Even before the financial crisis, there has always been a surprising number of ex-physicists who find their way to graduate study in economics. It could be that many of these math-physics people have simply concluded that they no longer like physics and are interested in economics instead. (Moreover, the job market for economics Ph.D’s is much better than the job market for physics Ph.D’s.) I suspect however that some of them are here because they have some incorrect perceptions about the field. A student with a mathematical-physics background could easily convince himself that he has superior mathematics abilities than typical economists and superior statistical and computational skills than most economists. He might go on to conclude that, as a consequence of his superior mathematical and computational abilities, he should be able to enter economics and start contributing quickly and easily. He might also anticipate that he could easily adapt established models or techniques in physics to study economic phenomena and impress the profession.
If you are one of these people, let me try to disabuse you of these notions. Your mathematical abilities are actually not that much better than most economists (if they are better at all). You will have to spend a lot of time acclimating to the subject and the path to actually making contributions will be long and difficult. In all likelihood, there are very few (perhaps zero) off-the-shelf models or techniques in physics (or engineering, or chemistry, …) that will produce meaningful economic results. High-tech methods and approaches will be valued only if they can be described in simple, direct ways.
Economists are not held back because of a deficiency of mathematical tools and techniques. As soon as I hear a physicist (or a mathematician or whoever) start talking about the need for economists to use “the right mathematical techniques” I immediately think that the person has absolutely no idea what the main problems and questions in economics actually are.
Eric Weinstein for instance has somehow managed to convince himself that the instability of preferences is a huge problem for economics (it’s not) and that the application of Gauge Theory to economics will improve things. Now, I don’t know anything about Gauge Theory but I would be willing to bet that it has virtually nothing to add to economics. If Eric Weinstein has some insight that he wants to share then fine – send it my way and I’ll listen but I’m not going to listen just because the math is difficult. The fact that it’s difficult to understand something does not mean that it is important to pay attention to it. Eric Weinstein might be perfectly well-intentioned but if he thinks that because he knows some fancy mathematics, economists are obligated to grant credence to his work, he is sorely mistaken.
As another example, take Mark Buchanan, an ex-physicist who writes the blog The Physics of Finance. Mark seems genuinely interested in economics, particularly macro, but it doesn’t sound like he has a good grasp of the field or of the problems in the field. In a recent column in Bloomberg, he calls for a new age of pluralism in economics:
[M]acroeconomists should learn to speak the languages of other fields, including sociology and psychology, as well as neuroscience and engineering.
An appeal to pluralism like this is usually a sign that the appellant’s ideas are probably not particularly helpful. If you have a good idea, it won’t need affirmative action to get a hearing. True, new ideas and techniques are often met with hostility from the establishment but the reason to listen isn’t that the ideas come from outside the field. The reason to listen (if there is one) is that the ideas are good.
Lee Smolin is yet another physicist who has (at least in the past) waded in to economic waters. Smolin, like all physicists, is clearly very smart and he is asking good questions, but they are the questions of a smart undergraduate. According to Smolin, the “well known fault” of neoclassical economics is the possibility of multiple equilibria and the possibility that equilibria may be “path dependent.” Like Mark Buchanan, Professor Smolin thinks that researchers from other fields (he mentions complex systems as one) are needed to help push economics forward. Also, despite apparently having no direct familiarity with finance, Smolin is prepared to offer his diagnosis of the current state of economics as it relates to the financial crisis:
[T]he whole thing is a disaster if I can say that as an outsider. And it [was one of the reasons] why regulations were lifted on markets and trading through the decades, but when people were making arguments to Congress, to the President’s office that the economy would be better off without regulation, this was the “scientific rationale for it” and led to the very unstable situation of the last economic crisis.
Mathematical fads often pop up in economics but they won’t last unless they have concrete value. Chaos theory was tried briefly but it produced essentially nothing of any value in economics. Neural-networks, agent based modeling, path-dependent equilibria, knot theory … the list of techniques that sound impressive is long but the list of accomplishments associated with these techniques is decidedly quite short.
Naturally, there are counter-examples. The introduction of calculus in economics represented a huge leap forward. Dynamic Programming, originally developed by the mathematician Richard Bellman in the 40s and 50s, is one of the most widely used mathematical tools in economics today. The famous mathematician John Von Neumann  was instrumental in the early development of Game Theory, and so on. However, guys like Alfred Marshal, Richard Bellman and John Von Neumann and don’t come along all that often.
If you are a physicist and you want to work in economics, you had better strap yourself in and prepare for a long challenging path – one that is only worth following if you are really interested in the subject itself. There isn’t very much low-hanging fruit left (as Lones Smith once said, ‘there certainly isn’t any low-hanging fruit left in the middle of the yard!’). Don’t think that after watching Inside Job you can jump in to economics and save the day just because you understand the Navier–Stokes equations.
 For brevity I will use ‘he’ for this post when the gender is unknown.
 I’m not sure I should necessarily classify Von Neumann as a mathematician. It’s a testament to his genius that he is routinely “claimed” by so many fields. He’s a physicist! No, wait, he’s a mathematician! No, wait, he’s a computer scientist! No, wait, he’s an economist! No wait …
Chris, great post and completely agree with it, but had a question – where do you think network theory comes into this? Obviously it’s not physics. But given that some of it seems to be application of fairly standard epidemiology models, it comes more broadly under the category of techniques from natural sciences being applied in standard ways, to come up with interesting results in economics. This (http://rspa.royalsocietypublishing.org/content/466/2120/2401.short) is a good example.
