Future scenarios of climate change depend on the projections of greenhouse gas emissions, which are highly uncertain. A framework for projecting emissions should focus on the sources of economic growth and the changing structure of the global economy over time. It also requires an understanding of key historical and statistical issues, including the role of convergence assumptions and the appropriate basis of comparisons between countries. This paper presents a methodology developed using the G-Cubed multi-country model in which the economic structure and emissions outcomes are determined simultaneously. In order to illustrate the importance of the assumptions underlying the way these long term carbon projections are produced we also explore the debate around the "Castles and Henderson Critique" of the Special Report on Emission Scenarios (SRES) regarding the use of Market Exchange Rates (MERs) rather than Purchasing Power Parity exchange rates (PPPs) for benchmarking income differentials in the world economy. We show that under one scenario, emission projections based on convergence assumptions defined in MER terms are 40% higher by 2100 than emission projections based on PPP comparisons of income differentials. While this result can not necessarily be generalized to all forecasting frameworks, the potential magnitude of the difference suggests that this is a significant issue for such projection efforts.