My previous post on assumptions was not quite rigorous enough in its definition of assumptions, and attracted some skeptical feedback from the commenter named isomorphisms. Allow me to reiterate my point more clearly.
The distinction between hypotheses and assumptions was intuitively appealing, but of course all assumptions could be said to be hypotheses in a sense. However, I think most scientists would agree that a useful assumption has definitive characteristics, even if it’s difficult to pin down exactly what those are. I think they’d also agree that counter factual prepositions about the mechanics of a system are not useful assumptions. So what are?
At their heart, assumptions are intended to simplify analysis – this is an oft-used defence of economists. But the crucial way in which assumptions are able to do this is by eliminating a specific complication. Of course, this alone is not a sufficient condition. Assumptions also need to have a clear impact on the analysis, too, so we can be sure what happens when they are relaxed.
How many economic assumptions meet these two criteria?
Firms equating marginal cost to marginal revenue certainly doesn’t, as it’s a preposition about the nature of the firm, rather than an assumption that simplifies the nature of the problem – in fact, cost-plus pricing is far easier to calculate and also appears to be used far more widely used.
Perfect information can’t be said to eliminate a specific complication – it’s simplifying in a sense, but it potentially ‘simplifies’ the analysis to the point of undermining it, hence creating its own complications (you’d eliminate most real-world firms). Analysis is entirely possible without this assumption – ‘Schumpeterian’ economics uses imperfect information to its advantage.
Rational self maximisation, on the other hand, is a good example of an assumption that is defensible, as it allows us to simplify how people make decisions and has clear implications. Furthermore, it can easily be modified to include behavioural characteristics such as loss aversion (though economists seem unwilling to do this any time soon).
I stand by the idea that assumptions are an appropriate target for criticising economics, and feel this is a much more coherent and useful definition of what makes a good or bad assumption.