Thanks for the link — I’ll take a look at the paper and see if I can make any sense of it.
I don’t know too much about the history underlying neural networks. My understanding is basically that these are models that perhaps existed in other disciplines and are now being applied in economics. I know that there are a couple of researchers who are using networks to try to understand complicated inter-connectivity between financial intermediaries but I don’t know if they have arrived at any insightful conclusions yet. (They have arrived at a lot of difficult math of course.)
There’s some interesting results even given how new it is – Acemoglu is very active there, if you’re interested http://www.nber.org/papers/w18727
Is this satire?
It should probably be considered such, imbued as it is with a degree of arrogance that everything has already been thought of. That economics would naturally be of interest to physicists should surprise no one, even if novel additions may be rare.
Some pretty strong assertions here that look to me to be overdone. I am not going to defend knot theory, but on several of the others I think we can say that either “it remains to be seen” or “there have been contributions, but now absorbed.”
So, chaos theory, the main reason why we do not see much lit on this now, ironically is predictable from chaos theory itself. So, the basic idea of chaos theory is the butterfly effect, but to the effect it holds then models subject to it will give bad predictions due to it. Thus is it not surprising that even though we observe positive Lyapunov exponents for many time series, when we estimate kernel models based on it they fail to outperform random walks in forecasting (I note that significance testing for such estimates is very hard). So, an observer like you says “nothing of any value for economics.” Really? One of the values that some of us see is that it seriously undermines rational expectations and the empirics support lots of apparent chaos out there, although ratex has been dimore rectly undermined empirically repeatedly, and you have previously here made it clear that such findings do not impress you and that macro models should continue to assume ratex anyway. So, in that regard I guess chaos theory is useless because, heck, macroeconomists are just going to assume ratex no matter how many arguments, theoretical and empirical, are brought to bear to show that it really is just false.
Agent-based modeling has made some scores, starting with the urban residential segregation model of Schelling dating back to 1971. I could list more. Getting back to macro, ABM has not replaced either NK DSGE or RBC DSGE or VAR models or even old fashioned ones derived from ISLM, even though many think they ought to. I am not sure why that is, but I think this is a situation where we should wait and see. There are ongoing efforts at ABMing macro, and they should be encouraged, not just sent to the woodshed. There are a lot of us (although I may personally be dismissable as a looney heterodox) who think that DSGE models of both types are deeply flawed, with the simply false ratex assumption at the top of the list of its problems.
And what is your evidence that path depence has been of no use or interest? I could note the multiple equilibria work of Farmer in macro, but in other areas of econ there have been many useful apps and it is seen as very relevant. I would suggest going back to look at Brian Arthur’s work on increasing returns, much of this applied to such matters as technological lock-ins and urban development patterns. This work pretty much holds up and is accepted. What are you talking about?
Guess I won’t bother with neural nets, but really, Chris, are you such a Whig that you have taken to wearing wigs? While the critiques of views of some outsiders may have punch, this dismissal of ideas listed above is just a joke that does not make you look like you actually know what you are talking about, sorry to be so blunt about it.
(Oh, and on the matter of econophysics itself, I point out that an awful lot of time series are power law distributed, which is the main hobby horse a lot of them like to ride, although it must be admitted that it was Pareto who first cooked those up, although going the other way, his original ed was in engineering.)
“Now, I don’t know anything about Gauge Theory but I would be willing to bet that it has virtually nothing to add to economics.”
This has to be satire.
Pingback: Why Are Physicists Drawn to Economics? – Orderstatistic | Marty Investor
I don’t disagree with your suspicions about physicists entering economics, but I’d suggest more caution about social science critiques of economics.
During the early years of topology, there was a brief enthusiasm amongst psychologists that topological theories would provide modeling opportunities for both human memory and imagination (it had to do with multidimensional volumes being arbitrarily contiguous with one another). Of course that phase passed, but much of what passes as economic theory strikes me as similar deterministic models proposed in other social sciences. A good grounding in the history of social sciences would probably make economic theory a bit more knowing when the next fad comes along.
You spent an awful lot of words basically saying “GET OFF MY LAWN!”
Not every technique or idea will bear fruit immediately. Sometimes they need time to be incubated & developed. Cross-pollination between multiple fields is always a good thing to attempt, even if the people going from one field to another field may seem naive to old-timers in the field they are entering. Obviously won’t always yield new insights or perspectives, but so what?
As for the instability of preferences not being a big thing… well if your models are based on stable preferences, then it almost certainly is a big thing if in the real world preferences are unstable. It’s definitely worth pursuing trying to build models from different bases of assumptions, from multiple equilibria (Farmer has some interesting work here), to unstable preferences, irrational behaviour, etc.
I think you mean “You Kids Get Off My Lawn!” in classic Letterman fashion.
“The purpose of studying economics is not to acquire a set of ready-made answers to economic questions, but to learn how to avoid being deceived by economists.”
― Joan Robinson
Having worked with physicists, the answer is clear: They are sure that they know more than anyone else, in any discipline.
