The Fundamental Error of Rational Expectations Proponents Is Their Claim that They Have Discovered True Statistical Models: Given that No Model Can Be True, Talk of Having a 'True Model' Is An Anti-Scientific Oxymoron Given Keynesian Uncertainty

Abstract

No model can ever be true. By definition, models are only, at best, approximations to reality. Some models are better approximations than others, so one can talk about one model being better than another model. However, to talk about a model yielding true predictions means that the speaker does not understand what a model is and what it is used for. This is especially true in the area of probability and statistics. George Box said it well when he stated that ‘all models are wrong, but some are useful.’

Rational expectations advocates violate basic scientific approaches to theory construction and model use when they claim that there is a true model of how the economy operates that consumers and producers can learn from experience. There can never be any scientific support for that claim or any of the following claims, given that all models are only approximations, which can never be true:

• There is a true(correct, right ,valid) probability
• There is a true(correct, right, valid) expectation
• There is a true(correct, right, valid) model
• There is a true(correct, right, valid) expected value
• Consumers and producers can learn the true(correct,right,valid) model
• There is a true(correct, right, valid), objective probability or probability distribution
• There are true(correct,right,valid) model consistent expectations