The paper computes the labor productivity and total factor productivity using a series of hours worked from 1982 to 2012. In order to do that we use a methodology that enables us to combine PNAD and PME data despite their methodological changes in the period 1982 to 2012, to construct a series of monthly labor hours worked. The comparison between people occupied and labor hours reveals a decrease in weekly labor hours, a phenomenon which is consistent during the whole period of analysis. This result allows us to reinterpret Brazilian productivity over the last three decades providing a different analysis of its evolution. We show that labor productivity losses (-0.6%) between 1982 and 1992 were the result of fewer hours worked per week/worker (-0.7%) while labor hour productivity was almost constant (+0.1%). The results show that labor productivity was underestimated by 35.8% in the period 1982-2012 while TFP was underestimated by 16.3%.
