Did the US consumer overreact? A test of rational expectations

Using data for the US economy from 1995 to 2008, I estimate a business cycle model in which consumers form expectations rationally, and an alternative model that explicitly allows for deviations from rational expectations in the form of overreaction to noisy signals. The second model does not have any advantage in explaining the movements in consumption and productivity in this period. Therefore, I cannot reject the hypothesis that consumers behaved rationally. The rather exuberant movements of this episode are rationalized by a story in which permanent income consumers find it hard to distinguish between permanent movements in productivity and temporary ones. Consumers who update their beliefs about trends in productivity on the basis of fairly noisy signals adjust their behavior very slowly. After the 1995–2000 productivity boom, consumers learnt very slowly about a subsequent decline in productivity growth, which led to financial and trade imbalances that ended in a correction starting around 2007. The whole boom-bust cycle was long, taking about 14 years to complete. ► Two business cycle models were estimated for the US economy 1995–2008. ► The first model features consumers that use Bayes’ rule to process information. ► The second model features consumers that overreact to current information. ► I could not reject the hypothesis that consumers use information optimally. ► Noisy information is able to account for the observed movements of consumption.