Having just read this made me realize what people are actually thinking about when they are making their assumptions on economic modelling:
"(..)economists have conventionally assumed rational behaviour. There's a reason for this. Such an assumption generates testable predictions, whereas if we assume people are mad then anything goes."
What people like the above do not realize is that it depends how you define rationality. Although mainstream economists say that they understand the limits of their models, it appears that do not actually do so. Most of the economic models I have seen assume homogeneity. That is basically stating that all people behave alike. This makes calculations easier but it doesn't make them more truthful. To be fair, people who actually use heterogeneous agents exist in the economic literature. Yet, they are actually imposing behaviours which do not really make them so distinct. For example, a recent paper (which I will not name) states that "we have heterogeneous agents who, for simplicity, we assume that they behave similarly." The calculus behind heterogeneous agents is indeed horrendous thus I will refrain from criticizing people for not using them any further.
Yet, the problem still remains: people do not behave similarly. Basic psychology tells you that. (actually a day in the world tells you that but how many economists who go out in the real world do you know?) Now, if I am presented with three different choices, I will most probably choose something different from what Joe, Jim, Sarah and Mary would. Which does not make my choice any more or any less correct than the one of the others.
Yet, economists propose that a "Rational Expectations Equilibrium" exists and it is the "optimal" choice. First of all, what is "Rational Expectations" and what makes it the correct choice? If given some data I reach conclusion A based on deduction and Mrs B reaches conclusion B on a whim, what good will rational expectations do to me if B happens? Nothing. The point is that in real life there is no "rational expectations" because shocks do exist (call them Black Swans or Normally Distributed errors or whatever else). If economics is trying to explain the real world then what is the point of making a false interpretation of it?
Is a rational expectations equilibrium reachable? Maybe if all the people in the world were machine-like, operating in the same way every time. Oh, wait: machines break; and make mistakes. Not to mention what we would also have to force the earth not to create hurricanes, earthquakes, tsunamis, too much rain or too much sun. Since perfect forecastability for these events is not really an option, then we cannot really obtain a perfectly forecastable equilibrium.
Rationality is not some notion which should be totally discarded though: one point is true. People are rational, yet in their own individual, unique way. What do I mean by that? Well people believe that they are doing the best they can in every situation. If you ask a person why he chose option A instead of B he will justify it. In that person's mind the best choice given the knowledge he had at the given time and the outcome he wanted to reach (which is distinct in every person I would remind you!) has been made. No person would choose to make a wrong choice (some may do but they are sure to be sure that the wrong choice would make them happier, e.g. letting a child win at a game).
It is simple to see that we are all rational. Nevertheless, we are not the same and the future cannot be forecasted accurately. This are the main points that economic modelling is lacking.
Oh, and a steady-state doesn't really exist either.