Economics and the arrow of time
from Bruce Edmonds
The question of the importance of the “arrow of time” has come up in the “Unpicking the anti-neo-classical” thread here. It really deserves a seperate thread so I am “restarting” the discussion. It was raised as an argument as to why physics and economics are different, that the arrow of time was inherent in economic phenomena but not physical.
1. This is somewhat irrelevant as a difference, since except for possibly sub-atomic physics the arrow of time is essential and fundamental to most systems that physics studies. Indeed the second law of thermodynamics (where the arrow of time is essential) is often thought of as one of the most basic physical laws. All complex systems physics includes the arrow of time. Whether there is an arrow of time or not does not change how physicists go about their subject.
2. The astounding oddity is that economists seem to want to avoid temporal dynamics in their analysis. People have memory – that is a knowledge of events that extends only one direction in time. Many economic processes undergo “lock-in” where the particular and somewhat arbitrary history determines a fairly stable patterns of goods and institutions into the future (e.g. the QWERTY keyboard). Equilibria are not observed in our economic life except in the most limited and constrained circumstances (e.g. some behavioural experiments). Our economy is awash with devices, tools and ideas that clearly are undergoing some sort of evolutionary pattern (variants are usually of past items, forces of selection mean only some are successful) which do not have any equilibria at all.
3. Physicists have a whole range of analytical approaches to characterising the behaviour of the systems they study. I read quite a lot of physics since I collaborate with physicists every week – I can not remember seeing a paper that even bothered to prove an equilibrium in the last 10 years.
Thus the fundamental difference between physicists and neo-classical economists is, once again, that their models do not represent (indeed are not even concerned with) what is observed. At best this makes them an obscure and not very useful branch of applied mathematics (in which case they should not be awarded many grants and certainly not let loose advising people on real world problems), but more accurately (in my opinion) they are just bad scientists who do not change the modelling assumptions when they are wrong.