Ok, here we go again… someone thinks it is funny to compare economics with astronomy…
Worse, than that, it is Chad hisself, hovering near one of the antipodes of the blogosphere.
Why, some of my best friends are economists…
So, is economics really like astronomy?
Well, except for our mutual affinity for unreasonably large numbers, mesoscale problems and models that are mostly too messy to actually solve for realistic situations.
Seriously: economics has two current limits – they can do mathematically “rigorous” toy models, with over simplified assumptions, and generate qualitative sketches of asymptotic behaviour and existence of trivial properties (it’d be cruel to mention the Laffer Curve here – as much as I adore the first theorem of calculus, the existence of a maximum tells you nothing about where it is, nor its uniqueness…);
the other limit is that they have access to not inconsiderable amounts of real world data, and they can correlate and scatter plot it to their hearts’ content, but never really tie it to their mathematical theories, such as they are.
Bit like biology really, I suppose.
As proof of this, I note that the typical economists plot will not have any actual numbers on their axis, but just show crossing points and extrema and their qualitative behaviour as the input variables change.
Unless they are doing dorky multiparameter fits and showing from data how one parameter weighs in to the observed variance of the other, in which case they show ridiculous numbers of significant figures and assume they can freely extrapolate outside their data set, with little or not sanity checks on the asymptotic behaviour.
There may well be a beautiful underlying theory of economics, it may even have interesting, stable and applicable solutions.
But, the current approach will not get there: the solvable theories are over simplified, and the data driven models are generally perturbative, and fail to capture the large scale dynamics.
In physics language, economics has a strong backreaction, worse than that, it is a non-local theory, with sources and sinks – that is to say, responses are not generally some function F(t), but of some integral over F(t – τ).
Now, if you’re lucky, this can lead to damped systems with stable points, that can in principle be identified, but more generally this leads to dynamical systems with heterogenous feedback, variable lag, and non-linear responses that may jump between basins of attraction.
There may be local chaos, but, unfortunately, probably not enough mixing to get true egodicity, that’d be far too simple.
But, and I cannot repeat this point too often: electrons can’t read – they can’t ponder the equations of QED and decide, just for the hell of it, to do something different.
Er… read the theories of human economic behaviour, that is, and decide to sod it all.
If we have free will, and the system is not at the point where large N limits drive it to statistical equilibria (so Trantor may be simpler than California, in that way).
Seriously, we could, enough of us, decide to do something irrational, just because we knew that it was something not at all predicted by any ensemble of solutions of economical models.
Now that’d be beautiful.