Both Noah Smith and Mark Buchanan have posted replies to my earlier post on the tendency for physicists to be attracted to economics and both make good points.
After posting my original entry I thought about whether there might be areas of economics in which physicists would be able to contribute quickly and it occurred to me that perhaps finance – particularly high-frequency trading – might be such an area. Here lots of trades and market patterns might be governed by the technology of trading and could present an environment in which the skills of physicists are particularly valuable. So I was pleased to see that Mark seems to also focus on finance as an area where physicists have found a grip.
He mentions one particular area – distributions of random variables that have “fat tails” – which has specific value in modern economics. Several researchers are using such distributions to analyze economic phenomena. Xavier Gabaix has used power laws to study “granular” business cycles – the idea that random shocks which affect large players might have substantial consequences for the aggregate economy. Robert Barro, Francois Gourio, and others consider the possibility that such distributions place weight on extreme outcomes which in turn influence asset pricing (asset pricing implications of rare disasters). Trade theories, models of urban (spatial) formation, etc. use power laws. If physicists are actually the source of some of the background on fat-tailed distributions then kudos– that’s a point for physicists in economics.
Mark’s post also presents an example of what you might call ‘Deepak Chopra’ mode. This occurs when a writer (or speaker) talks about an issue which is complicated but then takes the opportunity to start hurling around complicated-sounding ideas and jargon with at best tangential relevance to the topic and at worst no relevance to the topic whatsoever. (Deepak Chopra is known to start talking about quantum healing and other new-age concepts that have the word “quantum” attached to them …). Here is Mark veering dangerously close to Deepak Chopra territory:
What the physicists DO believe, however, is that markets and economies are great examples of what scientists have come to call “complex systems” — systems of many elements (people, firms, etc.) with strong interactions between those elements which create webs of non-linear feedback. The elements learn and adapt, their interactions create “emergent” coherent structures and fluctuations at the collective level, and these structures then act back downwards to influence the behavior of the elements.
He quotes a similar sentiment from the essay by Brian Arthur:
It is this recursive loop that connects with complexity. Complexity is not a theory but a movement in the sciences that studies how the interacting elements in a system create overall patterns, and how these overall patterns in turn cause the interacting elements to change or adapt. … Complexity is about formation—the formation of structures—and how this formation affects the objects causing it.
(… and this is why we need physicists?)
OK, the economy is “complex”. There are many players on the stage and they take a variety of actions which together contribute to the aggregate behavior of the economic system. In turn, aggregate outcomes influence individual behavior. This simultaneity is indeed one feature that makes analyzing the economy so difficult. The thing is, this type of two-way feedback is what economists are doing already. We often reduce environments like this to “fixed point problems” – the players take the market conditions as given when they make their decisions and in turn, these decisions generate the perceived market conditions. In fact, we often prove that equilibria exist by appealing to mathematical “fixed point theorems” for exactly this reason. You might think that this rules out learning and adaptation to an environment. It doesn’t. Economists have been analyzing environments with learning for decades. Learning does make things a bit more difficult. In learning models, the market participants start with some subjective beliefs upon which they base their actions. These actions lead to market outcomes and beliefs are refined. The interplay between beliefs, actions and outcomes is analyzed simultaneously (and it plays out over time). A bit more difficult, yes. But not an insurmountable obstacle for the field.
If economists already do this stuff, why do smart outside observers like Mark Buchanan think we do not? Why are we being lobbied by physicists like Eric Weinstein to adopt techniques which are of such speculative value? My guess is that techno-heterodox ideas (Technodox economics?) like gauge theory, agent based models, chaos (or complexity) theory are advocated in the hope that it might either (1) provide outsiders with some authority even though they don’t really know much about the field and (2) because most practicing economists will have to admit (if they are honest) that they know essentially nothing about these techniques – an admission which could deprive them of some authority even though they do know quite a bit about the economy.
Incidentally, the jargon which comes along with a lot of these Technodox ideas allows even those who don’t understand these speculative techniques to masquerade as though they do understand them. Jargon is a very problematic part of academia. Often academic jargon is particularly imprecise on top of the fact that it allows the speaker to try to “pull rank” on the audience.
