Expectations, Heterogeneous Forecast Errors, and Consumption: Micro Evidence from the Michigan Consumer Sentiment Surveys

The household data underlying the Michigan Index of Consumer Sentiment are used to test the rationality of consumer expectations and their usefulness in forecasting expenditure. The results can be interpreted as characterizing the shocks that hit different types of households over time. Expectations are found to be biased and inefficient, at least ex post. People underestimated the disinflation of the early 1980s and the severity of recent business cycles. People's forecast errors are also systematically correlated with their demographic characteristics, in part because of time-varying, group-level shocks. Further, sentiment helps forecast consumption growth. Some of this rejection of the permanent income hypothesis is due to the systematic demographic components in forecast errors.