all axioms are a fabrication of a human fancy. And, so are all the models, theory and what not.

      Delete
    6. KV, you should check out some of the things I listed above, which really do work. Google makes their profit using a kind of auction designed by economists.

      Delete
    7. Anonymous6:42 PM

      Oh come on Noah you well know that discrete choice theory and stable matching are rather the exception and far from central to the economics core; that's just nit-picking.

      Economics as a field is an abject empirical failure and you know it. But then you make a living off all this cognitive dissonant chicanery so of course you have to tell yourself otherwise so you can sleep at night.

      Delete
    8. They are exceptionally good, yes, but really successful theories are always exceptional.

      Lots of mundane economic theory is decently useful. Tax economics does a decent job of predicting the effects of a lot of tax changes. Finance theory can improve portfolio allocation a lot, if used carefully. Contract theory does a decent job at creating incentives. Basic Econ 101 supply-and-demand graphs are not a terrible guide to policy in many cases. Game theory has lots of successes that are not as spectacular as auction theory. Behavioral econ has done a good job getting people to sign up for pension plans. And so on.

      So there are other theoretical hits besides the famous, exceptional few.

      Delete
    9. Anonymous3:25 AM

      To be fair, you can't just cherry pick greatest hits that overlap substantially with harder fields like operations research or computer science. Economics gives some fairly good intuitions, but those aren't especially unique to economic thinking.

      And what are we to make of the harmful theories and implied policies? Rogoff's 90% threshold, the false confidence of risk management and efficient markets, the great moderation that wasn't, the wisdom of deregulation, excessive executive compensation as a principal-agent solution, the tidal wave of money floating around the world due to loose central banks, austerity programs, privatization of medicine and education and other public services, the explosion of student debt (it makes them more accountable!), incentivized healthcare in the UK, privatized prisons in the US, the whole supply side tax cut fiasco, etc., etc., etc. ?

      I expect that people follow blogs like yours because we're hoping for better from the newer economists.

      Delete
    10. Noah,

      May I remind you that you do not like that darn macro model or that Bayesian formulation as it involves belief system. As I said before, force-feeding math is not the way of Econ. Google's use of auction theory probably works because the demand for Google's ad space is so large and hence efficient and they can use for their advantage. That does not mean that auction theory is correct and applies universally.

      Delete
    11. To be fair, you can't just cherry pick greatest hits that overlap substantially with harder fields like operations research or computer science. Economics gives some fairly good intuitions, but those aren't especially unique to economic thinking.

      But why should they be unique to econ? Lots of fields overlap. The point is, the theories work.

      And what are we to make of the harmful theories and implied policies?

      What do you mean, "what are we to make" of them? Bad theories are bad theories.

      the tidal wave of money floating around the world due to loose central banks

      I'm sure that you don't really know that that's a bad thing.

      I expect that people follow blogs like yours because we're hoping for better from the newer economists.

      And you'll get it. But not because the younger generation is any smarter or more reasonable. As conditions changed and new evidence became available, new generations internalized the lessons, while older generations as always found it difficult to do so. Ideas evolve generationally.

      That does not mean that auction theory is correct and applies universally.

      Of course. No theory applies universally. Some people unfortunately tend to ignore the scope of their theories and pretend as if the theories are universal, but smart people know that this is never the case.

      Delete
  17. Anonymous9:10 AM

    an interesting post, one I've considered lately, from the view of the physicist. I think what attracts the physicist to economics is the sense that here is something that is very important to everyone, but is the framework really on solid ground. Both physicists and economists make heavy use of models. I think there is a feeling by the physicist that the economic models are seriously lacking. The problem is that the physicists have yet to be able to do any better, but the feeling remains. Probably one of the central points of discomfort from the physicist about economic models is that the models seem to lack a clear coupling to the real world of finite limits. The feedback and self-limiting mechanisms of the physical world seem to be missing. It is as I said however, the physicists do date have yet to do any better, other than arm wave.

    ReplyDelete
    Replies
    1. The thing is, I think that physicists (and almost everyone outside econ) hears mostly about macro. Actually, there are a lot of economic theories that work pretty well - random-utility discrete-choice models, auction theory, and stable matching algorithms, to name three prominent ones. If physicists heard mainly about these, they might still want to come into econ, but not to "fix" it.

      The problem is that nearly everyone hears only about macro, and macro probably doesn't have the data quality to be "fixed" by clever new theoretical approaches.