BAM. I don’t think physicists are especially interested in economics more than any other discipline. They just happen to have a harder time infiltrating it; they’re all over biology/chemistry/etc, so why not economics?
I’m not a professor of economics or physics (my economics knowledge extends only to the undergraduate level), but I think a lot of this article is correct, if perhaps a bit overblown.
I’m currently working my way through Kartik Athreya’s excellent “Big Ideas in Macroeconomics” which has demystified high-end macro and explained to me a lot of the modelling trade-offs.
I think the major error every outsider (physicists more than most) makes is that they think macroeconomics is about describing literally what goes on in the economy. It’s not, and bringing along some new modelling tool from another discipline is just a fancy way of continuing to make that same mistake.
Theoretical macroeconomics is instead about obtaining analytical insight through making assumptions about the economy and modelling the impact of those models.
Macroeconomists often deliberately choose simple, unrealistic models not because it makes the maths easy, or that they are lazy or tied to some absurd neo-anarchist worldview, but because it means that the outcomes are more powerful. It is easy to explain pretty much any outcome if you assume the right pre-conditions. What is more interesting, and useful, is by using the barebones assumptions and then slowly building on them to gain insight.
For example, take the Breeden-Lucas model of asset pricing. This is a popular choice of model not because it uses some fancy branch of mathematics, or that it attempts to delve into the psychology of every decision maker. In fact by using the assumption of aggregation it only believes that there is one consumer in the entire economy, and that this consumer has rational expectations and is a price-taker. Then, by asking an abstract question in turn (what prices would mean that no trade occurs from initial positions), insight is gained as to how the fundamental value of an asset is related to its price.
From what I understand, the mathematics in macro are entirely secondary to this process of argument: laying out initial conditions and assumptions, using mathematics to analyse the implications of those outcomes. This is then followed by slightly changing the assumptions and seeing how those affect the outcomes. This process overall gains insight as to the behaviour of the economy and, equally importantly, the implication of new untested policies or policy changes on the future state of the economy.
Gauge theory, a bone of contention above, might have something to add to economics. But if it does, it will only be as a way of allowing people to analyse a more exotic or complex set of assumptions than is currently possible.
“I think the major error every outsider (physicists more than most) makes is that they think macroeconomics is about describing literally what goes on in the economy. ”
“From what I understand, the mathematics in macro are entirely secondary to this process of argument: laying out initial conditions and assumptions, using mathematics to analyse the implications of those outcomes. This is then followed by slightly changing the assumptions and seeing how those affect the outcomes. This process overall gains insight as to the behaviour of the economy and, equally importantly, the implication of new untested policies or policy changes on the future state of the economy.”
Exactly how do you know an economic insight has been gained when you are not actually interested in describing the economy?
Sure, I could come up with a whole bunch of fancy models about all sort of stuff. For example, my model could be that umbrella use causes rain. Then I tweak the assumption to have some mythical creature, that no one has seen, invoke both the rain and the use of umbrellas. I could do many things, and in fact macro does have a model for just about everything.
Exactly what insights have I gained from these exercises? None at all until I actually test them as tools for making predictions!
My comment above set out my understanding of macroeconomic argument: you specify the actors, you specify their initial endowments of technology and preferences, you specify how they all interact, and then you determine what the equilibrium outcome is.
You could as you say develop a model that, through a set of assumptions, means that “rain” is caused by “umbrellas”. I would like to see what those assumptions are, and challenge your use of those assumptions. We would then be able to adapt your model and see how changes to your assumptions alter the outcome of the model.
It would be superior if we could laboratory test each model and isolate relevant factors to assess their implications, but in macroeconomics at least, this is impossible. There are simply too many moving parts and insufficient data to separate signal from noise. Therefore you are left with the inferior methodology I outlined above.
All models are wrong, but some are useful. In macroeconomics models are precisely wrong. In the same way, maps are never literal descriptions of their territory, but still generate insight, probably more insight than a photograph might.
Let me give an example of where realistic assumptions can lead to useless modelling. Nobody seriously challenges the notions that consumers and firms are all different with different outlooks on life. They therefore have totally different preferences about things and technological capabilities. But these preferences and capabilities, which are crucial to any outcome, are inherently unobservable.
It is therefore sometimes of little use to attempt to set up heterogeneous consumers, because you would be able to explain anything away. For example you could claim that Asian-American children have a tendency for hard work and are more future-orientated than the general population, therefore this is why they are less likely to go to prison. But this would lead us away from the insight that this group has received institutional advantages which would mean that any member of the group, no matter what their preferences for hard work and the future were, would be less likely to go to prison. Assuming homogeneous consumers means you avoid this trap.
PS Eric Weinstein below neatly answered the question about gauge theory. Gauge theory is only useful if it allows us to analyse a set of previously un-analysable assumptions. Eric says it solves one relating to intertemporal welfare. Job done.
What “institutional advantages” do Asian Americans have regarding the matter of avoiding crime? This is questionable; in fact, the converse seems true.
It’s a bit funny to read the nonsensical balderdash that is Charles Barry’s writing. Little of what you write makes any sense, and considering that you are not even a college graduate, I find it hilarious to read your statements, such as “I think the major error every outsider (physicists more than most) makes is that they think macroeconomics is about describing literally what goes on in the economy.”
