It is shown that time-series of US productivity and hours are apparently affected by a structural break in the late 60s. Moreover, the importance of technology shocks over the business cycle has sharply decreased after the break.

VAR models with common cyclical features imply parsimonious univariate final equations, justifying the use of both time series aggregation, and homogenous panels with common factors and dynamic heterogeneity. However, conducting statistical inferences based on too restrictive models might be...