This paper shows that heterogeneity of expectations matters. The individual differences studied concern the assessment of the input-output relationship. Output clearly declines with rising expectations errors. This occurs even - and strongly - when there are equal numbers of over- and under-predictors in the economy. This shatters the belief which is still widely held that market outcomes equal the predictions from rational expectations models as long as individual idiosyncrasies cancel out on average. Depending on the precise object of disagreement, it can be worse to have a majority of pessimists or a majority of optimists. The study suggests that similarity of assessments can mitigate the losses due to wrong assessments. Hence, conformity can be a social virtue and individualism can be a burden.