Economic growth has conventionally been modelled for space-less economies. Econometrically, growth models have mostly been estimated on time series of national economies with minimal distinctions between economies as large as Japan or the USA and as small as the smallest economies of Asia and Europe. This approach to the analysis of economic growth is especially dangerous when the impact of scientific and technological knowledge is important for the process of growth. Creative activities and the formation of knowledge are highly clustered in space. Thus, the spatial distribution of accessibility to knowledge capital and investments determines economic growth of nations and other spatial aggregates.
The Haavelmo paradox contrasts chaos as the generic property of non-linear dynamic models with the fact that most statistics on macroeconomic growth processes tend towards persistent constant positive rates of growth. The paradox can be resolved if the non-linear dynamic model is subdivided into fast, private variables and very slow, public variables. Modelling spatial accessibility of knowledge as a slow, public variable and machinery and similar material capital as a relatively faster, private variable ensures stable growth, at least in the short and medium terms of the economic growth processes.