Average hourly productivity has often been used to draw conclusions on long run per capita GDP growth, based on the assumption of full utilization of labour resources. In this paper, we argue that a failure to recognize the potentially significant wedges among the two variables – even in the long run - can be misleading. By applying both time series and panel cointegration techniques on data on 19 OECD countries, we fail to reject the hypothesis of absence of a long run common stochastic trend among the two variables in the period 1980-2005. Furthermore, we apply a simple decomposition of GDP growth into five variables, included some related to the supply-side and demographics, so to verify the single contributions to income growth and variance over our period of interest. We conclude that variables that have been so far absent in the growth literature have indeed a non-negligible role in explaining the dynamics of long run per capita GDP growth. In particular, these "forgotten factors" (that we identify with the employment and the activity rates and a demographic ratio) matter more in better performing economies, where we also highlight that productivity has been less important in determining GDP growth than in relatively bad performers.