Existing evidence for unconditional convergence in the OECD is mixed, and depends largely on whether time series or cross sectional methods are used. In this paper we reconsider the evidence for unconditional convergence by dividing the long run data into several subperiods. We use a two stage approach in this work. We first model the growth rate of output directly and use this model to estimate the long-run growth rate for the countries in our sample. We then use the estimates of long-run growth in output to test for unconditional convergence and to test for equality of long-run growth across countries. GLS is used to explicitly take into account the sampling uncertainty inherent in our estimates of the long-run growth rate we found in the first stage of the process. The results show strong evidence for unconditional convergence in the post WWII period 1951-1974, but no evidence of convergence in the periods preceding or following this period. Moreover, it is di±cult to reject the hypothesis that most of the countries in our sample had the same growth rate outside of this period. Thus find little evidence to suggest that absolute convergence has been a continuous long run process, and some evidence for the view that national policies mainly affect income levels rather than growth rate
