I have some problems with the rational expectations hypothesis. To hear some macro economists talk about it, you’d think that it was a wonderful scientific innovation for economists to start assuming that the agents in their models know the structure of the economy and only make random errors in forecasting. Such sentiments are entirely misguided.
Is it really the case that the market behaves as if the people in the market do not make systematic errors? If so, this is a highly interesting feature of the market economy, one that economists should explain rather than simply asserting.
It is entirely possible to construct a theory of markets without presupposing the specific types of errors (systematic or random) that people make. In doing so, we should ask ourselves what would happen to an entrepreneur who repeatedly and systematically failed in forecasting the future state of prices. Such a person would repeatedly earn losses. Faced with these losses, he would be forced either to revise his forecasting method such that his predictions would improve, or face continual losses and eventual bankruptcy. Thus, the market process tends, in the limit, towards something like rational expectations.
The advantage of the argumentation above is that it can tell us something about the circumstances under which we should not expect rational expectations to prevail. The selective market process that tends toward rational expectations does not happen instantaneously, so we should not expect entrepreneurs to immediately adapt to major structural changes to the economy. Nor can the market perform its selective function under any conceivable set of government interventions; a government that bails out unprofitable companies is directly interfering with the market’s ability to select good forecasters over bad.
I suspect that the economists who assume rational expectations as an ex ante condition of market actors will not disagree with much of the above analysis. They may even use some of the same reasoning to justify the assumption. The problem is that removing one of the market’s chief functions from one’s formal models can blind one to the possibility that this function could ever be impaired.