How many physicists have you even seen let alone spoken with? Do you even know what an integral means, what a logarithm is, or what a vector is? And yet you presume to “think” that the “major error” of “every outsider” is something they must all be blind to, something that you, the enlightened undergraduate who spews nonsense, are not. As if you yourself were not an outsider.
Regarding the matter of criminal activity amongst different demographics, it is entirely possible and likely that different groups have vastly different innate abilities. Cultures do differ, but in their essence they reflect the capacities of their people; in short, if one could somehow take ten randomly chosen groups of European toddlers and raise them all in blank-slate culture societies, then all ten cultures, while unique from each other, would still hold essential similarities. A culture reflects the innate tendencies of its people; it is naive to assume that culture has a near-monopoly on a people’s aggregate actions. Culture is something that people create and shape with each generation.
In the case of Asian Americans, not only is the above true, but there has also been the additional issue of much institutional oppression; if you knew a modicum of Chinese American history or Asian American history in general, you would realize this. The demographic was extremely alienated and oppressed in ways that have not faded today, and holding up the model minority “pedestal” only worsens the situation by giving leeway for continued oppression.
Can you please explain why you don’t think adaptive preferences are a problem for economics? Lionel Robbins told us that economics is the study of how people make choices about how to use scare means in order to pursue their ends, which are assumed to be fixed and which economics and economic models have no influence over. So the goal of economics as applied to policy is to figure out ways to maximize overall prosperity, so that people have more money to pursue their fixed ends — we generally think it becomes easier for people to satisfy their preferences if they are richer. If the models, after we diffuse them in hopes that they will encourage policies that increase prosperity, end up changing the very structure of society in ways that alters the ends people want to pursue — because of endogenous preferences — then that is highly problematic for the economist. Indeed, if this sort of process is true, then economists aren’t helping people satisfy their fixed preferences; economists are influencing people to hold certain preferences. What kind of preferences? Hmm, I don’t know, maybe ones that are conducive to succeeding in market based economies? To the extent that these types of preferences might undermine other preferences that are defensible on moral grounds, like for example preferences for community engagement or romanticism, then we have a huge problem.
There is no way I can discuss this adequately in a comment but let me give you a sense of what I mean.
There is no doubt that preference change over time. My likes and dislikes today are quite different from what they were when I was leaving college. The reasons for this change in preferences are somewhat random — I met people who exposed me to other foods, musical styles, whatever … I’m not contesting that preferences do change. The issue is whether drift in preferences is important to the main insights in economics and whether they compromise analysis in a fundamental way.
There are areas in which it might matter. Advertising is on potential area. It’s possible that advertising can change your preferences to some extent and this does present a challenge for how we want to think about that situation but it doesn’t undermine basic economic insights. It just means that you have to consider a model in which preferences are malleable. These models do open up many ancillary issues like whether people are aware that their preferences can be influenced and changed (behavioral guys refer to consumers like this as “sophisticated”) or whether they are unaware of this endogeneity (referred to as “naive”). There are also issues of how you solve such models. Typically solution methods look for Nash equilibria in which people are playing “games against themselves.” (It sounds weird but it actually works.)
Another possibility is that preferences change randomly over time. This is obviously easier to deal with.
The big philosophical issue is making welfare comparisons for people for whom preferences change because you might think (in line with the game theoretic solutions used above) that you are making utility comparisons across individuals where the individuals are actually different “selves” (different instances of a single person). Normally, such comparisons are not possible with ordinal utility. You could revert to cardinal utility which reduces the problem to one of simply directly calculating and comparing utility levels but its not clear that this is really a meaningful step to take.
As I said above, this is impossible for a comment. Food for thought.
Wow! Just saw this but I’m not sure I get your point about my take on gauge theory and preferences. Let me share my confusion about your argument.
First you write:
“Eric Weinstein for instance has somehow managed to convince himself that the instability of preferences is a huge problem for economics (it’s not) and that the application of Gauge Theory to economics will improve things.”
indicating that the lack of intertemporal ordinal welfare comparison isn’t a problem. But then you say:
“The big philosophical issue is making welfare comparisons for people for whom preferences change because you might think (in line with the game theoretic solutions used above) that you are making utility comparisons across individuals where the individuals are actually different “selves” (different instances of a single person). Normally, such comparisons are not possible with ordinal utility. You could revert to cardinal utility which reduces the problem to one of simply directly calculating and comparing utility levels but its not clear that this is really a meaningful step to take.”
which indicates that you realize just how big of a problem it is!
But then you said earlier:
“Now, I don’t know anything about Gauge Theory but I would be willing to bet that it has virtually nothing to add to economics.”
But one thing gauge theory does is to make intertemporal welfare *ordinal* in markets rather than cardinal.
So, if it solved the problem you put beautifully above……making ordinal welfare comparisons possible in markets where tastes are changing……could we physics guys get an invite to play on the lawn?
And thanks for taking a strong stand with lively prose.
Whether endogenous preferences undermine within-individual utility comparisons is indeed an interesting philosophical point. But I don’t think that’s the biggest issue at stake.
You’re still seemingly convinced that you are a describer of the social world. You’re basically telling me that endogenous preferences can be dealt with by using more sophisticated modeling techniques — e.g., by incorporating ideals from behavioral, by assuming random preference changes over time, etc. But what I’m saying is that if preferences are malleable, then that very fact means that people are wrapped up in the social world around them. All ideas from the social sciences end up influencing the social world in one way or another. Economic ideas are no exception.