      Delete
    2. Anonymous1:30 PM

      on point again. The problem is that macro is the part that appears to some of us, as running amuck. What appear to be the large developing inequalities all over the globe look as though they are tide to macro issues. Some of it is normative to be sure, but the normative issues would probably look less so if the basic contradictions in theory were more apparent.

      Delete
    3. This comment resonates with me (and I agree with it, mostly). I do think a lot of economic models are really lacking. For example, I studies Newtonian gravity and thought "This is a really great system and explains a lot." Then I got to wave-particle duality or the concept of a "field" or of how a particle too massive to exist at usual scale for more than a fraction of a fraction of a second could be the gauge boson for a force acting on galaxies and thought "wait .. what!?" I don't doubt physics at all. I believe that the results we're getting are the best attempts by very smart people dealing with very difficult systems. I love learning more as human knowledge expands.

      But it's like that in economics too. Economists know that many of the models are seriously lacking. Much of it is due to data horizons. Much of it is due to limited experiments (there are even natural experiments showing that lab experiments are invalid :/ )

      I think all science would benefit from the idea of "Chesterton's Fence". Rather than assuming that everything that came before was wrong, approach with a spirit of humility and realize that very smart people have thought about this before. When I read a book from the 1500's, I don't think "what an idiot," I think "I might have more data than this person did." Likewise with contemporary fields, maybe I think more clearly, but maybe I don't appreciate their challenges.

      Delete
  18. Anonymous9:57 AM

    A minor pedantic nitpick - if the only skill set that someone from a CS background brings to economics (or any other field) is the ability to program a computer, then they have a rather poor CS background, insofar as CS!=programming. At the applied-math core of computer science, some of the key skill sets involve techniques for formalizing (and/or modeling) both problems and their solutions (admittedly in a manner that makes said solution readily conveyable to a computer as a set of discrete instructions). Now, perhaps by "programming" you meant to include all of the various aspects leading up to program design, rather than merely the act of translating a human-level description of a solution into an algorithm expressed in terms that an idiot savant (the computer) can rapidly step through -- but I did say that this was a pedantic comment :)

    ReplyDelete
    Replies
    1. Not a minor nitpick at all! A major nitpick and a good point. I added a note about this. :-)

      Delete
  19. Zathras11:42 AM

    "Physics intuition is very different than econ intuition. Physics intuition is all about symmetry, and about finding elegant (i.e. easy) ways to solve tough-seeming systems."

    This is true of theoretical physicists (frankly more true than is healthy for the field, which has too many pure mathematicians masquerading as physicists). However, over 80% of physicists are experimentalists. Experimental physics is very much a different animal, with different intuitions that vary from subfield to subfield. Some spend most of their time building and calibrating new instruments. Others do analysis on massive datasets. These data analysis skills would likely translate better than those of a theoretical physicist.

    ReplyDelete
    Replies
    1. Phil Koop4:01 PM

      "This imagining process is so difficult that there is a division of labor in physics: there are theoretical physicists who imagine, deduce, and guess at new laws, but do not experiment; and then there are experimental physicists who experiment, imagine, deduce, and guess." -- R. Feynman

      Delete
    2. One difference would be that the subjects in physics don't sometimes lie to their measurers.

      Delete
  20. As a physicist who is very interested in economics I'd like to offer a few comments.
    I think that what physicists bring to economics is critical thinking. Not that economists don’t possess such thinking, just that physicists are, I think, more readily willing to discard the currently acceptable line of reasoning and try something altogether new. This comes from the emphasis on first principles as the best way of reasoning about a problem. That is, what are the fundamental causes, rules, and impetuses that govern a situation? By divining these principles and then deriving relationships to fit a particular problem using them, a useful and correct result can be found. Most others, it seems to me, apply reasoning by analogy. That is, they look for a similar situation and try to fit that solution to the current problem. Physicists of course do this too, but are mindful of the underlying first principles.
    I agree that physics is easier because it is a hard science based on theory and experiment. That is, physics is based on a belief system that has been well supported over the years. That belief is: that the universe is governed by immutable principles, that we can divine these principles through experimental methods, and that we can describe these principles by constructing quantitative theories. So, by making observations and experiments, reasoning about the universe to discover principles, and then describing those principles using the machinery of mathematics, we get to a useful and repeatable understanding of the universe.
    I believe that some of the angst that economist suffer comes from comparing what they do to that of what physicist do. The desire among some economists seems to be to find quantitative theories of economics which are similar to those created in physics. Theories which produce well defined predictions of the future for well described systems. But, maybe the techniques of physics and the expected results are not the same in economics? Rational thinking is certainly as important in economics, but perhaps the quantitative methods of physics are not appropriate to economics. The math used in physics is there because it is useful. That is, the math works for the problems to which it is applied. If it didn’t it would be discarded.
    The last point “If you allow an academic discipline to be judged only by insiders, you run into the problem of groupthink and vested interest. Smart outsiders are needed to look in from time to time and check on what other kinds of researchers are doing.” is I think most important. Particularly in a field which is so influenced by politics due to the close relationship of the field to the world of policy making.

