EN
This paper examines structural dependencies within the national innovation systems of 125 countries in period 2006 – 2008. Some 37 partial indicators from the World Economic Forum and World Bank databases are aggregated into 11 independent and 2 dependent variables. Variable relations account for distinctive non-linear dynamics and are modelled via two-step cluster analysis and artificial neural networks. Overall quality of education system, property rights, law and ethics, and competition forces are identified by major predictors for innovativeness. The paper also examines some assumptions by the varieties of capitalism theory on institution complementarities and level of innovativeness. No evidence is found for liberal (coordinated) economies having more efficient innovation systems than mixed ones.