You don’t seem very confident that economists have figured out yet how people’s preferences change over time. We have a lot of good leads from game theory, but “we’re still working on it,” is essentially what you’re saying. That’s fine. But if you bring a model into the policy world that doesn’t accurately account for how preferences change, then the very model itself can change people’s preferences in one way or another. In that case, you’re not doing positive science but rather normative philosophy. In which case, criticizing the, frankly, very superficial classical liberal/utilitarian assumptions behind most econ models is fair game.
What I admire mostly about Kahneman and others who are trying to expose all of the systematic bias that impede our decision making is that they truly believe that people would be better off if they learned about these biases and became less susceptible to them. In other words, the behavioral program is a normative project: it is trying to change the way people behave to make them more successful, or happier, or whatever. Those doing behavioral are clear about their goal to change the way the world works.Are most economists?
Hardly. Most are still under this absurd view that their endeavor is about positive empiricism, not normative philosophy. As such, they never want to debate the implicit normative views behind their models; they just want to have these petty little debates about, say, where the Wicksellian interest rate currently is, or whatever nonsense that usually gets debated on these blogs.
Thanks for the comment.
As I said before, I don’t know exactly what you are doing so it’s hard for me to comment on its value one way or the other. Based on your comment, I would be curious to know how to interpret welfare comparisons under changing ordinal preferences (quite apart from the technical aspect of “calculating” the welfare change. My guess would be that this has no direct interpretation but I could be wrong.
You are welcome to try and convince me (though I can’t commit to devoting a ton of time to it and I also would suspect that it’s not best done in blog comments).
Thanks for the reply. I’m reacting to the post.
I think you could be making two errors here. One is about mathematics and physics. The other is about researchers and motivations.
As far as the mathematics goes, I’m sure you are okay with differential calculus and linear algebra. Gauge theory is sort of their love child. It just doesn’t make sense to say ‘I’m okay with scalars, vectors, and matrices as well as derivatives and the occasional integral…..but that stuff that combines them is just fancy razzle dazzle and I’m willing to bet it has nothing to do with economics’.
Here’s how to invent Gauge Theoretic economics for yourself. Let’s say a function is constant if it’s derivative equals zero and that you have an agent’s wages in US dollars as a smooth function of time w(t). Without touching the function w(t), create 3 different derivative operators D1, D2, and D3 such that
D1 w(t)=0 implies US wages are constant in dollar terms.
D2 w(t) =0 implies US wages are constant in terms of the agent’s purchasing power.
D3 w(t)=0 implies US wages are constant in yen (relative to a smoothed USD/JPY exchange)
This is what gauge theory does (in a first simple situation). It absorbs the idea that constancy is situation specific and that the derivative operators should be set endogenously by the economists and their subject areas, rather than ‘at the factory’ by the mathematicians and physicists. It encodes different notions of constancy by letting linear algebra into the derivative operator to make sure the model is fully endogenous and tailored to the problem. If you don’t do this, of course, you end up one-size-fits-all derivatives which are not taking into account the different human motivations for the modeling or the different problems under study.
This is *actually* what you are railing against when you write:
“The fact that it’s difficult to understand something does not mean that it is important to pay attention to it. Eric Weinstein might be perfectly well-intentioned but if he thinks that because he knows some fancy mathematics, economists are obligated to grant credence to his work, he is sorely mistaken.”
Something very curious has broken down in our intellectual life. I’m sorry about that. If you are still willing to make the bet you state above, I’m willing to take it. But remember, the calculus you already know in economics is a form of gauge theory. Think twice before betting the family silver and college funds.
If you wanted to admit that you spoke too soon about ‘fancy mathematics’, changing preferences not being a problem, or being eager to bet that gauge theory has nothing to contribute, I would totally accept that. Otherwise we should structure the bet you suggest as an exercise in revealed preference. I’m not against making money as an inferior way of changing minds if we can’t find a better academic solution.
As I said in the post, I don’t know anything about Gauge theory so I really can’t be sure whether it has the ability to make contributions to economics or not. My skepticism is really just a reaction to a common pattern. People show up making obscure references to some mathematics that is not common in the field and they say we should adopt this way of doing things because it will have big payoffs. Gradually it is revealed that the big payoffs aren’t going to materialize and the only consequence is that some practitioners got “tricked” into wasting a bunch of time.
I’m willing to be convinced but the burden of proof is going to be on you.
Can you explain your insights / contributions without reference to the mathematics? That would be a good starting point. If, on the other hand, the initial explanation requires me to trudge through a bunch of math that doesn’t have direct connection to my own work, you are going to find me somewhat unsympathetic.
If you need a betting payoff to make this real, we can do a “Noah Smith” bet — one pepperoni pizza to the winner. I am comfortable conducting the wager in terms of the honor system if you are.
To the average Joes that are paying attention…
This is what it looks like when academics argue with each other. It’s cute, isn’t it? Dr. House is annoyed about something so he reacts about it on a blog. Dr. Weinstein reads about the reaction and reacts to the reaction. It doesn’t look a whole lot different from little brothers arguing with each other, when you break it down, does it?
Dr. Weinstein makes a statement:
“indicating that the lack of intertemporal ordinal welfare comparison isn’t a problem.”