As an aside, Eric Weinstein commented on my earlier post by arguing that gauge theory was actually valuable. I told him that I was willing to be convinced but I asked him to convey his insights without referring to the mathematics. He hasn’t got back to me yet. (Eric also commented on Noah’s post with a lengthy comment with a bunch of mysterious mathematical notation. I doubt he is going to convince people by taking this route.)
I think you are making many excellent points. And it is extremely frustrating when outsiders talks about economics as if we missed the whole idea that the economy is a “complex” system. Propagation, feedback from expectations to outcomes (and vice versa), coordination problems, externalities, etc. etc., are all at the heart of economics. But for some reason it is overlooked.
Anyway, I am also willing to listen to Eric Weinstein if he could come up with a simple setting in which gauge theory leads to new insights that could not have (or at least have not yet) been reached by conventional tools. Say some application in a two period, deterministic, setting. I get the feeling that it is more a “beautiful mathematical language” that can rehash most of modern economics. But the usefulness of that is also doubtful (in particular if that language is divorced from the way we think about economic aggregates, as, for instance, real, or natural numbers).
I also included a link in a comment to your previous post referring to Eric Weinstein’s TOE. This is actually quite an embarrassing story for both Eric and Oxford, but illustrates well that Eric is not shy of making extremely bold claims with very little substance behind it. At the same time Eric discusses in a nice tone, and seems very clever in many dimensions, so I am still very curious!
Weinstein’s wife wrote a thesis applying gauge theory to inflation measurement problem, leesmolin.com/wp-content/uploads/2013/06/MalaneyThesis.pdf
But that was almost two decades ago, and apart from that, I don’t think Weinstein has written down an actual paper about his theories. Which is a bit strange, given that he’s been making these rather grand claims about gauge theory in economics for years now (search for “economics manhattan project”).
To fill out the discussion, I thought I’d just offer a few other Deepak Chopra-related links: 🙂
Click to access speech386.pdf
This is sometimes Deepak Chopra impersonator Andy Haldane (Bank of England) on complexity and the stability of financial networks.
Here you’ll find some not unrelated thoughts by a team of Deepak Chopra mimics including Joseph Stiglitz, Bruce Greenwald and other economists (plus one physicist). They discuss complexity and the consequences of dense interconnections (and “nonlinear” interactions!) in financial networks.
Click to access WhomOrWhatDoesRepIndRepresent.AKirman1992.pdf
Of course, this is famous Deepak Chopra imposter Alan Kirman on complexity, agent interactions and the general failure of economic models to tackle them in any realistic way.
Finally, why not mention alleged Deepak Chopra admirer Michael Woodford, who suggested that economics needs to get a lot more serious about modelling how people form and change their expectations, and how those expectations interact in an ongoing way to create collective outcomes.
The above offered in the spirit of a gracious “we can agree to disagree.”!
Life is a little bit too short to read all your links, so I decided to focus on Woodford. There is nothing Chopra-esque about his text (which I, in case anyone is interested ;), agree with very much).
Of course Mark though I conjecture that we probably don’t disagree as much as you might think. Cheers.
you’re probably right. We’re arguing over the 20% we disagree on, ignoring the 80% we agree on. I bet we DO have very similar reactions to Deepak Chopra himself!
Well, this physicist can’t seem to figure out what the benefit of Weinstein’s application of gauge theory is. Yes, the properties of “utility” — if it behaves in the precise mathematical way it is defined — has the structure of a fiber bundle. It also has the structure of thermodynamics. See e.g. here:
But if economic agents had well-defined utility that behaves in the precise mathematical way it is described, then economics would have been already completely sorted out by Walras and Pareto. The interesting part about economics is that it doesn’t, and it hasn’t!
I will also venture the rather bold claim that there are no topological non-trivial solutions in economics. No instantons or Aharonov-Bohm effects. Therefore the fiber bundle picture adds nothing to the understanding.
Let me first apologize for making inappropriately intemperately personalistic remarks about you in discussing these issues earlier. I wish to retract them. That said, I continue to disagree with much of what you say here, although like Mark Buchanan I suspect that there is more agreement than disagreement. I also applaud Mark’s set of links, which are indeed to various well-known economists. It is not widely known, but Michael Woodford was originally a complexity guy who then turned into a DSGE guy, but who has recently somewhat drifted back to his complexity roots.
