This paper argues that cumulative causation processes are fundamental to understanding growth and development. Such processes derive from spatially concentrated increasing returns to scale including thick market effects, knowledge spillovers, sectoral and urban clustering, and self-reinforcing improvements in physical and social infrastructure. These sources of agglomeration have been extensively analyzed in the economic geography literature. They imply that spatial unevenness in economic activity and incomes is an equilibrium outcome. Growth tends to be 'lumpy', with some sectors in some countries growing fast while other countries lag. The policy challenge is to lift potential new centers of economic activity to the point where they can reap the productivity and investment climate advantages of increasing returns and cumulative causation.