Revealed preference
actions over words · revealed preferences
The inference of what someone values from what they chose, rather than from what they said. Samuelson (1938) introduced it to rebuild demand theory on observable behaviour alone, eliminating any appeal to unobservable utility.
In practice
A client says price is the obstacle and then renews at the same price while declining the cheaper tier. The stated obstacle was price; the operative constraint was the cost of switching, and only the second is visible in what they did.
The common mistake
Reading choice as preference without reading the constraint. Someone who takes the only option available has revealed the constraint, not a preference; and Sen (1973) showed the inference also requires assumptions about consistency and self-interest that are substantive claims, not definitions.
Ask what someone wants and you get an answer shaped by what they would like to want. Watch what they do and you get something harder to fake and harder to interpret.
Samuelson's programme
Paul Samuelson (1938) set out to remove utility from consumer theory altogether. If preference is unobservable, build the theory from choices instead: if a bundle A is chosen when B was affordable, A is revealed preferred to BSamuelson, P. A. (1938). 'A Note on the Pure Theory of Consumer's Behaviour.' Economica 5(17). The weak axiom of revealed preference: if A is chosen when B is available, then in any situation where B is chosen, A must not have been available. Hendrik Houthakker (1950) added the strong axiom, ruling out longer cycles and securing the existence of a consistent utility representation.. Hendrik Houthakker (1950) completed it, showing that a consistency condition on observed choices is sufficient to reconstruct a utility function. The achievement is real: a theory of value built entirely on what can be seen.
The objections
Amartya Sen (1973) identified what the programme had smuggled in. Reading choice as preference requires that the chooser is consistent, that they choose in their own interest, and that the act of choosing carries no meaning beyond the object chosen. None of these is definitional and all three fail routinely. Choice can be shaped by commitment, by norms, or by what refusing would signalSen, A. (1977). 'Rational Fools: A Critique of the Behavioral Foundations of Economic Theory.' Philosophy & Public Affairs 6(4). Sen distinguishes sympathy — where another's welfare enters your own — from commitment, where you act against your own welfare. Commitment breaks the link between choice and preference entirely, and Sen argues it is common rather than exotic. — a person who declines the last item in a bowl has not revealed a preference for going without. Sen later showed choice is menu-dependent in ways no single ordering can represent (1993).
The constraint problem is separate and more common in practice. A choice made from a set of one reveals the set, not the preference. Most observed behaviour in employment, housing and purchasing is made under constraints tight enough that the inference is mostly reporting the constraint back — which is why revealed preference is far stronger evidence of what is available to someone than of what they want.
What it is good for anyway
Against stated preference it remains the better instrument, because stated preference is subject to social desirability, to poor introspective access, and to the gap between what people predict they will do and what they do. The honest formulation is comparative: behaviour is better evidence than testimony, and neither is direct access to a preference. The productive move in practice is not to choose between them but to treat the divergence as the finding — the gap between what a customer says and what they renew is usually where the real constraint is.
What it rules out
It rules out stated intention as evidence of future behaviour without corroboration. It rules out treating a survey as a measure of demand. And it rules out inferring preference from a choice made under a binding constraint, which is the error the concept is most often used to commit.
It does not rule out asking people what they want. It rules out believing the answer unweighted, and it supplies the check: compare it to what they did last time.
Sources
Houthakker, H. S. (1950). 'Revealed Preference and the Utility Function.' Economica 17(66). · Samuelson, P. A. (1938). 'A Note on the Pure Theory of Consumer's Behaviour.' Economica 5(17). · Sen, A. (1973). 'Behaviour and the Concept of Preference.' Economica 40(159). · Sen, A. (1977). 'Rational Fools.' Philosophy & Public Affairs 6(4). · Sen, A. (1993). 'Internal Consistency of Choice.' Econometrica 61(3).
Concept web
Open the full webQuestions
What is revealed preference?
Inferring what someone values from what they chose rather than from what they said. Samuelson (1938) introduced it to rebuild demand theory on observable behaviour alone, without appeal to unobservable utility.
What are the limits of revealed preference?
Sen (1973, 1977) showed the inference requires that the chooser is consistent, self-interested, and that choosing carries no meaning beyond the object — none of which is definitional. A choice made under a binding constraint also reveals the constraint, not a preference.
Is revealed preference better than asking people?
Usually, since stated preference is distorted by social desirability and poor introspective access. But neither is direct access to a preference, and the most useful signal is the divergence between them — the gap between what a customer says and what they renew.