Abstract: With the emergence of accelerated technological change and progress, economic growth has been exhibiting increasingly complex features. Trade within and across industries, countries and regions, has risen sharply, and the globalisation effects of economies of specialisation, low-cost information technologies and networks have taken a stronger grip on economies. Furthermore, long-run economic growth backed by conventional fossil-based technologies (within the context of an increasing population and escalating demands for goods and services) are exerting an unprecedented pressure on the environment, particularly in terms of greenhouse gas emissions and climate change. We investigate not only the historical dynamics of the world economic system, but also project likely long-term changes in the structure of economic output across economies, and assess co-movements between economic activities and greenhouse gas emissions, with an emphasis on the role of endogenous technological change. This is mostly a methodological paper with innovative large-scale econometric modelling techniques with dynamic structures being brought to the fore. We find that century-long projections of economic systems are seldom addressed in the literature, not least because of the inability of models to predict structural changes and the penetration of new technologies. We argue that four key features need to be adequately addressed when modelling long-run economic changes at the macro-level, with particular relevance for climate change analysis. First, in projecting economic systems one needs to first consider their behaviour in the past. Second, the channels of international transmission and the increasing economic interdependence across markets and countries have to be incorporated. Third, there is a need for explicitly modelling the effects of endogenous technological change on economic systems. And fourth, which is the focus of our future research, our understanding of the uncertainties underlying the projections should be explored to avoid misleading deterministic solutions based on the assumed properties of the systems.