Unfortunately, if you haven’t had any training in formal mathematics there’s a good chance that this statement will go completely over your head. “intertemporal ordinal welfare comparison” –> “to-compare a-thing-existing-in-a-set-that-has-order within-time”
And through use of this kind of lingo you have already lost maybe 90% of the population and agitated most of the rest. And all of this to eventually make a point:
“So, if it solved the problem you put beautifully above……making ordinal welfare comparisons possible in markets where tastes are changing……could we physics guys get an invite to play on the lawn?”
Do you really need an invitation? If your techniques start to yield useful results, won’t other people start to pay attention? Is it an issue of how to get the research funding to do just that?
The point is… using your own interdisciplinary jargon really muddies the water and makes it difficult for experts within their own disciplines and non-experts alike from understanding what you are talking about. The concepts that you are discussing are difficult enough as it is.
For Dr. House:
Can it really hurt to have a few more smart people looking at economic issues from a different angle?
Regarding Dr. Weinstein’s later explanation of gauge theory:
This is one of the best attempts that I’ve seen to try and communicate a complex and previously highly specialized subject to a person from another field. Can we see more of this kind of thing in academia? Can we work on trying to make this kind of communication more efficient and more transparent, please?
You’re right, physics stuff can rarely be taken off the shelf and applied to econ problems. Also, physics intuition does not translate at all in to econ intuition.
But also note that lots of top econ people did have backgrounds in undergrad physics. Samuelson, Stiglitz, and Cochrane are three I know of, and I’m sure there are more. Maybe it’s just because IQ is IQ, but I think people who switch fields can often bring fresh perspectives.
I completely agree. Switching fields is totally fine — smart people coming into econ who are ready to work hard will always be good. Just make sure you are coming in for the right reasons not because you think you have found an “easy A.”
Yes, would not do to eliminate a good former student from being an economist, :-).
BTW, Chris, gauge theory is basically a harder, more formalistic version of what Miles teaches in his 609 class – it’s just about symmetry transformations of Lagrangians. You’re almost certainly right that it’s not going to help much with econ, because there are so few believable symmetries. But it was neat to see habit formation solved in a couple lines with a symmetry transformation.
Thanks for your post. I do hope there aren’t many physicists who think as you suggest they do! I certainly don’t. Anyway, I’ve written a post in response here: https://medium.com/the-physics-of-finance/b91a9e6ee184
Thanks for the comment Mark.
I’ll have a look at your post and reply if I have anything useful to say.
Pingback: Overcoming Bias : Social Science Exists
The math in economics may be harder — at least it seems harder from this physicist’s perspective. In physics, there are lots of symmetry principles and universal laws that really help narrow down the possibilities. Economists are working without a net. The math that one could argue is “harder” (say, string theory) isn’t applicable to economics. Economic theory’s “substrate” (economic agents, money) doesn’t support e.g. esoteric topological arguments like theoretical physics substrate (the mathematical constructs we call particles).
In defense of the more respectful physicists out there, I’d like to offer some proof that were not all ego-maniacal tools — a quote from some advice to physicists I wrote several months ago:
“Theorems and fundamental laws that always apply don’t translate well to the human domain. … And be nice: make sure your model reduces to stuff economists already know (for some reason people get angry when you declare their entire grad school education was all for naught).”
There’s a relevant quote from Samuelson expressing annoyance at attempts to find analogies between physics and economics at the same link.
In my armchair theorizing, I’m excited when I come up with the IS-LM model, or Okun’s law, supply and demand, or the quantity theory of money as particular limits:
PS For the other physicists out there — I have no idea what gauge theory Weinstein plans to apply. What gauge symmetry could economics have? The absolute closest I’ve come is this idea here:
but that is effective field theory using the long run neutrality of money as a symmetry to restrict the terms in the model, not gauge theory.
Pingback: Links 3/25/14 | Mike the Mad Biologist
I sympathize with your unpleasant experiences with those coming from mathematical backgrounds not believing that there are useful tools which they have not yet heard of, which can contribute to a certain arrogance. Because of the generality and power of mathematics, its practitioners are often confident, and often justifiably so. Mathematics has a wonderful amount to offer, for instance gauge theory, which I believe promises to further improve the useful applications of calculus and linear algebra to economics in the past two centuries (Eric Weinstein gives a great explanation of the relations between calculus and gauge theory). Beyond gauge theory, there is so much more.
In short, I believe there is no doubt that many, many fields of mathematics have something (if not a lot) to offer economic theory. The influx of new tools and insights from mathematics can only help.
That being said, historical and current literature cannot and should not be ignored to the extent that it is by the mathematically confident.
Pingback: Blank you, Mankiw | Origin of Specious
Pingback: “Narrow” Self-Interest | Economic Thought
It is not a question of economist vs physicist. Both need each other. Physicists, if anything, should be able to predict the movement of physical matter. Now imagine standing in an inventory warehouse filled with packages, and attempt to predict where that package will be in one week. An economist will ask, where does the owner of the package want it to go? A physicist will ask, what are the physical characteristics of the package? In this case the economist is likely to have a more accurate prediction. In essence, while physicists have had great success in creating a compact model of the physical laws of the universe, in economics each person acts through a localized force. Those forces are called preferences. These are constrained, in part, by our physical reality. What an economist calls a decision or choice, is in essence the collapse of a wave function.
