Economics has come under a lot of criticism recently, and proponents sometimes try to defend themselves by pointing to Milton Friedman’s methodology of positive economics. In this essay, Friedman
loses his mind argues that the assumptions of a theory do not matter as long as its predictions are correct. Of course, even if you accept this there are still plenty of criticisms of neoclassical economics – the theories internally contradict themselves, and it is also incredibly hard to verify empirical predictions in social science, making Friedman’s litmus test somewhat of a damp squib. However, let’s put these objections to one side.
In the essay, Friedman takes us on a typical Friedman logic train to his preordained conclusion and leaves you to puzzle over how you got there. He argues that to be completely realistic a theory would have include everyone on earth’s eye colour, qualifications, etc. But how do you test whether or not to include these things? You see whether evidence corroborates the theory without them! Fantastic – assumptions don’t matter.
In this case, Friedman’s sleight of hand lies in not properly defining the word ‘assumption’. This is, in fact, so significant that it means his paper is effectively advocating any methodology whatsoever (if I assume throwing darts at this board will give me the GDP figures for next year…). The crucial characteristic of assumptions in engineering or science is that they eliminate specific variables. A perfect gas is one where many of the smaller forces between molecules are ignored. Assuming a vacuum eliminates air resistance. This gives us an appropriate method, as ‘relaxing’ an assumption means adding in more variables, and this process can continue for as long as it is practically feasible.
However, many economic assumptions could not be argued to be eliminating a certain variable – assuming that people are rational self maximisers, or that firms calculate expenditure based on marginal costs and revenue, are actually hypotheses, not ‘assumptions’ in the scientific sense of the word. As such, the ‘assumptions’ themselves are empirically falsifiable and cannot be swept under the rug.
Furthermore, in science theories are only deemed as valid as their assumptions are realistic. A theory can always be improved by making the assumptions resemble reality more accurately. So even if we were to accept Friedman’s premise, we could still improve theories by abandoning rationality based on behavioural evidence, or abandoning marginal cost based on surveys*. Unsurprisingly, in the case of widely used economic models such as Arrow-Debreu, it is incredibly difficult to relax assumptions before the theory collapses – if this were true in physics, the model would be abandoned.
To be honest, it is a sad indictment of economics that an essay which contains the passage:
The articles on both sides of the controversy [regarding marginalist analysis]…concentrate on the largely irrelevant question of whether businessmen do or do not in fact reach their decisions by consulting schedules, or curves, or multivariable functions showing marginal cost and marginal revenue.
has to be critiqued formally. A theory’s assumptions are always relevant to its conclusions, and improving them will always yield more accurate results. That much is obvious to the man on the street, but clearly not to economists.
*Friedman argues surveys are as useless as asking octogenarians how they account for their long life. I can only interpret this as him saying businesses have no idea what they are doing, which sort of undercuts him intellectually elsewhere.