    ReplyDelete
    Replies
    1. Good points.

      But also, I think the persistent dominance of string theory and supersymmetry has tarnished this reputation a bit. In particular, after the LHC didn't find "natural" supersymmetry, the theorists seemed to just move the parameters up to higher energy level, just out of sight of the LHC.

      Delete
    2. As a former high energy theorist (well, really a phenomenologist, but I don't think anyone outside of high energy uses that word), the real problem is the combination of

      1. everyone knows the standard model breaks down somewhere, so there must be something beyond

      2. over the last two decades, people have managed to show that almost all the interesting extensions to the standard model all look alike. Adding extra gauge groups (more forces) looks exactly like adding extra (compact or large) dimensions, which looks a lot like super symmetry (the best way to think about super symmetry is probably to think about adding extra dimensions that anti-commute, as weird as that sounds).

      So the LHC hasn't seen anything other than the Higgs, BUT we know the standard model isn't the end all, be all. What do we do? Its not just SUSY that wasn't seen, nothing apart from the Higgs was, thus far. Hopefully when the LHC runs at spec, they'll see something. Otherwise, the field will probably be killed dead.

      Delete
    3. Hopefully when the LHC runs at spec, they'll see something. Otherwise, the field will probably be killed dead.

      Yeah. Seems unlikely that any countries will shell out the billions required for a bigger collider, without the strong expectation of seeing something there.

      Delete
    4. On the energy scale, LHC is so many order of magnitude below the levels the real cosmos may be working so that it is a fallacy to say if LHC did not find such and such, it is the end of that venue in the science.

      Econ is necessary evil of the society, and evilness is basically driven by the power structure - or fabricated laws of the society. If this is not the case, explain why every 20 years or so, we go through a financial crisis, screw the masses through inflation, higher rates, QEs, bailouts of corporations and even governments all over the world. All this while saying that this new guy or gal is going to save our behinds from over own follies, perpetuated through econ arguments - conservative or liberal or anything in between.

      Delete
    5. The problem is that if the LHC fails to lead to any new discovery (not just the verification of the Higgs everyone knew existed) then no one will spring for a new collider.

      There is certainly something there (see my post above), but funding for physics is already at super low levels (my entire grad school cohort has left science),and funding for particle physics is even lower. It will be tough to push funding for a new collider after a collider generation that found nothing.

      Delete
    6. Will,

      It may be that we will not be doing collider physics on the Earth. Or, we may even give up physics by smashing particles.However, there are many many stars, clusters and black holes doing super physics right now and we may turn our eyes and mind to them. Besides, LHC was oversold; energy levels are what we can do technologically now. And, did you see that 20 years cycle in Econ crisis? Same thing happens in almost all sciences.

      Delete
  21. I feel Macro goes wrong when they rely too much on deductive methods of logic to try and identify higher order relations. The laws of physics allows this to a certain extent while there are not true laws in how people act.

    Macro is inherently inductive in logical application. This is why people often reason from analogy. This seems messy because large scale experiments can't be constructed to better refine macro relationships.

    This is why I believe ACE models are the future. You can construct agents that are based on actual experimental data from psychology and sociology. Learning algorithms can be refined to better approximate reality. Scale up a model that consists of these agents and you can then start to ascertain the higher order effects and emergent properties of an actual economy.

    Models like DSGE try to be founded in micro but to scale them up the assumptions require a smoothing over of the messy realities of human decision making. This smoothing over may eliminate the exact emergent properties that may be responsible for things like bubbles and business cycles.

    I think there are physicists who work with fluid dynamics and materials that may be much more prepared for that type of macro. Things like protein folding or crystal growth are the exact type of problems that may lend itself to economic modeling and are not dragged down by theory.

    Computation gets exponentially cheaper and our abilities to construct models of sufficient complexity are only going to get easier.