Advanced mathematics provides some wonderful, even fun, tools for us to use. However, just using more sophisticated tools will not improve results if the fundamental paradigm is faulty.
The use of gauge theory to deal with estimates of CPI (which some understand as a version of chained CPI), still does not get at what can be called “behavioral transition costs”. These are not changes in preferences. Rather it is the cost of changing from one pattern of behavior to another. As an example, in chained CPI there might be implied substitution from home ownership to home rental (in the case that home prices increase faster than rents). This ignores the fact that there is a steep cost to sell a home, and move into a rental. Those costs are transactional, but they are also psychological. Imagine someone who has lived in their home for 40 years, selling it, and moving to a new neighborhood to live in a smaller apartment. The stressors could lead to increased demand for counseling, therapy, and time to maintain the same level of welfare. A chained CPI would be biased toward underestimating the price for an indifferent outcome. This might inadvertently be interpreted as dynamic preferences. It could also be that attempting to change the essence of CPI (which is the price, over time, required to buy the same basket of goods), which does not assume anything about the utility to the consumer, into a measure that supposedly takes into account changes in utility, just opens a door where more mathematics does not close.
It is unfortunate, but increases in the quantity of genius only tends to decrease its marginal utility.
Dear Eric. You say
“So, if it solved the problem you put beautifully above……making ordinal welfare comparisons possible in markets where tastes are changing……could we physics guys get an invite to play on the lawn?”
I don’t think Chris, or any other economist for that matter, would say that you cannot make ordinal comparisons when tastes are changing. Of course one can – in many, many different ways. The problem is that because it can be done it so many ways, it is vacuous In other words, there is no point in even trying to making ordinal welfare comparisons in markets where tastes are changing.
“The problem is that because it can be done it so many ways, it is vacuous In other words, there is no point in even trying to making ordinal welfare comparisons in markets where tastes are changing.”
That is the point of view of Fisher and Shell and has been uncontroversial since the early 1970s. It was also Arrow’s assumption that his impossibility theorem about comparing many different individuals at a single instant of time was dual to the impossibility of comparing a single evolving consumer over many instants of time. This has recently been repeated by the Federal Reserve in work of Travis Nesmith and is the basis of Carl Christian von Weitzacher’s program for endogenous dynamic welfare.
The claim is that *all* of this is misdirected. Therefore, it should be interesting to learn that there appears to be a *unique* natural extension of intertemporal *ordinal* consumer welfare when prices and preferences evolve in time. That is, if preferences don’t evolve, our measure agrees with standard Konus COLA theory. If they do, the measure varies smoothly, naturally and canonically with the input without recourse to data beyond preferences and prices. In fact you don’t even need prices as income expansion paths (Engle’s curves) are sufficient.
Now you can take issue with this by questioning the math or the smoothness or other such hypotheses. But such simplifying assumptions are made all the time with no one batting an eye in economics ….. so if one wants to invoke that, I’m happy with it if they will throw out marginalism as well. Yet, they seem to want to make wrong statements about the standard framework and then go into complaints about the realism of ordinal utility or marginal analysis. But it is my contention that what you are saying that there is ‘no point in even trying to [make] ordinal welfare comparisons when tastes are changing’ is simply false….by techniques of geometry and physics. And, even better, the techniques are applied *inside* the standard assumptions without alteration.
I would imagine this would be interesting. What is even more interesting is that this is uninteresting as economists claim to know better! You will notice however that while there are many assertions that there is no unique extension of cardinal welfare comparison to changing tastess, there is no *proof* in the literature that ordinal cost-of-living can’t be uniquely extended to evolving ordinal preferences. If you will show me such a claimed proof of your assertion above, I will be happy to show you its error.
[From past history I normally don’t reply to anonymous comments but your point is solid and tone professional. I’m trying to respond professionally and respectfully as well. If it goes sideways, I have to pull back because I’m out here using my own name and the asymmetry can be problematic.]
Thanks Eric. I honestly think that if you want to convince economists that conventional wisdom is incorrect, you should develop the simplest possible example of that incorrectness. For example, a simple numerical example of the “unique” ordinal comparison in a model with 2 goods, 2 time periods, and different preferences in the two time periods.
I strongly agree with economist2001.
Hi again Eric. How about I make the question even easier. Two goods, x and y.
In period 1, px = 1, py = 1, Income = 1
The consumer’s utility in period 1 (up to an affine transformation, which is all that could ever be known) is:
U1 = x^(1/2) * y^(1/2)
(Obviously, consumer maximizes utility in period 1 by choosing x =0.5, y = 0.5)
In period 2, px = 1, py = 2, Income = ?
The consumer’s utility in period 2 (up to affine transformation) is:
U2 = x^(1/3) * y^(1/2)
Can you please tell me what your Gauge Theory says is the unique period 2 price index (i.e. the unique income level that would make the consumer indifferent between the period 1 and period 2 outcomes)?
My claim is:
Theorem: There is no unique period 2 price index
Proof: Follows directly from observation that utility can only be known up to affine transformation. QED
Hmmm…. it’s been three weeks with no answer to the unique price index in the simple example. Maybe the emperor has no clothes…..
[Note: In the last paragraph, the lone use of the word ‘cardinal’ is in error. The word ordinal was intended. ‘Tastes’ is also misspelled.]
