Revealed Preference

What people say they want and what they actually choose are often different. Revealed preference is the principle that you learn someone's true preferences from their choices, not from their statements. If a customer says they value sustainability but consistently buys the cheapest option, their revealed preference is price. If a board says it wants long-term value but fires CEOs who miss quarterly targets, its revealed preference is short-term results. Actions reveal; words can mislead. The economist Paul Samuelson formalised the idea in the 1930s: preferences are inferred from observed behaviour under constraints. You don't need to ask people what they want; you need to see what they do when they have to trade off.

The power of the model is that it cuts through stated preferences, social desirability, and self-deception. People misreport their preferences for many reasons — they want to look good, they don't know themselves, or they're answering a hypothetical. When they act, they put something at stake. The choice reveals what they're willing to sacrifice for what. In negotiation, watch what the other side does with their time, their offers, and their walk-aways. In strategy, watch how customers allocate money and attention, not what they say in surveys. In hiring, watch what candidates have done, not what they say they'll do. Revealed preference is the discipline of trusting behaviour over words.

Revealed preference has limits. Behaviour is noisy; one choice can be a mistake or a one-off. You need a pattern of choices under comparable constraints. And sometimes the constraint is hidden — someone might "choose" A over B because they never had a real chance to choose B. The model works when you observe consistent choices and you understand the constraints. Then you can infer preference, predict behaviour, and design offers that match what people will actually do.

Revealed preference shows up whenever someone's actions contradict their words, or when the only reliable signal of what they want is what they choose when it costs them something.

Common misapplication: Ignoring stated preference entirely. Stated preference can signal intent, aspiration, or constraint (e.g. "I'd pay more if I could"). The mistake is to only use stated preference. The correction is to weight revealed preference more — especially when stakes are high and words and actions diverge. Use both; trust actions when they conflict.

Second misapplication: Reading one choice as full preference. A single decision can be noise, a mistake, or a one-off. Revealed preference is most reliable when you see a pattern of choices under comparable constraints. Don't over-interpret one data point. Look for consistency across time or across similar trade-offs.

Revealed preference sits with models about incentives, signalling, and how we infer intent from behaviour. The connections below extend the lens to strategy, negotiation, and agency.

Preference is inferred from choice. We don't observe utility or desire directly; we observe which bundle was chosen from the set of available options. The hypothesis is that the chosen bundle is the one that was preferred. That's the entire move: behaviour is the data, preference is the inference. The same logic applies beyond the consumer: observe choices under constraints, infer what was preferred, predict future choices when constraints are similar.

Summary: Revealed preference is the principle that you learn what someone wants from their choices under constraint, not from their words. Actions reveal; words can mislead. Use it to read customers, partners, and organisations: watch what they do when it costs them something. Design offers and incentives around revealed preference. When words and actions conflict, trust actions. The model has limits: look for patterns of choices, and be clear about constraints. Connected ideas include incentives, signalling, and information asymmetry.