… is from page 285 of Hayek’s 1978 collection, New Studies in Philosophy, Politics, Economics and the History of Ideas; specifically, it’s from Hayek’s January 1966 essay in the Oriental Economist entitled “Personal Recollections of Keynes and the ‘Keynesian Revolution'”:
He [J.M. Keynes] certainly from the beginning was much given to thinking in aggregates and always had a faible for (sometimes very tenuous) global estimates. Already in discussion of the 1920s arising out of Great Britain’s return to the gold standard his argument had been couched entirely in terms of price- and wage-levels, in practically complete disregard of the structure of relative prices and wages, and later the belief that, because they were statistically measurable, such averages and the various aggregates were also causally of central importance, appears to have gained increasing hold upon him. His final conceptions rest entirely on the belief that there exist relatively simple and constant functional relationships between such ‘measurable’ aggregates as total demand, investment, or output, and that empirically established values of these presumed ‘constants’ would enable us to make valid predictions.