There is not one reason that physicists find economics interesting. There seem to be multiple reasons on the part of the physicists.
There are not very serious reasons. The most superficial of all is mathematics. If a physicist thinks that mathematics is what is preventing economics from going further, they are incredibly naive about the subject and in my opinion can be ignored.
If an economist thinks that a physicists are attracted to economics because they think just the right mathematical technique will open up new vistas then they are equally naive.
The more serious physicists that find economics interesting do so because they suspect deep flaws in the theory. I am limiting my view to macroeconomic theory. The serious physicists look at economic data and then look at economic theory and say, this can’t be right. I think economics is still waiting for its “ether” revelation.
“waiting for its ‘ether’ revelation” … that is good. Some potential examples …
Mathew Yglesias called this economic “phlogiston”, but ether is a good analogy as well …
Here’s another one …
Both of these appear in Mankiw’s “Principles of Macroeconomics”, so they’re hardly outside the mainstream. The best example by far are so-called microfoundations, which to me looks like coming up with models of the ether … which physicists used to do! See the diagram at the top of this link:
Maybe developing an “economics of physics and physicists” would be the place to start. I often hear the phrase theory of everything, and yet it can not be such, unless it also explains the evolution of scientific experiment and theory as well.
As far as an “ether” revelation, you might want to study up on the phrase “surplus value” most commonly used in the 18th and 19th centuries, before the marginalist revolution. In the 20th century there was the so-called Cambridge Controversies, concerning the nature of capital. Even how we measure GDP has its issues (e.g. we do not include the value of leisure time that is produced).
Maybe it turns out the economic ether is money itself.
There is no theory of everything yet — some people think it is string theory, but that is not yet established.
But when you say “[a theory of everything] can not be such, unless it also explains the evolution of scientific experiment and theory as well” …most theories in physics do! For example, Einstein’s general relativity does describes the entire history of physics starting from Galileo. Galilean invariance (as described in his early work on motion) is what Lorentz invariance looks like when you’re not moving near the speed of light. Newton’s law of gravity is likewise a limit of general relativity with weak gravitational fields. General relativity also reduces to special relativity in flat space-time. It doesn’t tell us about the incorrect theories (e.g. ether), but mistakes are selected from a somewhat larger set than truth.
The scales of applicability of those limits then tell us the history of experiment. The speed of light is large; therefore, until we developed experimental equipment with enough precision to see effects on the order of v/c, we were in a Newtonian world.
Jason, those are good point you make. What I was thinking in terms of a TOE, wasn’t so much physicists that explain how their theories could incorporate previous theories, as it was TOE giving predictions of what theories will be in the future. In other words, if the observed behavior of matter is dependent on the models/theories we have (as in the case of artificial satellites, which would not exist if did not have “rocket science”) then there could not be a TOE without it.
Of course, if a theory was to predict that certain masses of various molecules would be ejected by planet earth (especially those used in building rockets), that would be very interesting.
Pingback: Technodox Economics and ‘Deepak Chopra Mode’ | Orderstatistic
Chris House made the statement:
“Eric Weinstein for instance has somehow managed to convince himself that the instability of preferences is a huge problem for economics (it’s not) and that the application of Gauge Theory to economics will improve things. Now, I don’t know anything about Gauge Theory but I would be willing to bet that it has virtually nothing to add to economics.”
I will remind you that argument against a “thing” predicated upon ignorance of that “thing” is a very weak basis for an argument.
Pingback: Why Are Physicists Drawn to Economics? | Bamboo Innovator
Pingback: Big data, prediction, and scientism in the social sciences | Theory, Evolution, and Games Group
Pingback: Coming into econ from physics (and other fields) | roleis.com
IHi, anyone interested in getting I.T training for non I.T’s at affordable rates. http://bit.ly/cheap-online-learning
Pingback: Economics Isn’t Physics: Scottish Edition » The Capital Spectator
Pingback: Arrogant physicists — they think economics is easy? | download Song
physicists are right.physics is the most basic and fundamental science
and the hardest also..
Reblogged this on The grokking eagle.
Pingback: Arrogant physicists — they think economics is easy? | Finance Solutions
Pingback: Coming into econ from physics (and other fields) | The Conscience of A Libertarian
Some well known scientists – Einstein, Soddy, even Edison – delved into economics. We need more of them. Neoliberal capitalism, and indeed capitalism as a whole, hinges upon dangerous, harmful silly, psuedoscience. What we need is a branch of thermodynamics which maps heat flows under capitalism, and provides some kind of final proof that, yes, capitalism must produce more debt than “value”. The anti-capitalists have been saying this for eons, but we need some sort of Physical Law now that can verify it in terms of heat flows.
Economics is a cult buried in fake science. The sole purpose of neoclassical economics is thought control of the masses by teaching bs theories in High School we end up with ill informed idiots like Louie Gohmert professing Laissez Faire policies promoted by fake supply and demand curves. It’s called group think. It’s a serious illness so far with no known cure for the economist.
Pingback: BLOOMBERG: Economic Theory Is Wrong - Московские Вести
Pingback: Everything We’ve Learned About Modern Economic Theory Is Wrong | James Alexander Michie
Pingback: Doc. »La théorie économique moderne est fausse. » – brunobertez