For those who are not sure what complexity economics is all about, let me throw out a somewhat out of date but still pretty readable and not totally wrong piece by me from Fall 1999 Journal of Economic Perspectives, “On the Complexities of Complex Economic Dynamics.” My views have changed somewhat since then, but basics are there and a lot of people subsequently have adopted some of my terminology and arguments made there, but will not waste time citing my later stuff. Anybody who is really curious can just go peruse my website where most of it is at http://cob.jmu.edu/rosserjb .
So, I think that this is less about bringing in the physicists, although I think most of us agree on where they are currently being the most useful, particularly on the matter of distributions that are power-law, whether city size of time-series financial or wealth distributional, etc. As Mark notes, the movement is mutlidisciplinary, indeed I have argued it is transdisciplinary (yes, there are meaningful distinctions between “multi-disciiplinary,” “interdisciplinary,” and “transdisciplinary,” but will not go further on that here). The issues of nonlinearity, positive feedbacks, emergence, and so on are not just coming out of physics, with if anything the somewhat fuzzier, albeit important, one of emergence, which seems to set your hackles going, Chris, comes much more from biology than it does from physics, with some of the harder core econophysicists such as McCauley denouncing the idea for its fuzziness and failure to abide by invariance laws. I have already commented on chaos theory and ABMs and will not do so here further.
On Eric and his gauge theory, as near as I can tell it is a relative of the divisia index advocated by Bill Barnett, who has been as much of a supporter as anybody in econ that I am aware of of Eric’s position. I stand to be corrected on that by anybody who wishes to do so.
As for Chopra, I certainly have never taken him remotely seriously, and I am not aware of anybody in the general complexity community or the narrower economic complexity community that has either. OTOH< "technodox" is a cool neologsim. Congrats on coining it, :-).
To turn this question on its side, why is economics so hostile to physicists?
I have a running-joke that economics may be science, but it is not one of the sciences. In every other field that I have participated in – biology, neuroscience, chemistry, etc – ideas from physics are welcomed or at least unremarked. There is a certain level of intellectual discussion between the different ‘science’ fields. Yet economics always seems…distant. Like it doesn’t want to be one of us.
I don’t think I would say that I am hostile to physicists but I am skeptical of nebulous results that can’t be explained directly in plain English.
It would be one thing if we were talking about an “open problem” in economics which a physicist then tried to solve using some technical apparatus which could be explained and verified openly.
It’s quite another for someone to claim that we are doing XYZ all wrong and that if we only adopted their techniques we would be much better off but when asked to explain the technique or the problem or the intuition all we get is techno-babble.
Sorry, I didn’t mean to imply that you were hostile to physicists in particular, but rather that the field as a whole was. And I do think that the econophysicist ‘you guys are idiots’ approach is not appropriate (and all too common from physicists). otoh, I find the requirement that it be explained in totality in plain english weird; as seen in physics and other fields, often complex systems require complex mathematics to understand. Quantum mechanics or alignment of magnetic fields (ie Ising models and symmetry breaking) can be difficult to explain in plain english, no?
I’m actually curious about your opinion on a related question (to turn this a bit further around): why are economists not interested in using their techniques in other fields. In other words, why aren’t they ‘pulling a physicist’ and entering ie ecology, psychology (which I suppose they are), cog sci, neurobiology, etc? negative points for using the word ‘humility’, we are all scientists after all 😉
I think there are very few contributions in most fields that cannot be explained (at least at a very simple level) with ordinary language. Even Einstein’s theory of relativity can be explained in simple terms. There are certainly counter examples. The Taniyama-Shimura Conjecture for instance requires some initial exposure to eliptic curves and modular forms. If someone has an idea or a result that can’t be explained in simple terms then this becomes much more costly for them. It’s difficult to get people to pay the up-front cost of learning a bunch of mathematics when the payoff is just “you’ll see it when you get there.”
As for economists entering other fields, it does happen. Economic techniques are used frequently in Political Science for instance. Obviously, as you mentioned, there are connections with psychology. Evolutionary biology uses game theory to some extent. But in each of these cases I would guess that the host field (e.g., Political Science) understands the nature of the contributions of the economics. In many cases, the host field solicited the external techniques.
