Cardinal Utility

Cardinal utility is a utility measure that encodes not merely the ranking of preferences but the intensity or strength of those preferences. Where ordinal utility tells us that A is preferred to B, cardinal utility tells us by how much — or at least that the difference between A and B is twice the difference between B and C. This additional structure makes cardinal utility indispensable for any theory that requires weighing outcomes against probabilities, costs against benefits, or individual gains against collective losses. Without cardinality, expected utility theory is algebraically empty; with it, the theory becomes computationally powerful and conceptually treacherous.

The concept has a contested history. Nineteenth-century economists — Jevons, Menger, Walras — treated utility as inherently cardinal, measuring psychological pleasure on a numerical scale. Early twentieth-century economists, particularly in the ordinalist revolution led by John Hicks and Roy Allen, rejected cardinality as psychologically unfounded and methodologically unnecessary. The ordinalists argued that only preference rankings are observable, and that any theory requiring cardinality was relying on metaphysical assumptions about mental states. The counter-revolution came with von Neumann and Morgenstern's 1944 theorem, which proved that cardinality can be derived from behavioral axioms about choice under uncertainty, without requiring introspection about psychological intensity.

Yet the derivation does not resolve the deeper problem. Cardinal utility functions are unique only up to positive affine transformation — a shift and a scale. This means that statements about ratios of utilities ("A is twice as good as B") are meaningless, while statements about differences ("the gain from A to B is worth the same as the gain from C to D") are meaningful. This "interval scale" property is sufficient for expected utility but insufficient for many applications that economists and policy analysts casually assume cardinality supports. The measurement theory of utility — whether utility is an interval scale, a ratio scale, or merely an ordinal ranking — remains unresolved, and the casual treatment of utility as if it were a ratio scale (as in "this policy produces twice the welfare of that one") is one of the most common errors in applied welfare economics.