    As a physics discipline I look to meteorology to drive the type of thought that would be most useful to future econ models. Weather and climate models have a closed system of limited input and output (radiative balance). They are path dependent and have very rough current data at low resolution. Initializations cause wide swings in outcome. Economic modeling is much the same.

    I could construct a simplistic model of earth that shows large patterns of equilibrium. Winds generally go this direction, storms rotate this way in this hemisphere, etc... They may have some value, but you ignore the feedback loops, emergent properties, and inherent variations within the system. The earth can't be described as trying to reach some idealized equilibrium. Why should the economy be different.

    Abandoning the concept of equilibrium is the next step in econ. It will be hard, but it is necessary.

    ReplyDelete
  22. Wonks Anonymous2:41 PM

    "outsiders have been instrumental in forcing the theoretical physics community to realize that string theory hasn't really gotten anywhere"
    I was under the impression the theoretical physics community has not recognized any such thing.

    ReplyDelete
  23. Surdas3:34 PM

    You seem to be assuming here that physics = theoretical physics. ("Physics intuition is all about symmetry, and about finding elegant (i.e. easy) ways to solve tough-seeming systems".) There are many other fields within physics that have nothing to do with this and actually yield *useful* intuitions and skills.

    For example, I have a doctorate in geophysics. Most of my work on this field focused on planetary science, which has (surprising) parallels with economics: incomplete or otherwise not very convenient datasets (e.g. macro-scale data with no corresponding micro-scale data), a need for simple models, inability to run controlled experiments (obviously), etc.

    Four years ago, I took an opportunity to switch fields - I now work in public policy for the Canadian government. To cut a long story short, I haven't done any degrees in economics (although I have taken upper-level undergrad macro and micro), but I deal with economics every day in my job. For example, I read econ papers for my policy research activities, and interact with economic modeling and modelers in other parts of my job.

    I find that my previous physics experience is actually quite helpful, and I'm much more capable of working at the economics/policy interface than most of my non-economist colleagues. Further, I would say that even non-PhD economists frequently don't have a great understanding of macro models, what the assumptions are, and whether they're appropriate to the situations for which they're being used. This is my biggest beef with macro (which you seem to share) - not that equilibrium models should never be used, but that the convenient package of assumptions (which is generally associated with neoclassical) that everyone uses is not a realistic model of economic behaviour and shouldn't be used uncritically. Also, I find that economists (in government and out) tend to rely far too much on the assumption that "the market" is achieving optimal results in a given situation (often in the face of evidence to the contrary).

    With that said, I think there is plenty that is useful in economics, particularly at the non-macro level (such as the things that you mentioned). I think that economics is a tricky subject where it's easy to get tripped up by lazy thinking, and it's very common for people not to put in the effort required. Someone with a PhD from an empirical branch of physics can actually do well and even bring new insights if he/she brings an open mind. I think that having a basis in a slightly less (but not too much less) fuzzy domain can be helpful.

    ReplyDelete
  24. Anonymous6:36 PM

    Obligatory mention of Phil Mirowski, of course. He wrote a few good histories of the influence of physics on economics. Also Wade Hands.

    We've been talking about symmetry and group theory, but a more basic issue is conservation: how can you make physical analogies or even use algebra when it's not clear what is being conserved?

    Scientists are taking a renewed interest in economics I suspect because 2008 happened. The popular impression, which I think is fair, is that economists were not doing their jobs. The flurry of "we were so!" articles only confirms this belief--the economist doth protest too much.

    So the problem space turns out to be murky. Scientists will find that out, but they will not, by and large, pretend to know things they couldn't possibly know.

    ReplyDelete
    Replies
    1. Didn't Emmy Noether prove that a symmetry implies a conservation law and vice versa? In other words, you're probably more right than you know.

      Interestingly, accounting is all about symmetries. They're a fundamental part of double entry bookkeeping and a major reason for its superiority over older systems. Unfortunately, a lot of economics at the macro and micro levels makes no accounting sense whatever. Perhaps if economists were better versed in accounting, they would pick up on more symmetries in their field.

      Delete
    2. Noether's Theorem is the main reason I'm skeptical about the usefulness of symmetry in econ...because there are so few things that are conserved outside of special cases.

      Delete
  25. Anonymous7:10 PM

    What about Tinbergen

    A guy from Holland

    ReplyDelete
    Replies
    1. Sure, but he's a bit older, so people might think that when he contributed, econ was just not sufficiently "mathed up" yet.

