None of the elements in the Classical equations represents a directly observable entity. The coefficients, as usually interpreted, represent physical inputs for “best-practice” techniques (which at most only a few firms yet use). Data are actually collected as costs, that is, in monetary terms, not in physical terms. The products are likewise idealized and defined in physical terms, whereas output data are also mostly collected as revenue. Neither firms nor households, nor any other decisionmakers appear. Prices are uniform – the “law of one price” is assumed – and prices are expressed in barter terms; rates of profit and wages are similarly smoothed out and hold in barter value. No trace of the unruly market or the hurly-burly of actual competition can be found. Nor could there be, for the Classical equations do not represent behavior at all. They portray, in simplified and idealized form, the essential structure of the capitalist system, the core relationships between wages, prices and profits, on the one hand, and consumption, quantities, and growth on the other.

Yet this poses an epistemological problem. If all our empirical data come from the market, while the Classical equations are independent of the market, how can we verify that those equations correctly describe the core relationships of the economy? An answer was supplied by Smith and Ricardo: market prices and other market variables “gravitate” toward, or around, the corresponding “natural” prices or other variables, where “natural” values are those determined by the Classical structural equations.