Joan Robinson, 1955

“The purpose of studying economics is not to acquire a set of ready-made answers to economic questions, but to learn how to avoid being deceived by economists.”

Joan Robinson wrote this characteristically sharp sentence in a 1955 essay on teaching economics. A major Cambridge economist and an architect of the theory of imperfect competition, she understood both the power and the limits of formal analysis. Economics could reveal hidden relationships, but its authority could also make assumptions look like facts and political choices look inevitable. She defined education defensively as well as constructively: a student should learn enough about models, evidence, and institutions to resist being overawed by professional confidence or by a tidy answer detached from its conditions.

For business leaders, the warning applies to forecasts, valuations, pricing models, productivity claims, and market research. A number gains influence when it appears precise, even if the inputs depend on uncertain behavior or omit a group that bears the cost. Critical economic literacy means asking what is held constant, whose incentives are represented, how sensitive the result is to a different assumption, and whether the model explains causation or merely describes a pattern. The goal is not to reject expertise. It is to become a better client of expertise and a more responsible user of its conclusions.

Robinson’s phrasing can sound as though economists intend deception. Often the more ordinary danger is that a model built for one purpose travels into another without its caveats. Every useful abstraction leaves something out. The relevant question is whether what was omitted changes the decision. Leaders should also resist the opposite error of treating personal intuition as superior simply because analysis is imperfect. Transparent methods, competing models, disconfirming evidence, and independent review create stronger judgment than either blind faith or cynical dismissal.

The lesson matters when algorithms translate economic assumptions into automated choices about credit, hiring, inventory, and prices. Once embedded in software, a contested theory can operate at scale while appearing neutral. Decision-makers can require model cards, scenario ranges, subgroup outcomes, and a named person accountable for override and remedy. They can ask affected people to identify effects the dashboard does not record. Robinson’s purpose for study becomes an operating principle: understand the machinery of an argument well enough to see where judgment entered, who benefits from its framing, and when reality has escaped its answer. A decision document can show the base case, alternative interpretation, range of plausible outcomes, and variables with the greatest influence. Naming the model’s owner and intended use prevents an analytical artifact from becoming universal policy by accident. When the result affects access or opportunity, publishing outcome differences across groups adds a reality check that aggregate accuracy cannot provide and invites correction before authority hardens around the number.