      Delete
  26. I don't think physics is a panacea for economics, but I do think economics is horribly inbred, and that this has hurt it. Outside methods and ways of thinking have a lot to offer. For example, economics could benefit from some signals and systems theory. Economists always seem surprised that differential equations behave like differential equations. In fact, they often argue that they don't.

    Evolutionary biology and population biology also have a lot to offer. The former as it is a historical science like cosmology or geophysics. Even a solid grounding in history and historical research would benefit the field. History may be one of the humanities, but it requires solid reasoning and sifting of evidence. Looking at population biology would also be useful as its application of mathematics to the real world seems to work so much better than whatever it is that economists have been using.

    Economics isn't the only field that has gotten awfully inbred. I've been following the computer science curriculum at Stanford - helping out a BS candidate - and I've been rather surprised at the lack of progress. I graduated 40 years ago, and it seems that most of the interim has been spent forgetting, rather than moving forward. In their defense, their formula does seem to please the local venture capitalists.

    ReplyDelete
    Replies
    1. Anonymous2:36 AM

      Undergrad anything is pretty much a joke anywhere. Just speaking from experience. Coming into grad school with a 4.0 and struggling anyway seems to be fairly typical. To be honest, high school was harder than undergrad.

      Delete
  27. Good article. I'm intrigued at your idea of transferring disciplines; I did the same—from chemistry to economics—though all at the Bachelor degree level.

    An interesting side note: Daniel Kahneman won a Nobel prize in Economics, and he was a psychologist.

    ReplyDelete
    Replies
    1. Kahneman just turned 80 and there is a nice write-up at edge.org.

      Econ is full of those darn heuristics of thinking fast guys that invariably fail especially when one slows down.

      Delete
  28. Anonymous3:05 AM

    Noah,

    "A physics background will not help you in econ as much as one might imagine. Actually, this is true not just of physics, but of all subjects, math and computer science included."

    How many times do an economist do use Group Theory ?

    Econ : Not that much.
    Physic : You should.
    Informatic : Every time you make an addition.

    Facetious but true :)

    ReplyDelete
  29. Anonymous4:02 AM

    Kaleberg strays into the opposite question - what does econ need to be better? But I think in a lot of ways it's the more interesting and important question.

    Certainly as an engineer used to systems analysis who then moved in to complex systems study I find that econ as a field lacks even a basic terminology for some of the things that go on inside "the system" that is an economy. And that's due to the silo effect, which is natural in academia, but kind of depressing.

    I had a short conversation with Nick Rowe in comments on his blog where it was apparent that the econ terminology was getting in the way of understanding a system of decision making - but even more (sadly) clear that unless I was prepared to list out a whole set of references and request him to learn a new terminology (and let's face it, how well do those suggestions go down when commenters make them) then there was no way to communicate to him what he appeared to be missing...

    ReplyDelete
    Replies
    1. Thanks for concurring that Econ is charlatan field. Much of the Econ stuff is take someone/thing else's formulation and force feed to the rest of us.

      I have seriously researched complexity in developing profitable investment strategies. In my opinion, stock like MSFT is nearly ideal for experiment as the sheer trading volume allows one to isolate - like we do in physics. Even then, almost all math tools fall apart; all models eventually fail (become inefficient, may be!), etc. etc. Only thing that sometimes works s when you begin with an objective that is well within the rational bounds of trades and trading.

      Delete
    2. Hey! I did not put that 5:47PM -KV.

      Impostor did! Anyway, it does not matter. Econ = Economics. Happy?

      Delete
  30. Physics intuition is all about symmetry, and about finding elegant (i.e. easy) ways to solve tough-seeming systems.

    That may depend on where you did your physics degree. We were drilled in dimension analysis and order of magnitude calculations. We were taught to visualize the results from formulas and consider whether they made any sense and then to reconcile the formula and the expectation.

    Economics could use more of the "does this make any sense" reality check that has to resort back to intuition. When I read Cochrane, for example, I see someone who loses sight of his assumptions and trusts in the implications of his formulas without asking if the result makes any sense.

    ReplyDelete
    Replies
    1. Actually, economists often do that sort of check. Order-of-magnitude checks are quite common. As for Cochrane, he was trained in physics...

      Delete
    2. Noah - I know Cochrane was trained in physics. After posting my comment it occurred to me that the experience of physics students might differ radically from department to department. My department was heavily into solid state and related experimental physics. A department focused on theoretical particle physics, cosmology or other theoretical branch of physics might produce different mindsets.

