It’s hardly the most significant issues in normative neoclassical welfare economics, but there seems to me still quite a lot of confusion floating about relating to the role of the assumptions of perfect information and perfect rationality, which I suppose is only to be expected giving the confusing and idiosyncratic mix of the positive and the normative, fact and value, science and ethics associated with that theory. Let’s discuss that this week.
The main point I wish to establish is that the significance of the assumptions of perfect information and perfect rationality is entirely different in a normative context from in a positive context. In a positive, empirical, scientific context, perfect information and perfect rationality can be seen as simplifying assumptions that needn’t be true for the theory in which they are embedded to be highly evaluated on a scientific basis by its ability to predict. The use of such assumptions is just an analytical choice that may be useful in constructing predictive, scientific models without getting bogged down in complicating details that may add little to the predictive accuracy of the results. Certainly, those particular assumptions make possible a great deal of mathematical model building that might otherwise be infeasible, although of course the scientific, empirical evaluation of the resulting models is another matter entirely.
In contrast, in a normative context, the “assumptions” of perfection information and rationality are more properly classified as “factual premises” that must be true if the conclusions are to be relevant to reality. The method for evaluating arguments or theories proper to ethical philosophy involves logic and precision. If one is evaluating an ethical proposition about what to do about other people eating apples that may be poisoned, and one knows ninety-nine out of every one hundred people know the apples are poisoned and one doesn’t, then that’s the correct factual premise that should appear in one’s ethical theory. The correct factual premise is not the approximation “everyone knows the apples are poisoned.” That’s a false factual premise. If one bases one’s ethical conclusions on that premise, one’s normative conclusions will not be applicable to the real world. One’s theory will not be highly evaluated according to the relevant criteria. That’s the way ethical philosophy works.
Nor can I can’t fix it by saying, look, I understand the problem. I dutifully call instances where the factual premise is false in realistic situations “theory failures,” and I try to address those theory failures by telling everyone I meet the apples are poisoned. My ethical theory would still be garbage, if not as a funny parlor game at least as a serious ethical theory, because the conclusions would be still inapplicable to the real world. The theory would remain garbage until I introduced true factual premises.
It’s also important to realize that in normative neoclassical welfare economics, the factual premises of perfect information and perfect rationality are relevant to the evaluation of the fundamental ethical or normative proposition we should maximize total social utility. It’s a theoretical mistake to suppose those factual premises are only relevant in the context of specifying the conditions of a chimerical ideal market structure everyone understands is difficult or impossible to attain in a literal sense. It’s relevant to the entire normative program. It’s not the case that our theory can establish we reach pretty good normative outcomes even if we never attain those conditions exactly and hence arrive at ethical perfection. That’s not the way ethical philosophy works. My ethical theory of what I should do about people eating poisoned apples using the false factual premise everyone knows the apples are poisoned isn’t pretty good because the premise is approximately correct.
And if we’re thinking of perfect information and perfect rationality as hypothetical conditions rather than factual premises per se, then we not only need to identify them as such to avoid confusion with factual premises, but our normative or ethical theory needs to contain propositions about what we’re meant to do when those hypotheticals are not met; otherwise, again, it’s simply an example of a bad ethical theory, in that case because of incompleteness. In the context of the ethical or normative program of neoclassical welfare economics, are we still meant to accept the goal of maximizing total social utility when people don’t have perfect information and don’t exhibit perfect rationality, or are we meant to be interested in doing something else in that case? Because based on how we’ve defined “utility,” it sounds to me like that might be a bit controversial.
When one tries to address science and ethics, facts and values, the positive and the normative at the same time, confusion and conflict is the likely if not inevitable result. We should fix bad economics by taking the normative content out and allowing economics to become a real science.
