… is from page 61 of Vol. 9 (Contra Keynes and Cambridge) of Hayek’s Collected Works – and in particular from his 1963 essay entitled “The Economics of the 1930s as Seen from London”:
To me it seems as if this whole effort [begun in earnest in the mid-20th century to scientistically ‘scientificize’ economics] were due to a mistaken effort to make the statistically observable magnitude the main object of theoretical explanation. But the fact that we can statistically ascertain certain magnitudes does not make them causally significant, and there seems to me no justification whatever in the widely held conviction that there must be discoverable regularities in the relation between those magnitudes on which we have statistical information. Economists seem to have come to believe that since statistics represent the only quantitative data which they can obtain, it is these statistical data which are the real facts with which they deal and that their theories must be given such a form that they explain what is statistically ascertainable. There are of course a few fields, such as the problems of the relation between the quantity of money and the price level, where we can obtain useful approximations to such simple relations – though I am still not quite persuaded that the price level is a very useful concept. But when it comes to the mechanism of change, the chain of cause and effect which we have to trace in order to be able to understand the general character of the changes to be expected, I do not see that the objectively measurable aggregates are of much help.