Some reflections on econophysics and the way it's perceived

Neoclassical economics is failing. I would go so far as to say that this is no longer controversial; its flaws have been known about for many decades now. Many, many fields have sprung from this failure - I would say that this is an extremely good thing.

Some readers might wonder why I'm writing about economics at all. Importantly, I'm not an economist and have no formal training in the subject - I'm a vaguely interested party. I'm about as qualified as the pub bore (or most politicians) to tell you about the efficacy of various models; the difference is that I know I'm out of my depth. I am going to be discussing the perception of one particular field within economics, based on my experience with related fields.

A more general concern is: why talk about the economy at all? Everyone talks about the economy. Everyone else is fed up of hearing about the economy, fed up of hearing about why the economy means that the rich have to get richer and the poor have to get poorer, fed up of hearing about bullshit models when they're having to choose between heating their homes and feeding their families.

Against this backdrop, yet another middle-class schmo from the chattering classes offering yet another ill-informed opinion on what amounts to arbitrary emergent phenomena is at best useless and at worst actively harmful. My reasons for discussing economics boil down to this: when we try to think about economics, marginalised people suffer. When we say "to hell with thinking" and pursue policies based on ideology rather than evidence, marginalised people suffer more. This is something you can quantify and something you can spin heart-rending stories out of - whatever floats your boat. The point is that this phenomenon is real and not just some crap I made up because as a member of the educated metropolitan elite I love confusing people with numbers, or something. The point is that it is in our power to stop this phenomenon and use data to make a better world, and that we should be doing that instead of uncritically following ideology.

(I do actually confuse people with numbers a lot, but my brain is built that way. People have tried to socialise me and make me more normal, but it hasn't worked particularly well.)

One of the fields which gets more attention is econophysics: applying techniques developed in physics to solve problems in economics. The rationale behind using these techniques is that they work much better for the real world than the traditional tools of economics (which work for homogeneous agents at or close to equilibrium, rather than what we have - heterogeneous agents and plenty of situations far from equilibrium...like, let's face it, the economy never reaches some stable point of equilibrium).

Okay, so beyond more maths, what's there to complain about?

I'm going to leave the actual field of econophysics alone for the most part, because I have no experience with it. I have experience in two things that are related, though: evaluating the relevance of a particular technique and people's perceptions of econophysics.

I'm not in a position to analyse the use of every single technique in econophysics ever - it's a huge field and physics is even bigger. Certainly on the surface most of the uses of the techniques check out: large numbers of agents interacting with each other, heterogeneity, not in a stable state...we have tools for all of these things and I hope we can keep refining them.

However, nonlinear systems (like the ones I am discussing right now) can be problematic: many, many, many times you can come up with or apply some model to fit your data, and it will fit. On the surface, that seems like something you want to happen. Of course you want your models to fit the data. The problem is that in nonlinear dynamics, it is quite possible to fit a model to some data and for that model to not represent what's actually happening at all. In fact, it's something people spend quite a lot of time discussing. An example I like to bring up is the critical positivity ratio: two psychologists tried to use differential equations from fluid dynamics to describe human psychology with no theoretical or empirical justification. This last part is important: some justification for your model is needed before you apply it. In the case of econophysics, I think it's mostly justified.

At the same time, I feel like this need for justification is hardly stressed outside of specialist books; textbooks will definitely stress this. Some very good popular science books do stress this, such as Chaos (James Gleick) and Critical Mass (Philip Ball), but still the message doesn't get through.

I think part of this is due to the perception of laws of physics as fundamental and immutable, something built on more solid a foundation than any other science could dream of. This isn't something I've made up: I have literally seen people make the argument that econophysics is better than neoclassical economics because physics is based on fundamental principles.

Discussing the immutability of physical laws and whether we can ever truly know them, or simply make a series of better and better approximations (for a start, physics is still developing and that suggests we still haven't got a handle on all its laws), would open up a whole can of philosophical worms. For the moment, I will simply say that even if you believe in the immutability and fundamentality of physics, that in no way implies that every single application of physics is correct. For example, I could describe a bacterium in quantum mechanical terms. While quantum biology is a real discipline, writing down explicit equations with no approximations would be hellish because there are so many particle interactions going on. You choose your tools based on what is appropriate for your problem.

Something else usually left out of the discussion of econophysics is actually the most critical part of the field: how well does it work?

According to a quick snoop around Wikipedia, econophysics is good at explaining and modelling fat-tailed distributions (think 80-20 rule), which crop up everywhere in economics regardless of how big your data set is (this is called scale invariance). This is good - it's a demonstration that econophysics actually does work in some areas! At the same time, I'd be critical of that on a political level: when you frame issues as products of mathematical, immutable laws rather than consequences of policy, people become much more reluctant to do something about them. I can very easily see this being used to argue that gross wealth inequality arises from mathematics rather than from politics, for example.

I would like to stress that I'm not saying econophysics doesn't work. I am urging scepticism and restraint in using the theory. People's lives and livelihoods are too important to hype a theory to the stars without testing it to death first - and we know this. We've done this. Actually, we've done this all wrong by pushing neoclassical economics hard even as it failed again and again. We can't afford to make the same mistakes.

Comments