Where does rational expectations collide?

In economics, rational expectations is the hypothesis that the forecasts people hold about an uncertain variable match, on average, the predictions of the correct economic model, so that a person's subjective expectation equals the objective conditional expectation given the information available. John Muth introduced it in 1961 to close a gap in models of price movements, where firms had to guess future prices: he proposed that these guesses should not be systematically wrong, because informed agents would learn from and eliminate any predictable error. The claim is about the accuracy of forecasts, not about the sensibleness of goals or the consistency of preferences. It does not say people are never wrong, only that their errors carry no exploitable pattern relative to the information they had. This is the narrow, original meaning, and it is a strong assumption about how expectations are formed rather than a plain observation about behaviour.[1]

Unit

A population's forecast of an economic variable, set against the model's prediction

Recovery means

Not applicable

Measured by

Tested by checking whether forecast errors are unbiased and uncorrelated with information known at the time

In macroeconomics, rational expectations became a modelling discipline rather than a single testable claim: the rational-expectations revolution of the 1970s required that the expectations written into a model be consistent with the model's own predictions, so that agents are not assumed to be fooled in ways the modeller can foresee. Robert Lucas drew the sharpest consequence in 1976. Because estimated relationships between economic variables reflect how people expected policy to behave, changing the policy rule changes behaviour and breaks the estimated relationship, so those relationships cannot safely be used to evaluate new policies. Sargent and Wallace pushed the logic further in 1975 with a policy-ineffectiveness result: if expectations are rational, systematic monetary policy that people can anticipate has no effect on real output. Here the word carries a methodological standard for building and using models, which is a different use from Muth's statistical property, and it is the meaning most often meant by the phrase today.[2][3]

Unit

A macroeconomic model's treatment of how agents form expectations

Recovery means

Not applicable

Measured by

Assessed by whether a model's expectations are consistent with its own solution and hold up when the policy rule changes

A large body of work treats rational expectations as an idealised benchmark that real people do not meet. Under adaptive expectations, agents extrapolate from past values and so make predictable errors when trends turn; under adaptive learning, they estimate the model as they go and only sometimes converge on the rational solution. Christopher Sims's 2003 theory of rational inattention keeps optimisation but adds a limit: attention is a scarce resource, so agents process only part of the available information, and their expectations come out rationally sluggish and noisy. This is the empirical and behavioural rebuttal, and it connects to bounded rationality, the view that cognition and information are limited; it should not be confused with rational choice, which concerns the consistency of preferences rather than the accuracy of forecasts. The disagreement is not about whether people have goals but about whether their forecasts can be treated as statistically optimal.[4]

Unit

The gap between observed forecasts and the rational-expectations benchmark

Recovery means

Not applicable

Measured by

Studied with survey forecasts, laboratory learning experiments, and models of inattention or adaptive learning