      Delete
    3. Now keep in mind, Cochrane is mostly an empirical guy, not a theorist. I've definitely seen him talk about whether the order of magnitude of his results seems reasonable.

      Delete
  31. I totally agree Noah. Being an Econ/math high school, econ/stat major undergrad turned econ grad, I had fun studying economics and enjoyed doing macro, micro and econometrics throughout. The econ knowledge gained since high school was useful fo macro while my advanced math was useful in micro and econometrics. That being said, I saw my friends and classmates with physics/math background struggle and had to put some extra studying hours especially on macro.

    ReplyDelete
  32. From what I have seen, economists don't do too well with economic intuition either.

    ReplyDelete
  33. Great post Noah,

    As a math/econ undergrad turned Econ PhD student, I can say that, just as you describe, someone with a strong math background (even those with mostly pure math background, as opposed to applied math) can "sleep through" most of the required math courses during the first and second year of an Econ PhD.

    But as you mention, the economic intuition is a different thing completely. In my class there are people with Masters in Statistics, Masters in Math and people who were stats/math majors and then chose Econ for grad school. When they all saw the IS-LM for the first they all experienced some version of: "What's going on..?"
    The math was beyond elementary, but the intuition regarding the economic mechanism to explain a simple "shift of the curves" eluded many of them.

    But also, as you mentioned, they are quickly catching up. The playing field has leveled quite quickly. This does not mean that we can all on the spot come up with a "perfect" analysis regarding economic mechanisms when we encounter a new model for the first time. Even those of us with an econ undergrad degree sometimes struggle to articulate these mechanisms, the so called "intuition," when we encounter a new model. But we are getting better and better at it.

    In econometrics the stats people feel right at home. In micro the pure math people feel this way. But even in micro, a simple analysis of the Marginal Rate of Substitution can demonstrate in how many ways can economists think of this simple equation to extract useful information about consumer behavior. It's not self-evident at first for most people, but that's exactly why we need the kind of training offered at the level of an econ PhD. No amount of knowledge of topological spaces is going to help you give economic meaning to the MRS.

    ReplyDelete
  34. Some physicists know that any life only happens on the small border between chaos and order (I take Stephen Wolfram as reference). And that there is computability beyond math.

    But back to practical things. Physics has informative data, econ has not?

    In The Blank Swan of Metal Forming I tried to explain that the dynamics of metal forming is not so different from the dynamics of, say, derivatives markets.

    http://unriskinsight.blogspot.co.at/2010/09/blank-swan-of-free-metal-forming.html

    It refers to the book "The Blank Swan", E. Ayache (how to ride on the price waves of derivatives)

    To do constant recalibration right you need to know that inverse problems are ill-posed and how to overcome that ill-posedness (by regularization schemes)

    So, there is insight that can be shared from the general philosophy of "speculative realism" down to numerical techniques.


    ReplyDelete
    Replies
    1. Herbert,

      The main contribution of physicists to biology was Schrodinger's argument that life is an anti-entropic process. The argument about the edge of chaos was not due to physicists but Santa Fe people, notably biologist Stuart Kauffman, and some computer scientists such as John Holland with Wolfram not the initiator of this idea. I also note that what is involved here is not true mathematical chaos but mere general disorder.

      Delete
    2. Thank you, I confess, I am lousy in understanding the details of the theory of complex systems, but as a generalist I pick things like the field of Artificial Life, an abstract view on systems related to life, its processes and its evolution (Chris Langton naming the discipline)

      (If I remember right M Gell-Man and other physicists were among the founded of the Santa Fe institute?)

      IMO, there is something like "economic life" and with the border argument there is no such things as an equilibrium.

      And I think, the "battle" between the economists and the "econophysicists" is about this?

      Delete
  35. Something I see missing from this discussion is the difference in attitudes that physicists have toward math versus economists. Physicists tend to be willing to take intuitive leaps with their math and are ok with leaving the finer points of a proof to mathematicians. Economists in contrast seem obsessed with making sure you know how to prove things rigorously from axioms like mathematicians do. In this respect as a physics major in graduate econ I found I didn't have as big of a mathematical advantage as I thought I would.

    I found it kind of ironic that economists work so hard to reproduce the results of mathematicians given that they also preach the benefits of comparative advantage. I also kind of got the sense that economists were trying to compensate for their lack of empirical rigor by being more rigorous in their proofs. But maybe I'm just being crochety.

    ReplyDelete