I don’t think this is a completely correct perception. For example, here is a great post on this in ‘network science’ (i.e. sociologists pushing back against physicists). Personally, I have experienced this push-back in ecology, and have pushed back on behalf of psychology. If you want examples from neuroscience, then Chris actually gave you a great example in Deepak Chopra: just look at all the quantum mind nonsense.
However, there does seem to be a bit more push-back from econ, and I suspect it is an artifact of a (1) more developed blogging commuting, (2) respect for mathematical theory, and a (3) presence of non-physics maths.
To complain about something like a physics invasion, you need an informal medium like blogs, because papers can only be used for particularly obvious cases (but if they are so obvious, then why even bother?) like quantum mind in neuro/psych. As you pointed out, fields like neuroscience simply don’t have a blogosphere in which to voice their concerns.
Overall, to much of biology and psychology (not as sure about ecology and neuroscience), mathematical theory is more or less irrelevant. The fields are dominated by experimentalists. Physicists can’t often enter with better experiments (at least not anymore, it seems, I guess they did back in the 50s with crystallography; it would be interesting to see the historical record there) because that actually requires skill in the discipline they are entering. The theorists in these fields, are too busy trying to justify to experimentalists why they matter, and so they don’t have time to complain about theoretical physicists encroaching, instead they prefer to co-opt them into their case for “we should exist and be heard”.
In econ, on the other hand, it seems to me like the experimentalists are much less prominent, and theory is taken very seriously. As such, there is an established community that can protest and who has protest in their interest.
Finally, this brings me to my last point which is closely connected to the previous. Economics has developed its own mathematical tools relatively independently of physics. As such, the mathematics of the two fields differs, and econ has its own established flavour. When physicists climb in, it tastes strange.
In biology and neuroscience, this doesn’t seem to be the case. The theory that was built was actually largely built on top of physics math to begin with. In genetics there is an exception with a bit of a connection to non-physics-y statistics, and I don’t know of any non-physics-y ,math in neuroscience, but maybe you can give me some examples. As such, when physicists wander in, their math has a very similar flavour and is thus less objectionable.
Reading the comments above, I think I might be out of my depth here. Still…
I thought Dr. Buchanan’s comments were quite relevant. The reason why there’s considerable failure in being able to predict economic outcomes reliably is because economic environments may not lend themselves to being analysed as fixed point problems. Here’s an important problem in economics: What will be the likely effect of QE? I’m not sure if current economic theory has the tools to answer that question with a reasonable amount of confidence. I’m not saying physics has those tools either; complex systems are hard to model. We’re only now getting a handle on meteorology, which studies a system whose elements and interactions are relatively easier to model. But how about, say, planetary ecology/evolutionary biology? If a researcher in complex systems were to go back 65 million years, would he be able to predict the likely course of planetary evolution in the case of a giant meteor strike? Near-term effect of extinction of dinosaurs may be easier to predict, but would he be able to predict the evolution of human beings, something that took much longer? An analogy can be drawn as to the impact of QE. Do such scenarios submit themselves to being analysed as fixed point problems? I would like to know.
I think, with exponentially increasing computing power we might be able to answer questions relating to complex systems by modelling them correctly. Some efforts have already been made, of course, in the form of agent-based computational economics. In my opinion, the ACE approach to answering predictive economic questions may hold greater promise than the current one.
Can you teach me a little about agent-based models, please? I’m ignorant about how they differ from micro-founded macroeconomics.
How do the agents in agent-based models behave? Are they pre-programmed, or can they use some rule to change their actions if they would benefit from so doing? I ask because pre-programming sounds a lot like assuming the answer, and “changing actions based on some rule” sounds a lot like rational expectations macro.
Is the difference just that ABMs will get you a time-path, whereas most macro looks for an equilibrium/endpoint?
I’m nothing more than a dabbler and other commentators should be able to address this far better, but as far as I understand, the crucial difference between ABMs and microfoundations is that the latter looks for an equilibrium/endpoint as you’ve mentioned, while the former doesn’t.
The agents are pre-programmed, of course, otherwise how would they respond to their environments? But, they’re also programmed for learning and can alter their responses based on previous experiences.
You may find a synopsis of ABM theory here: http://en.wikipedia.org/wiki/Agent-based_model#Theory and can check out a model here: http://p.seppecher.free.fr/jamel/.
Pingback: In Praise of Linear Models … | Orderstatistic